Lecture Notes in Civil Engineering

Cuong Ha-Minh · Dong Van Dao ·

Farid Benboudjema ·
Sybil Derrible · Dat Vu Khoa Huynh ·
Anh Minh Tang   Editors

CIGOS 2019,
Innovation
for Sustainable
Infrastructure
Proceedings of the 5th International
Conference on Geotechnics, Civil
Engineering Works and Structures
Lecture Notes in Civil Engineering

Volume 54

Series Editors
Marco di Prisco, Politecnico di Milano, Milano, Italy
Sheng-Hong Chen, School of Water Resources and Hydropower Engineering,
Wuhan University, Wuhan, China
Ioannis Vayas, Institute of Steel Structures, National Technical University of
Athens, Athens, Greece
Sanjay Kumar Shukla, School of Engineering, Edith Cowan University, Joondalup,
WA, Australia
Anuj Sharma, Iowa State University, Ames, IA, USA
Nagesh Kumar, Department of Civil Engineering, Indian Institute of Science
Bangalore, Bangalore, Karnataka, India
Chien Ming Wang, School of Civil Engineering, The University of Queensland,
Brisbane, QLD, Australia
Lecture Notes in Civil Engineering (LNCE) publishes the latest developments in
Civil Engineering - quickly, informally and in top quality. Though original research
reported in proceedings and post-proceedings represents the core of LNCE, edited
volumes of exceptionally high quality and interest may also be considered for
publication. Volumes published in LNCE embrace all aspects and subfields of, as
well as new challenges in, Civil Engineering. Topics in the series include:
• Construction and Structural Mechanics
• Building Materials
• Concrete, Steel and Timber Structures
• Geotechnical Engineering
• Earthquake Engineering
• Coastal Engineering
• Ocean and Offshore Engineering; Ships and Floating Structures
• Hydraulics, Hydrology and Water Resources Engineering
• Environmental Engineering and Sustainability
• Structural Health and Monitoring
• Surveying and Geographical Information Systems
• Indoor Environments
• Transportation and Traffic
• Risk Analysis
• Safety and Security
To submit a proposal or request further information, please contact the appropriate
Springer Editor:
– Mr. Pierpaolo Riva at pierpaolo.riva@springer.com (Europe and Americas);
– Ms. Swati Meherishi at swati.meherishi@springer.com (Asia - except China -
and Australia/NZ);
– Ms. Li Shen at li.shen@springer.com (China).
Indexed by Scopus

More information about this series at http://www.springer.com/series/15087

Cuong Ha-Minh Dong Van Dao
• •

Farid Benboudjema Sybil Derrible

• •

Dat Vu Khoa Huynh Anh Minh Tang

•

Editors

CIGOS 2019, Innovation

for Sustainable Infrastructure
Proceedings of the 5th International
Conference on Geotechnics, Civil Engineering
Works and Structures

123
Editors
Cuong Ha-Minh Dong Van Dao
University of Paris-Saclay University of Transport Technology
Cachan, France Thanh Xuan, Hanoi, Vietnam

Farid Benboudjema Sybil Derrible

LMT Cachan University of Illinois at Chicago
University of Paris-Saclay Chicago, IL, USA
Cachan, France
Anh Minh Tang
Dat Vu Khoa Huynh Ecole des Ponts ParisTech
Norwegian Geotechnical Institute Marne-la-Vallée, France
Oslo, Norway

ISSN 2366-2557 ISSN 2366-2565 (electronic)

Lecture Notes in Civil Engineering
ISBN 978-981-15-0801-1 ISBN 978-981-15-0802-8 (eBook)
https://doi.org/10.1007/978-981-15-0802-8
© Springer Nature Singapore Pte Ltd. 2020
This work is subject to copyright. All rights are reserved by the Publisher, whether the whole or part
of the material is concerned, specifically the rights of translation, reprinting, reuse of illustrations,
recitation, broadcasting, reproduction on microfilms or in any other physical way, and transmission
or information storage and retrieval, electronic adaptation, computer software, or by similar or dissimilar
methodology now known or hereafter developed.
The use of general descriptive names, registered names, trademarks, service marks, etc. in this
publication does not imply, even in the absence of a specific statement, that such names are exempt from
the relevant protective laws and regulations and therefore free for general use.
The publisher, the authors and the editors are safe to assume that the advice and information in this
book are believed to be true and accurate at the date of publication. Neither the publisher nor the
authors or the editors give a warranty, expressed or implied, with respect to the material contained
herein or for any errors or omissions that may have been made. The publisher remains neutral with regard
to jurisdictional claims in published maps and institutional affiliations.

This Springer imprint is published by the registered company Springer Nature Singapore Pte Ltd.
The registered company address is: 152 Beach Road, #21-01/04 Gateway East, Singapore 189721,
Singapore
Organizing Committee

Chairmen
Ha, Minh Cuong, ENS Paris-Saclay, France
Dao, Van Dong, University of Transport Technology, Vietnam
Adeli, Hojjat, Ohio State University, USA
International Members
Bui Nguyen, Hoang, AVSE-Global & EDF, UK
Dang, Thuc Trang, Scene Plus Co. Ltd., Vietnam
Dinh, Van Nguyen, ANTS & University College Cork, Ireland
Do, Tuan Anh, Sorbonne University, France
Huynh, Dat Vu Khoa, Norwegian Geotechnical Institute, Norway
Le, Thi Xiu, ENPC, France
Le, Van Nhat, EGIS, France
Nguyen, Giang, University of Adelaide, Australia
Nguyen, Hai Nam, ENS Paris-Saclay, France
Nguyen, Quang Huy, INSA Rennes, France
Nguyen, Thuong Anh, Geodynamique et Structure, France
Nguyen, Tien Long, Tractebel, France
Nguyen, Van Minh, Tractebel, France
Pham, Cao Hung, University of Sydney, Australia
Pham, Van Hung, University of Paris-Est Marne-la-Vallée, France
Ta, Ngoc Binh, Institute of construction economics, Vietnam
Tang, Anh Minh, AVSE-Global & ENPC, France
Vu, Xuan Hong, University Lyon 1, France
Local Members (UTT)
Le, Hoang Anh, Faculty of Civil Engineering
Le, Nguyen Khuong, Faculty of Civil Engineering
Le, Thanh Hai, Department of Science, Technology and International Cooperation
Ly, Hai Bang, Faculty of Civil Engineering

v
vi Organizing Committee

Ngo, Quoc Trinh, Department of Science, Technology and International

Cooperation
Ngo, Thi Thanh Huong, Faculty of Civil Engineering
Nguyen, Long Khanh, Department of Science, Technology and International
Cooperation
Nguyen, Nam Hai, Faculty of Information Technology
Nguyen, Thi Thu Hien, Department of Science, Technology and International
Cooperation
Pham, Thai Binh, Faculty of Civil Engineering
Phi, Luong Van, Department of Science, Technology and International Cooperation
Tran, Trung Hieu, Department of Science, Technology and International
Cooperation
Hoang, Vu, Department of Science, Technology and International Cooperation
Truong, Thi My Thanh, Department of Economy of Transportation
Vu, Duy Tung, Center of Technology of Communication
International Scientific Committee

AMCS—Advanced Modelling and Characterization of Structures

Argoul, Pierre, IFSTTAR, France
Azenha, Miguel, Minho University, Portugal
Becque, Jurgen, University of Sheffield, UK
Benabou, Lahouari, Université de Versailles Saint-Quentin-en-Yvelines, France
Bui, Quang Hien, STECHCO Inc., Canada
Bui, Tan Trung, INSA Lyon, France
Couchaux, Mael, INSA Rennes, France
Dang, Cong Thuat, University of Danang, Vietnam
Elgawady, Mohamed, University of Missouri, USA
Ferrier, Emmanuel, University Lyon 1, France
Gilbert, Benoit, Griffith University, Australia
Herve-Secourgeons, Guillaume, IMSIA, EDF R&D, France
Jalil, Alan, EGIS, France
Labbe, Pierre, ESTP RENON, France
Lanos, Christophe, IUT/INSA Rennes, France
Le, Thien Phu, Université d’Évry-Val-d’Essonne, France
Meftah, Fekri, INSA Rennes, France
Mirza, Olivia, Western Sydney University, Australia
Nguyen, Giang, University of Adelaide, Australia
Nguyen, Quang Huy, INSA de Rennes, France
Nguyen, Tien Long, Tractebel France, France
Nguyen, Van Bac, University of Derby, UK
Paultre, Patrick, Université de Sherbrooke, Canada
Pham, Cao Hung, University of Sydney, Australia
Poologanathan, Keerthan, Northumbria University, UK
Potin, Gildas, Tractebel Engineering, France
Ragueneau, Frédéric, ENS Paris Saclay, France
Sansour, Carlo, University of Nottingham, UK
Tan, Ee Loon, Western Sydney University, Australia

vii
viii International Scientific Committee

Teh, Lip, Wollongong University, Australia

Thai, Huu Tai, University of Melbourne, Australia
Tran, The Truyen, University of Transport and Communications, Vietnam
Turmo, Coderque José, School of Civil Engineering of Barcelona, Spain
Vo, Thuc, Northumbria University, UK

Building Information Modelling (BIM)

Chi, Seokho, Seoul National University, Korea

Chia, Fah Choy, Universiti Tunku Addul Rahman, Malaysia
Han, Sanguk, Hanyang University, Korea
Han, Hoang Mai, The BIM Factory, Vietnam
Hussain, Afifuddin Husairi, Universiti Kebangsaan, Malaysia
Kang, Taewook, Korea Institute of Civil Engineering and Building Technology,
Korea
Kim, Jung In, City University of Hong Kong, China
Likhitruangsilp, Veerasak, Chulalongkorn University, Thailand
Ngo, Quang Tuan, Hanoi Architectural University, Vietnam
Nguyen, Nam Trung, Power Engineering Consultancy Jsc. 3, Vietnam
Nguyen, The Quan, National University of Civil Engineering, Vietnam
Nguyen, Viet Hung, University of California, USA
Park, Jongwoong, Chung-Ang University, Korea
Park, Manwoo, Myongji University, Korea
Pham, Hong Luan, Ho Chi Minh City University of Technology, Vietnam
Preece, Christopher, Abu Dhabi University, UAE
Shim, Changsu, Chung-Ang University, Korean
Shin, Dohyoung, Inha University, Korea
Suryanto, Krishna, Institute Tecknologi Bandung, Indonesia
Ta, Ngoc Binh, Institute of Construction Economics, Vietnam
Thang, Anh Long, Structemp, UK
Wu, Wei, California State University, USA
DML—Data Mining and Machine Learning
Bui, Tien Dieu, University of South-Eastern Norway, Norway
Chou, Jui-Sheng (Rayson), National Taiwan University of Science and
Technology, Taiwan
Le, Hoang Son, Vietnam National University, Vietnam
Liu, Zhongqiang, Norwegian Geotechnical Institute, Norway
Ngo, Ngoc-Tri, National University of Singapore, Singapore
Tang, Pingbo, Arizona State University, USA
International Scientific Committee ix

GEE—Geotechnics for Environment and Energy

Bouvard, Anne, INSA de Lyon, France
Bui, Quoc Bao, Ton Duc Thang University, Vietnam
Choi, Jung Chan, Norwegian Geotechnical Institute, Norway
Collin, Frédéric, University of Liège, France
Ddani, Mohamed, Tractebel, INSA de Rouen, France
Dias, Daniel, University Grenoble Alpes, France
Dong, Jia-Jyun, National Center University, Taiwan
Gastebled, Olivier, Tractebel, INSA de Lyon, France
Huynh, Dat Vu Khoa, Norwegian Geotechnical Institute, Norway
Kaynia, Amir M., Norwegian Geotechnical Institute, Norway
Leo, Chin, Western Sydney University, Australia
Nguyen, Thanh Son, University of Ottawa and Canadian Nuclear Safety
Commission, Canada
Pham, Thai Binh, University of Transport Technology, Vietnam
Taherzadeh, Reza, Tractebel, INSA de Lyon, France
Tang, Anh Minh, Ecole des Ponts ParisTech, France
Tran Nguyen, Hoang Hung, Hochiminh City University of Technology, Vietnam
Wong, Henry, ENTPE, France
Wu, Li, China University of Geosciences, China
SCMT—Sustainable Construction Materials and Technologies
Ashraf, Mahmud, Deakin University, Australia
Beakou, Alexis, IFMA/SIGMA Clermont, France
Benboudjema, Farid, ENS Paris Saclay, France
Bui, Ha, Monash University, Australia
Bui, Quoc Tinh, Tokyo Institute of Technology, Japan
Dotelli, Giovanni, Politechnico Milano, Italy
Forth, John, School of Leeds, UK
Garibaldi, Luigi, Politecnico di Torino, Italy
Golinval, Jean-Claude, Université de Liège, France
Grande, Daniel, University of Paris Est Creteil, France
Grillet, Anne-Cécile, Université Savoie Mont Blanc, France
Ha Minh, Cuong, AVSE Global and ENS Cachan, France
Ly, Hai Bang, University of Transport Technology, Vietnam
Monchiet, Vincent, University of Paris Est Marne-la-Vallée, France
Monteiro, Eric, Arts & Métiers ENSAM Paris, France
Nakarai, Kenichiro, Hiroshima University, Japan
Nguyen, Van Bac, University of Derby, UK
Nguyen, Khanh Son, Hochiminh City University of Technology, Vietnam
Nguyen, The Duong, Duy Tan University, Vietnam
Poh, Leong Hien, National University of Singapore, Singapore
Proust, Gwénaëlle, University of Sydney, Australia
Regnier, Gilles, Arts & Métiers ENSAM Paris, France
x International Scientific Committee

Serra, Roger, INSA Val de Loire, France

Si Larbi, Amir, ENISE, France
Tran, Minh Tung, Ton Duc Thang University, Vietnam
Vu, Xuan Hong, University of Lyon 1, France
Yvonnet, Julien, University of Paris Est Marne-la-Vallée, France
UPTE—Urban Planning, Transport and Environment
Button, Cat, University of Newcastle, UK
Chan, Tak-Ming, Hong Kong Polytechnic University, Hong Kong
Dang, Thuc Trang, Scene Plus Co. Ltd., Vietnam
Derrible, Sybil, University of Illinois at Chicago, USA
Do, Thuan An, Thuy Loi University, Vietnam
Nguyen, Frédéric, University of Liège, France
Nguyen, Hong Quan, Vietnam National University—Ho Chi Minh City, Vietnam
Hsu, Mark Shu-Chien, Hong Kong Polytechnic University, Hong Kong
Le, Hai Vu, Monash University, Australia
Nghiem, Duc Long, University of Technology Sydney, Australia
Nguyen, Chuong, CSIRO—Data61, Australia
Nguyen, Xuan Thinh, TU Dortmund University, Germany
Quan Luna, Byron, DNV GL, Norway
Rilett, Laurence, University of Nebraska-Lincoln, USA
Trinh, Toan, University of California, USA
Organisers

xi
Partner

xiii
Patronage

xv
Preface

Following the great success of each of the four previous consecutive events since
2010, CIGOS (Congrès International de Géotechnique—Ouvrages—Structures)
has firmly established its international reputation and is perceived as an important
conference for promoting both academic and high-quality professional exchanges
in the fields of Geotechnics, Civil Engineering Works and Structures. This, the
fifth International Conference—CIGOS 2019, held in Hanoi, Vietnam, was
co-organized by the Association of Vietnamese Scientists and Experts (AVSE
Global) in collaboration with the University of Transport Technology (UTT) and in
partnership with the Ecole normale supérieure de Paris-Saclay (ENS Paris-Saclay).
CIGOS 2019 placed particular focus on the theme of Innovation for
Sustainable Infrastructure, aiming to not only raise awareness of the vital
importance of sustainability in infrastructure development but also highlight the
essential roles of innovation and technology in helping to plan and build sustainable
infrastructure. Having built on the success of CIGOS 2017, this year’s conference
saw almost double the number of published papers, which covered a broad spec-
trum within the following topics:
• Advanced Modeling and Characterization of Structures
• Sustainable Construction Materials and Technologies
• Geotechnics for Environment and Energy
• Urban Planning, Transport and Environment
• Data Mining & Machine Learning
• Building Information Modeling
The present CIGOS proceedings contain 201 peer-reviewed papers from 27
different countries, six of which are keynotes presented by internationally out-
standing experts in their respective fields. The choice of papers was made on the
basis of nearly 380 abstracts accepted. The first keynote presents the author’s wide
perspective on four decades of computing in civil engineering, which are highly
multidisciplinary—crossing computer science, multiple engineering fields, mathe-
matics and medicine. The second, third, fourth, and fifth keynotes address the recent
novel technologies and innovations in transportation and construction, which have

xvii
xviii Preface

been remarkable in helping to plan and build infrastructure that is increasingly more
sustainable. The sixth and final keynote emphasizes that society and standards
require increasingly more “risk-informed” based decisions, demonstrating that the
implementation of reliability and risk-based approaches can not only provide useful
complementary information both for design and for re-evaluation during the life-
time of the infrastructure but also efficiently assist the preparation of sustainable
engineering recommendations and risk-informed decision making.
We wish to acknowledge the wonderful support of the scientific committee and
the external reviewers, who provided valuable assistance in the review process and
undoubtedly made a tremendous contribution in raising the quality of the pro-
ceedings. We are thankful to the keynote speakers, authors and participants for their
enthusiastic efforts and time in preparing valuable papers as well as bringing
encouragement and inspiration to the conference. Our special thanks also to the
organizing committee, without whom it would have been difficult to run this
conference.
Last but not least, we would also like to thank all sponsors for their support and
generosity in making our CIGOS 2019 such a success.
On behalf of the CIGOS 2019’s committees, we believe that the present pro-
ceedings, which include a mixture of contributions from academia and practitioners,
will provide valuable up to date knowledge for all those interested in Geotechnics,
Civil Engineering Works and Structures.

Cachan, France Cuong Ha-Minh

Thanh Xuan, Hanoi, Vietnam Dong Van Dao
Cachan, France Farid Benboudjema
Chicago, USA Sybil Derrible
Oslo, Norway Dat Vu Khoa Huynh
Marne-la-Vallée, France Anh Minh Tang

Acknowledgements We would like to thank the following invited experts for reviewing the
papers: Brian Carlton, Finn Lovholt, Pascale Massaad, Thi Binh Nguyen, Thanh Tu Nguyen,
Thi Thanh Thao Pham.
Sponsors pages

Gold Sponsor

VinTech City, a member of Vingroup, was founded with the aim of supporting
the Vietnamese ecosystem of tech research and startups, following the success
model of Silicon Valley. To realize such mission, VinTech City has identified three
fundamental areas of activities: Tech talents, tech products, and an ecosystem of
supporting activities. Among these factors, the workforce and tech products with
competitive edges have been considered the stepping stones of the development
strategy of VinTech City.
Specifically, VinTech City has designed programs focusing on tech startups,
innovators, researchers, university lecturers and students, in order to support them
in creating innovative research and translating them into commercialized products
and companies based on applied research. In May 2019, VinTech City has officially
announced six such programs, namely: (i) VinTech Fund—a grant for applied
research; (ii) Research Laboratory sponsorship; (iii) University-Enterprise
Co-operative Semester; (iv) SAP Training; (v) Events, workshops, or seminars
sponsorship; and (vi) Tech Startup Student Clubs. Such wide range of support
activities and programs demonstrates the comprehensive cooperation between
Vingroup and the science and technology universities in Vietnam.
One of the most prominent programs of VinTech City is VinTech Fund—a grant
for creating innovative tech products from research, or in another word “bringing
lab ideas to market”, which has recently announced the first batch of applied
research projects receiving fundings of 86 billions VND in total (equivalent to
$3.7m). Learn more about them on www.vintechcity.com.
In brief, Vietnamese researchers, who works at prestigious research institutes
and universities, can contribute to the development of science and technology in

xix
xx Sponsors pages

Vietnam through research collaboration, tech product development, and educational

activities with domestic universities within the sponsorship framework of VinTech
Fund.
We believe that VinTech City can contribute to create a catalyst for “Make in
Vietnam” technology products in market and complete the mission of supporting
Vietnam startup and technology ecosystem.

Silver Sponsors

Established in 2008, National Foundation for Science and Technology

Development of Vietnam (NAFOSTED) has been one of the main funding agencies
for research in Vietnam. Affiliated with the Ministry of Science and Technology,
NAFOSTED’s mission is to build a favorable research environment that meets
international standards in order to enhance national Science & Technology
capacity, including improving research quality and developing high-quality human
resources. NAFOSTED funds both basic and applied research in all disciplines
from mathematics, physics, medical and biological sciences to engineering, earth
sciences, social sciences and humanities. Recently, NAFOSTED provides funds for
around 400 research projects with participations of 1.200–1.500 researchers each
year. With the aim of improving research capacity in Vietnam to meet international
standards, NAFOSTED also supports numbers of scientific activities, encouraging
international cooperation for Vietnamese scientists, through this contributes to
national economic growth and competitiveness.

Officially established in Vietnam in 2007 by a group of young Vietnamese

intellectuals living in Ukraine, Sun Group has now emerged as one of the leading
real estate and tourism developers in Vietnam with 61 members corporations and
more than 6000 employees nationwide. With a focus on strategically investing
across four main categories—Leisure Travel, Recreation and Entertainment, Real
Estate, Infrastructure—and placing an emphasis on the three core values of “Quality
—Luxury—Distinction”, Sun Group has created numerous prestigious and iconic
landmarks, many of which have been hailed as architectural masterpieces.
Sponsors pages xxi

Bronze Sponsors

STECHCO is an engineering firm specializing in building design, building

structure, and construction project management. Our firm also provides professional
services in the application of new construction technologies in Canada and the
United States. STECHCO’s rigorous and professional team is recognized and
appreciated for our expertise and experience in all types of residential, commercial,
industrial and institutional construction. We are members or licensees of profes-
sional orders and associations such as Order of Engineers of Quebec (OIQ),
Professional Engineer of Ontario (PEO), The Northwest Territories Association of
Professional Engineers and Geoscientists (NAPEG). Our professional team has
participated in more than 500 projects in Canada.

NUCE has undergone over 50 years of formation and development and over 60
years of training. NUCE becomes nowadays a leading centre for training, research,
technology transfer and creating high quality human resources for the country in the
field of civil engineering. In addition, NUCE was one of the first four universities in
Vietnam evaluated and accredited by HCERES (Haut Conseil de l’Evaluation de la
Recherche et de l’Enseignement Supérieur)—an international assessment organi-
sation for 5 years, without conditions until May 2022.
Contents

Keynote Lectures
Four Decades of Computing in Civil Engineering . . . . . . . . . . . . . . . . . 3
Hojjat Adeli
Industrialized Construction of Medium Span Concrete Bridges
Using Movable False Work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
Jose Turmo
Using Simulation to Estimate and Forecast Transportation Metrics:
Lessons Learned . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
L. R. Rilett
eHighway – An Infrastructure for Sustainable Road Freight
Transport . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35
Manfred Boltze
Innovation Reduces Risk for Sustainable Infrastructure . . . . . . . . . . . . 45
Suzanne Lacasse
Applying a National BIM Model to Vietnam’s National
Implementation of BIM: Lessons Learned from the UK-Vietnam
Collaboration for the Industry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57
Adam Matthews and Binh Ta

Advanced Modeling and Characterization of Structures

Advanced Analysis Software for Steel Frames . . . . . . . . . . . . . . . . . . . . 69
Phu-Cuong Nguyen, My Ngoc-Tra Lam, Duc-Duy Pham,
Trong-Nghia Nguyen and Phong Thanh Nguyen
Finite Element Modelling for Axially Loaded Concrete-Filled
Steel Circular Tubes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75
Duc-Duy Pham and Phu-Cuong Nguyen

xxiii
xxiv Contents

Fracture Analysis of Crack Propagation on the Diaphragm

of Steel Bridge Structures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81
Tran The Truyen, Nguyen Duc Hieu, Bui Thanh Tung,
Doan Bao Quoc and Nguyen Thuong Anh
A Study on Combination of Two Friction Dampers to Control
Stayed-Cable Vibration Under Considering its Bending Stiffness . . . . . . 87
Duy-Thao Nguyen and Duy-Hung Vo
Design Proposal and Behavior Simulation of Prestressed Concrete
Slab Track at Highway-Railway Grade Crossings . . . . . . . . . . . . . . . . . 93
Pham Van Hung, Tran The Truyen, Tran Anh Dung, Doan Bao Quoc,
Le Hai Ha and Nguyen Hong Phong
Effects of Random Road Roughness on Dynamic Impact Factor
of Multi-span Super T Girder Bridge with Link Slab due
to Moving Vehicles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 99
Xuan-Toan Nguyen and Duy-Thao Nguyen
Advanced Design Software for Steel Cable-Stayed Bridges
Using Nonlinear Inelastic Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105
Viet-Hung Truong and Seung-Eock Kim
Blind Source Separation Technique for Operational Modal Analysis
in Presence of Harmonic Excitation . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111
Van-Dong Do, Thien-Phu Le and Alexis Beakou
Investigation Into The Response Variability of A Higher-Order
Beam Resting on A Foundation Using A Stochastic Finite Element
Method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 117
Ta Duy Hien, Bui Tien Thanh, Nguyen Ngoc Long,
Nguyen Van Thuan and Do Thi Hang
Fatigue Analysis of Jack-up Leg Structures in Transit condition . . . . . . 123
Dinh Quang Cuong and Vu Dan Chinh
Finite element modeling of the TECCO protection system
for rock-fall under impact loading . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 131
Tran Van Dang, Tran Dong and Dennis Gross
The study on the behavior of the simply supported beam steel
bridge structure without the intermediate bracing system
in the construction stage . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 137
Van Nam Le, Anh Rin Nguyen, Lien Thuc Pham and Ngoc Thi Huynh
Seismic behaviour of rammed earth walls: a time history analysis . . . . . 143
Quoc-Bao Bui and Tan-Trung Bui
Contents xxv

Analysis of Two-Directional Seismic Deterioration of Steel

Box Columns . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 149
Tran Tuan Nam
Numerical analysis of double-layered asphalt pavement behaviour
taking into account interface bonding conditions . . . . . . . . . . . . . . . . . . 155
Minh Tu Le, Quang-Huy Nguyen and Mai Lan Nguyen
Parametric numerical study on a novel energy harvester using
iron-gallium alloy and strain response . . . . . . . . . . . . . . . . . . . . . . . . . . 161
Tuan Minh Ha, Saiji Fukada and Toshiyuki Ueno
Numerical investigation of derailment loading on composite fibre
transoms for implementation in the Sydney Harbour Bridge . . . . . . . . . 167
Olivia Mirza, Maryam Hosseini and Micah Fountain
Finite Element Analysis of Reinforced Concrete beams subjected
to combined actions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 173
Tuan-Anh Nguyen, Quang-Huy Nguyen and Hugues Somja
Resistance of Cross Laminated Timber Members Under Axial
Loading—A Review of Current Design Rules . . . . . . . . . . . . . . . . . . . . 179
Xin Li, Mahmud Ashraf, Mahbube Subhani, Bidur Kafle and Paul Kremer
A Current-State-of-the-Art on Design Rules Vs Test Resistance
of Cross Laminated Timber Members Subjected to Transverse
Loading . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 185
Xin Li, Mahbube Subhani, Mahmud Ashraf, Bidur Kafle and Paul Kremer
A Further Study on Stay Cable Galloping Under Dry Weather
Condition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 191
Vo Duy Hung and Nguyen Duy Thao
Stress-Dependent Permeability of the Fractured Rock Masses:
Numerical Simulation Based on the Embedded Fracture
Continuum Approach . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 197
Hong-Lam Dang, Duc-Phi Do and Dashnor Hoxha
Improved Rigid-Plastic Method for Predicting the Ultimate
Strength of Concrete Walls Restrained on Three Sides . . . . . . . . . . . . . 203
J. H. Doh, N. M. Ho and T. Peters
Flexural Capacity Accounting for SBHS500 Steel of Composite
Bridge Girders . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 209
Dang Viet Duc and Yoshiaki Okui
The Inadequacies of the Existing Structural Health Monitoring
Systems for Cable Stayed Bridges in Vietnam . . . . . . . . . . . . . . . . . . . . 215
Luong Minh Chinh
xxvi Contents

Riding Comfort Assessment of High-Speed Trains Based

on Vibration Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 221
Van Nguyen Dinh, Ki-Du Kim and Dinh Tuan Hai
Assessment of methods of riprap size selections as scour
countermeasures at bridge abutments and approach embankments . . . . 227
Huy Quang Mai and Noi Thi Doan
A review on protection methods against debris accumulation
for bridge in mountain areas . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 233
Huy Quang Mai and Phong Dang Nguyen
Numerical modelling of thermo-mechanical performance
of small-scale CFRP reinforced concrete specimen using near
surface mounted reinforcement method . . . . . . . . . . . . . . . . . . . . . . . . . 239
Phi Long Nguyen, Xuan Hong Vu and Emmanuel Ferrier
Experimental characterization of multi-full-culm bamboo to steel
connections . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 245
Nischal P. N. Pradhan, Themelina S. Paraskeva
and Elias G. Dimitrakopoulos
FE modelling of RC frames with Link Column Frame System
under in-plane loading . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 251
J. Joel Shelton, G. Hemalatha and V. Venkatesh
Mechanical Behaviour of Four-Leg Base Configuration
on Cold-Formed Steel Lattice Column . . . . . . . . . . . . . . . . . . . . . . . . . . 257
Mohd Syahrul Hisyam Mohd Sani and Fadhluhartini Muftah
Finite Element Simulation of Member Buckling of Cold-rolled
Aluminium Alloy 5052 Channel Columns . . . . . . . . . . . . . . . . . . . . . . . 263
Ngoc Hieu Pham, Cao Hung Pham and Kim J. R. Rasmussen
Lateral Buckling Tests of Cold-rolled Aluminium Alloy 5052 Zee
Beams . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 269
Ngoc Hieu Pham, Cao Hung Pham and Kim J. R. Rasmussen
Elastic Buckling Solution for Perforated Thin-walled Channel
Sections in Shear with an Aspect Ratio of 2.0 . . . . . . . . . . . . . . . . . . . . 275
Duy Khanh Pham, Cao Hung Pham and Gregory J. Hancock
The Behaviour of Cold-formed Channel Sections with Elongated
Web Holes in Shear . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 281
Duy Khanh Pham, Cao Hung Pham, Song Hong Pham
and Gregory J. Hancock
Experimental Investigation of Cold-Rolled Aluminium Alloy 5052
Columns Subjected to Distortional Buckling . . . . . . . . . . . . . . . . . . . . . 287
Le Anh Thi Huynh, Cao Hung Pham and Kim J. R. Rasmussen
Contents xxvii

Numerical Simulation of Cold-Rolled Aluminium Alloy 5052

Columns Subjected to Distortional Buckling . . . . . . . . . . . . . . . . . . . . . 293
Le Anh Thi Huynh, Cao Hung Pham and Kim J. R. Rasmussen
Use of Kriging Metamodels for Seismic Fragility Analysis of
Structures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 299
Cong-Thuat Dang, Thanh Tran, Duy-My Nguyen, My Pham
and Thien-Phu Le
Application of Weighted Latin Hypercube Sampling in Stochastic
Modelling of Shear Strength of RC Beams . . . . . . . . . . . . . . . . . . . . . . . 305
Ngoc Linh Tran and Khuong Le-Nguyen
Predicting onset and orientation of localisation bands using
a cohesive-frictional model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 311
Linh A. Le, Giang D. Nguyen and Ha H. Bui
Non-linear deformational analysis of reinforced concrete frame . . . . . . . 317
Tran Thi Thuy Van and Vu Thi Bich Quyen
Reliability Evaluation of Eurocode 4 for Concrete-Filled Steel
Tubular Columns . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 323
Huu-Tai Thai and Son Thai

Sustainable Construction Materials and Technologies

Preparation of low cement ultra-high performance concrete . . . . . . . . . 331
Norzaireen Mohd Azmee and Nasir Shafiq
Matrix dependent piezoresistivity responses of high performance
fiber-reinforced concretes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 337
Duy-Liem Nguyen, Thi-Bich-Nga Vu, Huynh-Tan-Tai Nguyen
and Van-Ben Nguyen
Influence of elastic modulus under uniaxial tension and compression
on the first-cracking flexural properties of UHPFRCs . . . . . . . . . . . . . . 343
Duy-Liem Nguyen and Minh-Thuan Duong
Investigation on shear resistances of short beams using HPFRC
composited normal concrete . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 349
Duy-Liem Nguyen, Van-Toi Do, Minh-Phung Tran and Luu Mai
Application of Fluidized Power Coating for Propellers . . . . . . . . . . . . . 355
Tuan Phan Anh and Huong Pham Thi Thanh
Design of a Drinking Water Disinfection Systems using Ultraviolet
Irradiation and Electrolysis Cell . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 361
Tuan Phan Anh and Tan Nguyen Minh
xxviii Contents

The effect of mineral admixture on the properties of the binder

towards using in making pervious concrete . . . . . . . . . . . . . . . . . . . . . . 367
Nguyen Van Dong, Pham Huu Hanh, Nguyen Van Tuan,
Phan Quang Minh and Nguyen Viet Phuong
Horizontal response of base-isolated buildings supported to high
damping rubber bearings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 373
Nguyen Anh Dung, Le Trung Phong and Tran Minh
A Study on Behavior of Reinforcement Concrete Beam using
the Recycled Concrete . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 379
Anh-Thang Le, Thanh-Hung Nguyen and Cong-Vu-Duc Phan
Geochemical modelling for prediction of chloride diffusion
in concrete exposed to seawater . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 385
Hoang Long Nguyen, Van Quan Tran, Long Khanh Nguyen,
Tuan Anh Pham, Quoc Trinh Ngo and Van Loi Giap
Application of Asphalt Concrete using Limestone with
Cement & Admixture . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 391
Nguyen Ba Hoang, Pham Van Hung, Tran Viet Khanh
and Nguyen Viet Huy
Experimental Investigation of a Self-powered Magnetorheological
Damper for Seismic Mitigation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 397
C Daniel, G Hemalatha, L Sarala, D Tensing, S Sundar Manoharan
and Xian-Xu Bai
Effect of Ground Blast Furnace Slag in Replacement of Cement
in Ternary Binder on Performance of Sand Concrete . . . . . . . . . . . . . . 403
Thanh Sang Nguyen and Long Hai Chu
Experimental Study on Effect of Ground Granulated Blast Furnace
Slag of Strength and Durability of Sand Concrete . . . . . . . . . . . . . . . . . 409
Nguyen Thanh Sang and Nguyen Tan Khoa
Mechanical properties of fly ash based geopolymer concrete using
only steel slag as aggregate . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 415
Dong Van Dao and Son Hoang Trinh
Effect of fly ash on the mechanical properties and drying shrinkage
of the cement treated aggregate crushed stone . . . . . . . . . . . . . . . . . . . . 421
Ho Van Quan, Nguyen Van Tuoi and Chau V. Nguyen
Strength of Granulated Blast Furnace Slag during Hydration
Reaction Process . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 427
Tran Thanh Nhan, Hiroshi Matsuda, Tran Xuan Thach, Nguyen Dai Vien
and Ho Trung Thanh
Contents xxix

Use of Coal Ash of Thermal Power Plant for Highway Embankment

Construction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 433
Tung Hoang, Viet Phuong Nguyen and Hong Nam Thai
Impact of the Porosity of Coarse Aggregates on the Structuration
of the Paste-Aggregate Interface: Elementary Model Study . . . . . . . . . . 441
Tran Duc Long, Cassagnabère Franck and Mouret Michel
Strength and Engineering Properties of Cementless Paste Produced
by GGBFS and MgO . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 447
Chao-Lung Hwang, Duy-Hai Vo, Khanh-Dung Tran Thi
and Mitiku Damtie Yehualaw
A Non-paraffinic PCM Modified Textile Reinforced Concrete
Sand-Wich Panel . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 453
Zakaria Ilyes Djamai, Amir Si Larbi and Ferdinando Salvatore
Design Method for Optimizing Geopolymer Concrete Proportions
Utilising Entirely Steel Slag Aggregates . . . . . . . . . . . . . . . . . . . . . . . . . 459
Dong Van Dao and Son Hoang Trinh
Experimental Measurement and Modeling of the Complex
Poisson’s Ratio of Bituminous Mixtures . . . . . . . . . . . . . . . . . . . . . . . . . 465
Quang Tuan Nguyen and Thi Thanh Nhan Hoang
Evaluation of Macro-Synthetic Fibre Reinforced Concrete
as a Sustainable Alternative for Railway Sleepers . . . . . . . . . . . . . . . . . 471
Christophe Camille, Dayani Kahagala Hewage, Olivia Mirza,
Fidelis Mashiri, Brendan Kirkland and Todd Clarke
The Repairing Effects of Lithium Silicate Based Material
to the Surface of Hardened Concrete . . . . . . . . . . . . . . . . . . . . . . . . . . . 477
Nguyen Xuan Tung, Ryoma Kitagaki and Yoshiro Yamakita
Research on Fabricating Flowforms from Ultra-High Performance
Concrete with Local Admixtures for Use in Processing
Contaminated Water . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 483
Thi Thuy Ha Ung, Tuan Hung Pham and Tho Bach Leu
Research on Application of Flowforms in Combination
with Planted Constructed Wetland for Improving Water Quality
of Urban Polluted Lakes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 489
Thi Thuy Ha Ung, Tuan Hung Pham, Tho Bach Leu,
Thi Hien Hoa Tran and Hong Nhung Chu
A Model of Local Kinetics of Sorption to Understand the Water
Transport in Bio-based Materials . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 495
N. Reuge, F. Collet, S. Pretot, S. Moissette, M. Bart and C. Lanos
xxx Contents

Air Permeability of Cover Concrete Quality of Precast Box Culverts

Affected by Casting Direction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 501
May Huu Nguyen, Kenichiro Nakarai and Saeko Kajita
Effects of Amounts and Moisture States of Clay-Brick Waste
as Coarse Aggregate on Slump and Compressive Strength
of Concrete . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 507
Phuong Trinh Bui, Xuan Nam Nguyen, My Ngoc Tang, Yuko Ogawa
and Kenji Kawai
Investigation of the Use of Reclaimed Asphalt Pavement
as Aggregates in Roller Compacted Concrete for Road Base
Pavement in Vietnam . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 513
Thi Huong Giang Nguyen, Tien Dung Nguyen, Trung Hieu Tran,
Van Dong Dao, Xuan Cay Bui and Mai Lan Nguyen
Comparison between Critical Path Method (CPM) and Last
Planners System (LPS) for Planning and Scheduling METRO
Rail Project of Ahmedabad . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 519
Viraj Parekh, Karan Asnani, Yashraj Bhatt and Rahul Mulchandani
Study and propose the size of cement concrete slabs for airport
road surface in Vietnam conditions . . . . . . . . . . . . . . . . . . . . . . . . . . . . 525
Nguyen Duy Dong, Nguyen Vu Viet and Nguyen Quoc Van
Experimental study of the thermomechanical behavior of the carbon
textile reinforced refractory concrete subjected to the constant
load and temperature heating . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 531
Manh Tien Tran, Xuan Hong Vu and Emmanuel Ferrier
A study on improvement of early - age strength of super sulfated
cement using phosphogypsum . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 537
Nguyen Ngoc Lam
Effect of calcium sulfate type and dosage on early strength
and porosity of self-leveling underlayments . . . . . . . . . . . . . . . . . . . . . . 543
Nguyen Ngoc Lam, Elodie Prud’homme and Jean-François Georgin
Heat resistant mortar using Portland cement and waste clay bricks . . . 549
Nguyen Ngoc Lam
Performance of Recycled Coarse Aggregate Concrete with Different
Nylon Fiber Content . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 555
Seungtae Lee, Youngkyu Chu, Sangwook Ha, Yong Kim
and Jaehong Jeong
Contents xxxi

Air permeability coefficients of expansive concrete confined

by rebars . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 561
Le Anh Van, Kenichiro Nakarai, Nguyen Huu May, Yasutaka Kubori,
Toshikazu Matsuyama, Hajime Kawakane and Shintaro Tani
Application of Six Sigma on METRO Rail Construction Project . . . . . . 567
Viraj Parekh, Kushal Solanki and Nishit Prajapati
Experimental investigation of loading rate effects on the shear
capacity of reinforced concrete deep beams . . . . . . . . . . . . . . . . . . . . . . 573
La Vanny, Kenichiro Nakarai, Halwan Alfisa Saifullah and Asuka Mizobe
Fatigue characterization of conventional and high rutting
resistance asphalt mixtures using the cyclic indirect tensile test . . . . . . . 579
H. T. Tai Nguyen, Anh Thang Le, Vu Tu Tran and Duy Liem Nguyen
Sustainable Measures to Achieve Better River-Crossing Access
in the Rural Areas of Developing Countries . . . . . . . . . . . . . . . . . . . . . . 585
Yoshinori Fukubayashi and Makoto Kimura
Assessment of Design Guidelines for Fiber-Reinforced Polymer
Shear Contribution of Prestressed Concrete Beams Strengthened
by Fiber-Reinforced Polymer Sheets . . . . . . . . . . . . . . . . . . . . . . . . . . . 591
Dien Ngoc Vo-Le, Chinh Ho-Huu and Long Nguyen-Minh
Development of Disaster Resilient and Sustainable University
Framework: Case of Bandung Institute of Technology (ITB) . . . . . . . . . 597
Krishna Suryanto Pribad, Rahma Hanifa, Eliya Hanafi
and Giovanni Pradipta
Experimental study on the valorization of poplar by-products
in cement-based materials . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 603
C. Djelal, J. Page, Y. Vanhove and H. Kada
Horizontal response of base-isolated buildings supported
to high damping rubber bearings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 609
Nguyen Anh Dung, Le Trung Phong and Tran Minh
Evaluating the change in the properties of bitumen during production,
transport and construction of hot mix asphalt in the conditions
of Vietnam . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 615
K. H. Quang Trinh, T. Nhan Tran and H. T. Tai Nguyen
Creep behavior and rutting resistance of asphalt pavements
by experimental testing and Finite Element modelling . . . . . . . . . . . . . . 621
Van Bac Nguyen, Chau Phuong Ngo, Ngoc Bay Pham,
Thanh Phong Nguyen, Van Phuc Le and Van Hung Nguyen
xxxii Contents

Influence of High Temperature on Non-Silicate Based Activated

Blast Furnace Slag . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 627
Virendra Kumar, Amit Kumar and B. K. Prasad
Influence of Linz Donawitz slag aggregates on strength development
of concrete . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 633
Brajkishor Prasad, Virendra Kumar and Prince Singh
Investigation of the effect of c irradiations on the mechanical
and physico-chemical properties of cementitious materials:
from the nanoscale to mortars’scale . . . . . . . . . . . . . . . . . . . . . . . . . . . . 639
Maxime Robira, Benoît Hilloulin, Ahmed Loukili, Gildas Potin,
Xavier Bourbon and Abdessalam Abdelouas

Geotechnics for Environment and Energy

Evaluation of the At-Rest Lateral Earth Pressure Coefficient of Fibre
Reinforced Load Transfer Platform and Columns Supported
Embankments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 647
Cong Chi Dang and Liet Chi Dang
Nonlinear Behavior Analysis of SFRC Foundation Considering
Homogeneous and Inhomogeneous Soil Interactions . . . . . . . . . . . . . . . 653
Tran-Trung Nguyen, Radim Cajka, Phu-Cuong Nguyen
and Thanh-Tuan Tran
Analyses on drainage capacity and sliding resistance of large
diameter vertical wells for deep-seated landslide stabilization . . . . . . . . 659
Ba Thao Vu, Van Minh Pham, Quoc Dung Nguyen,
Huy Vuong Nguyen and Thi Thanh Huong Ngo
Shape effects in the dynamical response of flints/boulders during
pile driving – a numerical study . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 665
Emilio Nicolini and Ivan Terribile
The failure envelope of a new shaped cross-plates foundation
for deep-water soft soils . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 673
Emilio Nicolini, Alessandro Alberio and Matteo Castelletti
Enhanced Oil Production in the lower Miocene reservoir
by Multistage Fracturing, Offshore Viet Nam . . . . . . . . . . . . . . . . . . . . 679
Huu Truong Nguyen, W. Bae and Van Hung Nguyen
Properties of geopolymer modified sludge generated in landslide
area designed by Taguchi method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 685
Minh Chien Vu, Tomoaki Satomi and Hiroshi Takahashi
Contents xxxiii

Granular flows through a model-scale forest: Influence of tree

density and implications for landslide mitigation . . . . . . . . . . . . . . . . . . 691
H. T. Luong, J. Baker and I. Einav
Dynamic properties of loose sand using numerical analysis-A case
at Hong Thai Tay coal transportation road project (Vietnam) . . . . . . . . 697
Nguyen Phi Hung, Nguyen Van Dung and Tan Do
Application of Box-Behnken Design in Optimization of Fracture
Treatment Design for Lower Oligocene Reservoir, Offshore
Viet Nam . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 703
Huu Truong Nguyen, W. Bae and Van Hung Nguyen
Experimental and numerical assessments of seepage effect
on embankment behaviours by the time . . . . . . . . . . . . . . . . . . . . . . . . . 709
Thanh Quang Nguyen, Truong Linh Chau and The Hung Nguyen
A numerical homogenized law using discrete element method
for continuum modelling of boundary value problems . . . . . . . . . . . . . . 715
Trung Kien Nguyen, Jacques Desrues, Gaël Combe and Duc Hanh Nguyen
Influence of heterogeneous fractured fault damage zones
on shear failure onset during fluid injection . . . . . . . . . . . . . . . . . . . . . . 721
Trung Kien Nguyen, Jeremy Rohmer and Ba Thao Vu
Influence of Compaction Factor on the mechanical behavior
of the soils used for the earth dams in the North Central region,
Vietnam . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 727
Van Toan Tran and Van Hien Tran
Enhanced oil recovery: A selection technique for the energy
and recovery of Bach Ho field in Vietnam . . . . . . . . . . . . . . . . . . . . . . . 733
Nguyen Van Hung, Nguyen Minh Quy, Hoang Long, Le Quoc Trung
and Le Vu Quan
Overview of Geomechanics and its applications to petroleum industry
– a case study for minimum overbalance pressure calculation . . . . . . . . 739
Nguyen Van Hung
Numerical model of hydro–mechanical coupling DEM–PFV
and application for simulation of settlement of soil saturated
in embankment due to static loading . . . . . . . . . . . . . . . . . . . . . . . . . . . 745
Anh Tuan Tong and Nguyen Hoang Phuong Luong
Abnormal pore pressure and fracture pressure prediction for Miocene
reservoir rock, field X in Vietnam . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 751
Nguyen Van Hung
xxxiv Contents

Study the influence of adherence edge to steel strip and soil

interaction in Mechanically Stabilized Earth Wall with
a self-made strip . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 757
Truong-Linh Chau and Thu-Ha Nguyen
Formation of Rust Scales on Mild Steel Pile in Brackish Water
of Phu My Port - Vietnam . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 763
Vinh-Dat Vuong, Anh Quang Vu, Marina Kawai, Kazuhito Tsujimoto,
Thang Van Le and Nguyen Danh Thao
Modelling Stress Distribution Around Boreholes . . . . . . . . . . . . . . . . . . 769
Khanh Do Quang, Thanh Nguyen Thi Tam, Phuc Kieu,
Nhan Vo Huynh and Quang Hoang Trong
A mathematical model for fault activation by water injection . . . . . . . . 775
Thanh Son Nguyen, Jonny Rutqvist, Bastian Graupner and Yves Gugliemi
A model for soil-structure interaction - Application to small
modular reactors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 781
L. Cui, M. Fall, N. Orbovic and T. S. Nguyen
Full-scale Pullout Testing of Ground Anchors to Evaluate
the Applicability of French Design Practice TA95 for Vietnam . . . . . . . 787
Nguyen Ngoc Thanh and Nguyen Phuong Duy
Durability Evaluation of a Geothermal Grout . . . . . . . . . . . . . . . . . . . . 793
Monnot Pascal, Poinclou Christophe, Blanc Philippe, Rojo Amandine,
Molez Laurent and Lanos Christophe
Assessment on the performance of EPB-TBM in the construction
of pilot metro line in Ho Chi Minh city (Vietnam) . . . . . . . . . . . . . . . . . 799
Thu-Hang Tran and Manh-Tuan Hoang
Reusability of muds dredged from lakes in Hanoi City from
its geotechnical engineering characteristics . . . . . . . . . . . . . . . . . . . . . . . 805
Pham Tri Thuc and Phan Huy Dong
Numerical study of pile reinforced slope-A case at Khe Cham
coal preparation construction site project (Vietnam) . . . . . . . . . . . . . . . 811
Nguyen Van Dung, Nguyen Phi Hung and Tan Do
Experimental Correlations for the Swelling Pressure of Expansive
Clays in the City of Tebessa, Algeria . . . . . . . . . . . . . . . . . . . . . . . . . . . 817
Adel Djellali, Behrooz Saghafi and Mohamed Salah Laouar
Design of Experiments (DOE) techniques to Predict Swelling
Pressure of Expansive Soils In Tebessa (Algeria) . . . . . . . . . . . . . . . . . . 823
Yacine Berrah, Abderrahmane Boumezbeur, Nouar Charef
and Serhane Brahmi
Contents xxxv

Analysis of Landslides in the Region of Souk Ahras (Zaarouria Sector)

North- east of Algeria Using pseudo-static method . . . . . . . . . . . . . . . . . 829
Nouar Charef, Yacine Berrah and Abderrahmane Boumezbeur
Solving the stability problem of vertical slope according
to the effective stress field . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 835
Do Thang
Finite Element Simulation of Water Content-Influenced Progressive
Failure of Sensitive Clays . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 841
Nhat-Phi Doan, Seung-Wook Woo, Yao-Long Hou and Sung-Sik Park
Non-microbial carbonate precipitation as an improvement
technique of sand . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 847
Sung-Sik Park, Trung-Tri Le, Seong-Heon Kim and Zhenzhen Nong
Modeling on Cuttings Transport in Inclined and Horizontal
Well Drilling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 853
Do Quang Khanh, Truong Trong Tuan Dat, Kieu Phuc,
Tran Thi Mai Huong and Hoang Trong Quang
Study on Cement-Treated Soil with RoadCem Additive
in Construction of Rural Roads: A Case Study in Viet Nam . . . . . . . . . 859
Ba Thao Vu, Thi Thanh Huong Ngo, Quoc Dung Nguyen,
Anh Quan Ngo and Lanh Si Ho
Application of Steel Pipe Piles with Wings installed in Soil
Cement Column for Building Structure in Vietnam . . . . . . . . . . . . . . . . 865
Pham Hoang Kien and Daisuke Ito
Modelling the hydromechanical behaviour of a granular expansive
clayey soil upon hydration using discrete element method . . . . . . . . . . . 871
Benjamin Darde, Jean-Noël Roux, Patrick Dangla, Jean-Michel Pereira,
Anh Minh Tang, Jean Talandier and Minh Ngoc Vu
Spatially-varying non-stationary seismic bedrock motions . . . . . . . . . . . 877
Van Nguyen Dinh and Biswajit Basu
An experimental investigation on bearing capacity behavior
of rectangular footing over reinforced soil slope . . . . . . . . . . . . . . . . . . 883
Raju Sarkar, Amit Kumar Shrivastava and Ankur Mudgal
Meshfree SPH modelling of shrinkage induced cracking
in clayey soils . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 889
Hieu T. Tran, Nhu H. T. Nguyen, Giang D. Nguyen and Ha H. Bui
Numerical predictions of post-flow behaviour of granular
materials using an improved SPH model . . . . . . . . . . . . . . . . . . . . . . . . 895
H. H. Bui and Giang D. Nguyen
xxxvi Contents

Stochastic Site Response Analysis in Consideration with Various

Probability Distributions of Geotechnical Properties . . . . . . . . . . . . . . . 901
Thanh-Tuan Tran, Phu-Cuong Nguyen, Seung-Ryong Han
and Dookie Kim
Analysis of strain measurements in large span soil-steel bridge
structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 907
Czesław Machelski, Adam Czerepak and Piotr Tomala

Urban Planning, Transport and Environment

A Landscape Structure Proposal for a Sustainable Development – Quy
Nhon City . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 915
Rute Sousa Matos, Huyen Thi Nguyen, Tho Van Phan
and Phan Thi Le Thuy
Travel behavior change patterns under adverse weather conditions - A
case study from Ho Chi Minh City (HCMC), Vietnam . . . . . . . . . . . . . 921
Anh Tuan Vu and Thi Thanh Huong Nguyen
Traffic Impact Assessment of Infrastructure Development Projects
for Sustainable Urban Growth . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 927
Trinh Dinh Toan and Dao Van Dong
Reasons for cyclists denying new-built bicycle lanes . . . . . . . . . . . . . . . . 933
Nguyen Duc-Nghiem, Nguyen Viet-Phuong and Hisashi Kubota
Sustainable and Health-Oriented Transport Planning and Urban
Planning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 941
Karin Menges and Manfred Boltze
Integrated Transport Planning for Sustainable Urban
Development – Singapore’ Approach and Lessons for Vietnam . . . . . . . 947
Trinh Dinh Toan and Dao Van Dong
Transitioning Different Stages of Transport Planning in Urban Areas:
Experiences of Singapore and Vietnam . . . . . . . . . . . . . . . . . . . . . . . . . 953
M. C. Rojas Lopez, T. D. Toan and Y. D. Wong
Analysis of Mode Choice Behavior under Adverse Weather
Conditions Using RA and SA Surveys - A Case Study
from Ho Chi Minh City (HCMC), Vietnam . . . . . . . . . . . . . . . . . . . . . . 959
Anh Tuan Vu and Thi Thanh Huong Nguyen
A Study on Motorcycles Overturning due to Strong Wind
and its Warning System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 965
Vo Duy Hung and Do Anh Vu
Contents xxxvii

Bus signal priority by active signal program – A case study

in Ho Chi Minh City (HCMC), Vietnam . . . . . . . . . . . . . . . . . . . . . . . . 971
Thi Thanh Huong Nguyen
Modeling land use change based on Remote Sensing, GIS
and Algorithm Cellular Automata decision support sys-tem
for urban sustainability planning in Quy Nhon, Binh Dinh
province central Vietnam . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 977
Tho Van Phan, Tu Anh Ngo, Huyen Thi Nguyen, Rute Sousa Matos
and Trang Thuc Dang
Modeling and applying heuristics for optimization of solid
waste collection under consideration of vehicle capacity . . . . . . . . . . . . . 983
Hai Dung Dinh
Effective Capital Usage Solutions for Road Infrastructure
Maintenance in Vietnam . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 989
Nguyen Thi Tuyet Dung and Nguyen Quynh Sang
Perspectives of the use of GPS in travel survey: Research
on identification of missing trips in a GPS pilot survey in Hanoi . . . . . . 995
Thanh Tu Nguyen
Satelite image application for assessing the effect of urbanisation
to temperature at Danang City, Viet Nam . . . . . . . . . . . . . . . . . . . . . . . 1001
Ngoc Duong Vo and Quang Binh Nguyen
Dambreak Simulation at Thuan Ninh Reservoir, Binhdinh,
Vietnam . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1007
Ngoc Duong Vo, Quoc Hoang Anh Huynh and Trung Dung Vo
Flood Mapping for Downstream of Ben Hai River, Vietnam . . . . . . . . . 1013
Ngoc Duong Vo and Truong Son Ho
A Framework for Last Planner System Implementation
in Egypt . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1019
Eyad Aboseif and Rana Khallaf
Climate Change Impact on Urban Flooding in Quynhon City,
Vietnam . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1025
Duc Phuoc Vo, Muhammed A Bhuiyan and Ngoc Duong Vo
Health Impacts of Traffic-related Air Pollution: Cause-effect
Relationships and Mitigating Measures . . . . . . . . . . . . . . . . . . . . . . . . . 1031
Huong Le Thi
Evaluation of the Effects of Traffic Management at School Areas . . . . . 1037
An Minh Ngoc and Khuat Viet Hung
xxxviii Contents

Policy Implications from Traffic Accident Analysis: A Study

Case from Vietnam . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1043
An Minh Ngoc and Truong Thi My Thanh
Analysis of Tourist Travel Behavior and Recommendation for Active
Transport Encouragement Strategies, the Case of Hue City . . . . . . . . . 1049
Truong Thi My Thanh, Dong Van Dao and Hai-Bang Ly
The Impacts of Urbanization on Urban Flooding in Danang City,
Vietnam . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1057
Thuy Nga To, Huy Cong Vu and Hung Le
Factors Influencing the Choice of Bike Sharing System: An
Investigation of Visitors and Local People in Vietnam . . . . . . . . . . . . . . 1063
Nguyen Thanh Tu and Le Thu Huyen
Flood risk assessment in the planning of new urban in Quangnam
province, Vietnam . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1069
Thuy Nga To, Hung Le and Huy Cong Vu
Low Impact Development, a novel technique for cutting
down urban flooding in Quynhon city, Vietnam . . . . . . . . . . . . . . . . . . . 1075
Nguyen Duc Phuoc Vo, Muhammed A. Bhuiyan and Ngoc Duong Vo
Vehicle Usage/Ownership control for a Sustainable Transport
system in the Motorcycle Dependent Cities . . . . . . . . . . . . . . . . . . . . . . 1081
Le Thu Huyen and Nguyen Thanh Tu
Typhoon Wave Modelling for Viet Nam Central Region . . . . . . . . . . . . 1087
Ngoc Duong Vo, Quang Binh Nguyen and Cong Phong Nguyen
Application of Win-TR55 for Calculating Designed Flood
in Small Watersheds in Mountainous Region of Vietnam . . . . . . . . . . . 1093
Noi Thi Doan
Using HCM Method in Estimation the Highway Capacity
and Recommendation Traffic Management for a Section
of National Highway #5 - VietNam . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1099
Tran Trung Hieu and Trinh Xuan Lam
Green Infrastructure Modelling for Assessment of Urban Flood
Reduction in Ho Chi Minh city . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1105
Nguyen Hong Quan, Nguyen Duy Hieu, Tran Thi Van Thu,
Matthew Buchanan, Nguyen Duc Canh, Mariana da Cunha Oliveira
Santos, Pham Dang Manh Hong Luan, Tran Tuan Hoang,
Huynh Luu Trung Phung, Khuu Minh Canh and Mitchel Smith
Review of Unmanned Aerial Vehicles (UAVs) Operation
and Data Collection for Driving Behavior Analysis . . . . . . . . . . . . . . . . 1111
H. Q. Pham, M. Camey, K. D. Pham, K. V. Pham and L. R. Rilett
Contents xxxix

New cooperative approach between universities and enterprises,

towards a reality urban planning. Lesson learnt from Nhon Ly,
Quy Nhon city . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1117
Thai Huyen Nguyen and Thuc Trang Dang
Smart Monitoring for Urban Biodiversity Preservation . . . . . . . . . . . . . 1123
Thi Hai Yen Pham, Isam Shahrour, Ammar Aljer, Alain Lepretre,
Celine Pernin and Sana Ounaies

Data mining and Machine Learning

A Daily Work Report Based Approach for Schedule Risk Analysis . . . . 1131
Chau Le and H. David Jeong
A Deep Learning-Based Procedure for Safety Evaluation
of Steel Frames Using Advanced Analysis . . . . . . . . . . . . . . . . . . . . . . . 1137
Manh-Hung Ha, Quang-Viet Vu and Viet-Hung Truong
Prediction of buckling coefficient of stiffened plate girders
using deep learning algorithm . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1143
George Papazafeiropoulos, Quang-Viet Vu, Viet-Hung Truong,
Minh-Chinh Luong and Van-Trung Pham
Monitoring Pedestrian Flow on Campus with Multiple Cameras
using Computer Vision and Deep Learning Techniques . . . . . . . . . . . . . 1149
Peter Kok-Yiu Wong and Jack C.P. Cheng
An Image Mosaicking Method Base on the Curvature of Cost
Curve for Tunnel Lining Inspection . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1155
Cuong Nguyen Kim, Kei Kawamura, Masando Shiozaki
and Amir Tarighat
Application of Neural Network to predict the workability
Parameters of Self-Compacting Concrete . . . . . . . . . . . . . . . . . . . . . . . . 1161
Cuong H. Nguyen, Linh H. Tran and Khoa N. Ho
Development of Artificial Neural Networks for Prediction
of Compression Coefficient of Soft Soil . . . . . . . . . . . . . . . . . . . . . . . . . . 1167
Binh Thai Pham, Manh Duc Nguyen, Hai-Bang Ly, Tuan Anh Pham,
Vu Hoang, Hiep Van Le, Tien-Thinh Le, Hung Quang Nguyen
and Gia Linh Bui
Real-time detection of asymmetric surface deformation and field stress
in concrete-filled circular steel tubes via multi-vision method . . . . . . . . . 1173
Yunchao Tang, Kuangyu Huang, Lijuan Li, Xiangjun Zou
and Wenxian Feng
xl Contents

A Possibility of AI Application on Mode-choice Prediction

of Transport Users in Hanoi . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1179
Truong Thi My Thanh, Hai-Bang Ly and Binh Thai Pham
Machine learning based tool for identifying errors in CAD to GIS
converted data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1185
Mohamed Badhrudeen, Nalin Naranjo, Ali Mohavedi and Sybil Derrible
Development of 48-hour Precipitation Forecasting Model using
Nonlinear Autoregressive Neural Network . . . . . . . . . . . . . . . . . . . . . . . 1191
Tien-Thinh Le, Binh Thai Pham, Hai-Bang Ly, Ataollah Shirzadi
and Lu Minh Le

Building Information Modeling

Building Information Modeling Based Optimization of Steel
Single-Plate Shear Connections Using Differential Evolution
Algorithm . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1199
Tran-Hieu Nguyen and Anh-Tuan Vu
BIM application for the design consultant on the irrigation
and hydropower projects in Vietnam . . . . . . . . . . . . . . . . . . . . . . . . . . . 1205
Van Toan Tran, Huu Hue Nguyen, Tuan Long Chu, Lam Tuan Mai,
Van Do Nguyen, Dinh Tai Nguyen and The Minh Ngoc Vu
BIM Adoption in Construction Projects Funded with State-managed
Capital in Vietnam: Legal Issues and Proposed Solutions . . . . . . . . . . . 1211
Thuy-Ninh Dao, The-Quan Nguyen and Po-Han Chen
BIM-based innovative bridge maintenance system using augmented
reality technology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1217
NgocSon Dang and ChangSu Shim
Perspectives on BIM Profession of BIM Specialists and non-BIM
Specialists: Case Study in Vietnam . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1223
Ngoc Quyet Le, Michael Er, Shankar Sankaran and Ngoc Binh Ta
The contextual influence on Building Information Modelling
implementation: A cross-case analysis of infrastructure projects
in Vietnam and Norway . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1229
Nam Bui
Application of BIM tools in technician training, a case of Ho Chi
Minh City Construction College, Vietnam . . . . . . . . . . . . . . . . . . . . . . . 1235
Tung Lam Nguyen and Thi Xuan Nguyen
Bridge assessment for PSC Girder Bridge using Digital Twins
Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1241
NgocSon Dang and ChangSu Shim
Contents xli

Application of Building Information Modelling, Extended tracking

technique and Augmented Reality in Building Operating
Management . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1247
Thanh Ban Nguyen, Anh Binh Tran, Manh Tuan Nguyen, Van Hoan Pham
and Khuong Le-Nguyen
Protocol for Standard Contract Forms for projects involving
Building Information Modelling (BIM) in Vietnam . . . . . . . . . . . . . . . . 1253
Nguyen Nam Trung and Dinh Tuan Hai

Author Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1259

 Part I
Keynote Lectures
 Four Decades of Computing in Civil Engineering

Hojjat Adeli

The Ohio State University, Columbus, Ohio USA

adeli.1@osu.edu

Abstract. This paper presents the author’s perspective on four decades of com-
puting in civil engineering. Examples of research by the author and his associ-
ates published during the past four decades are briefly described. They include
artificial intelligence and expert system technology, computer-aided design and
engineering (CAD/CAE), computer animation, object-oriented technology, da-
tabase management, solid modelling, parallel processing and supercomputing,
distributed computing on a cluster of workstations, neural networks, evolution-
ary computing and genetic algorithms, case-based reasoning, machine learning,
fractality and chaos theory, wavelet transform, and web-based computing. It is
argued that the introduction of novel computing ideas into the oldest engineer-
ing field has made the field more exciting. It has helped create new technolo-
gies such as semi-active vibration control and health monitoring of large struc-
tures and intelligent freeways, and automate processes that were unthinkable
otherwise.

Keywords: artificial intelligence; machine learning; structural health monitor-

ing

1 Introduction

This paper presents the author’s perspective on four decades of computing in civil
engineering. Selected examples of research by the author and his associates published
during the past four decades are briefly described.
The author’s approach for solution of any new research problem is: take a fresh
look, start from scratch, and break the boundaries. I have been an advocate of trans-
disciplinary research for the past four decades. The departmental boundaries (Civil
Engineering, Mechanical Engineering, Electrical Engineering, Computer Science,
etc.) needed to be broken and silo mentality overcome. In 1993, I founded the interna-
tional journal Integrated Computer-Aided Engineering "based on the premise that
interdisciplinary thinking and synergistic collaboration of disciplines can solve com-
plex problems, open new frontiers, and lead to true innovations and breakthroughs.
The focus of ICAE is the integration of leading edge and emerging computer technol-
ogies for innovative solution of engineering problems", as noted in my Editorial to the
inaugural issue of the journal. My research has been highly multidisciplinary, crossing
computer science, multiple engineering fields, math, and medicine.

© Springer Nature Singapore Pte Ltd. 2020 3

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_1
4 H. Adeli

2 Artificial Intelligence and Other Emerging Computing

Technologies (1980s)

In 1986, I founded the international journal Computer-Aided Civil and Infrastructure

Engineering to advance the application of artificial intelligence (AI) and other emerg-
ing computing technologies in civil engineering, the oldest engineering field. Once
upon a time, there were only two kinds of engineers: civil and military engineer.
In early 1980s, the author introduced AI and Expert System technology into civil
engineering [1-3] with a particular focus as computer-aided design and engineering
(CAD/CAE) [4-5]. Subsequent research included computer animation [6], object-
oriented technology [7-13], database management [14], solid modelling [15], parallel
processing and supercomputing [16-18] (The author published the first book on Paral-
lel Processing in Structural Engineering in 1993), distributed computing on a cluster
of workstations [19-20], neural networks [21-23] (the author wrote the first journal
article on civil engineering applications of neural networks in 1989), evolutionary
computing and genetic algorithms [24-26], case-based reasoning [27-29], machine
learning (ML) [30], fractality and chaos theory [31], wavelet transform [32-33], and
web-based computing [34]. A review of neural networks in civil engineering up to
2000 is presented in [35] and from 2001 to 2016 is presented in [36].

3 Computational Intelligence (1990s)

The three primary fields of computational intelligence (CI) are neural networks (NN),
fuzzy logic (FL), and genetic algorithm (GA). The author has been advocating a
multi-paradigm approach for solution of complicated pattern recognition problems
through adroit integration of NN, FL, and GA. For example, in early 1990’s Adeli and
associates developed hybrid machine learning algorithms for face recognition and
engineering design problems. The book Machine Learning - Neural Networks,
Genetic Algorithms, and Fuzzy Systems (Wiley1995) was the first authored book to
present the three main areas of CI in a single volume and demonstrate their
integration was a more effective way of solving complicated pattern recognition
problem than each one individually [37].

4 Large Scale Optimization (1980s & 1990s)

In 1998, together with his former Ph.D. student, H.S. Park (currently Professor at
Yonsei University, Korea), the author was awarded a rare U.S. patent (number
5,815,394) for an algorithm particularly suitable for large-scale optimization, known
as Neural Dynamics Model of Adeli and Park (NDAP) [38-39]. It was used for fully
automated optimum design of a 144-story superhighrise building structure with over
20,000 members subjected to the constraints of the American Institute for Steel Con-
struction (AISC), the tallest structure ever optimized at the time [40]. Use of a math-
ematical optimization algorithm for design of such a large structure was unprecedent-
Four Decades of Computing in Civil Engineering 5

ed then but has been gaining tractions among designers of highrise building structures
in recent years [41].

5 Construction Engineering and Management (1990s)

Employing neurocomputing and object technologies, Adeli and Karim developed a

new formulation and an object-oriented information model for construction schedul-
ing, cost optimization, and change order management and implemented it into a soft-
ware system called CONSCOM (CONstruction, Scheduling, Cost Optimization, and
Change Order Management) [42-44]. Adeli and Wu developed a regularization neural
network for construction cost estimation [45]. Senouci and Adeli developed a re-
source scheduling using NDAP mentioned earlier [46].

6 Intelligent Transportation Systems (1990s)

In 1990s, Adeli and associates developed novel multiparadigm algorithms for auto-
mated freeway incident detection [32, 47-48], workzone traffic management [49], and
workzone capacity estimation [50] for an Intelligent Transportation System (ITS).
These algorithms are now being implemented to create intelligent freeways.

7 Smart Structure Technology (1990-present)

In the past thirty years, Adeli and associates have developed a number of algorithms
for active, semi-active, and vibration control [51-55] as well as health monitoring of
large structures such as bridge and highrise building structures [56-58]. Earlier work
on vibration control was published in the book Control, Optimization, and Smart
Structures - High-Performance Bridges and Buildings of the Future in 1999 [59].
Subsequent research was published in two more recent books Intelligent Infrastruc-
ture – Neural Networks, Wavelets, and Chaos Theory for Intelligent Transportation
Systems and Smart Structures [60] and Wavelet-Based Vibration Control of Smart
Buildings and Bridges in 2009 [61].
Recently, Amezquita-Sanchez and Adeli presented a synchrosqueezed wavelet
transform-fractality model for locating, detecting, and uantifying damage in smart
highrise building structures [62], the ultimate objective in health monitoring of struc-
tures.

8 Machine learning (1985-present)

The author and his associates had been doing research on machine learning since mid-
1980s [21-23, 30, 37, 58]. In 1994, Adeli and Hung developed an adaptive conjugate
gradient learning algorithm for effective training of multilayer neural networks [63].
In 2010, Ahmadlou and Adeli presented the enhanced probabilistic neural networks
6 H. Adeli

(EPNN) as a new classification algorithm employing the concept of local decision

circles to overcome the shortcomings of the original probabilistic neural network
(PNN) and improve its robustness to noise in the data [64]. Local decision circles
enable EPNN to incorporate local information and non-homogeneity in the training
data thus resulting in a more accurate algorithm. EPNN has been applied to a number
of different classification problems such as diagnosis of the Parkinson’s disease [65]
and shown it is more accurate than other popular classification algorithms such as the
PNN, support vector machine, k-nearest neighbors algorithm, and classification tree.
The secrets for creation of a powerful classification algorithm are a) discovering
feature spaces with large margins between clusters and close proximity of the class-
mates and b) discovering the smallest number of the features that can yield accurate
classification. Recently, Rafiei and Adeli developed a new classification algorithm
dubbed Neural Dynamic Classification (NDC) using the patented neural dynamic
optimization model of Adeli and park to find the optimum number of features needed
for accurate classification [66]. A beauty of the new algorithm is the smoothness of
convergence curves, an outstanding characteristic inherited from the NDAP, a meas-
ure of numerical robustness and high performance of the new algorithm. NDC has
been used to develop NEEWS, a novel earthquake early warning system using seis-
micity indicators intended to forecast major earthquakes within a time frame of weeks
[67].
Recently, Rafei and Adeli presented a novel algorithm for global and local health
monitoring and damage detection in large structures such as highrise building struc-
tures through integration of two signal processing techniques, synchrosqueezed wave-
let transform and fast Fourier transform, an unsupervised machine learning tech-
nique, the Restricted Boltzmann Machine, and NDC algorithm [58].

9 Nature-Inspired Computing

Nature is an excellent source of inspiration. In recent years, a host of nature-inspired

computing algorithms have been proposed mostly as an optimization tool based on
the phenomena observed in the nature or the behavior of various species [68]. The
early example is genetic algorithm that became popular in the 1990’s. More recent
algorithms are Simulated Annealing [69], Particle Swarm Optimization [70], Harmo-
ny Search Algorithm [71], Spiral Dynamics Algorithm [72], Water Drop Algorithms
[73], Gravitational Search Algorithm [74], Central Force Metaheuristic Optimization
[75], Nature Inspired Chemical Reaction Optimization Algorithms [76], and bacteria
foraging algorithm [77].

10 Conclusion and Perspective

It is argued that the introduction of novel computing ideas into the oldest engineering
field has made the field more exciting. It has helped create new technologies such as
semi-active vibration control and automated health monitoring of structures and intel-
ligent freeways, and automate processes that were unthinkable otherwise. As an ex-
Four Decades of Computing in Civil Engineering 7

ample, to solve the concrete mix design problem, engineers traditionally employ the
guidelines of the American Concrete Institute (ACI), and a costly labor intensive and
time-consuming trial-and-error approach. Recently, Rafiei et al. [78], presented a
novel approach for concrete mix design through the integration of the neural dynam-
ics optimization model of Adeli and Park [38-39] and the enhanced probabilistic neu-
ral network classification model of Ahmadlou and Adeli [64] used as a virtual lab.
The authors note “The outcome of this research is an entirely new paradigm and
methodology for concrete mixed design for the 21st century” and “The cost savings
for large-scale concrete projects can be in the millions of dollars.’

References
1. Adeli, H. and Alrijleh, M.M., "A Knowledge-Based Expert System for Design of Roof
Trusses", Computer-Aided Civil and Infrastructure Engineering, Vol. 2, No. 3, 1987,
pp179-195.
2. Adeli, H. and Balasubramanyam, K.V., Expert Systems for Structural Design - A New
Generation, Prentice-Hall, Englewood Cliffs, New Jersey, 1988.
3. Adeli, H., Ed., Expert Systems in Construction and Structural Engineering, Chapman and
Hall, New York, 1988
4. Adeli, H. and Fiedorek, J., "A MICROCAD System For Design of Steel Connections -
Program Structure and Graphic Algorithms", Computers and Structures, Vol. 24, No. 2,
1986, pp. 281-294.
5. Adeli, H., Interactive Microcomputer-Aided Structural Steel Design, Prentice-Hall, Eng-
lewood Cliffs, New Jersey, 1988.
6. Konicki, M.G. and H. Adeli, "A Three-Dimensional Animation System for Seismic Re-
sponse of Multistory Structures", International Journal of Imaging Systems and Technolo-
gy, Vol. 8, No. 3, 1997, pp. 313-321.
7. Adeli, H. and Hung, S.L., "An Object-Oriented Model for Processing Earthquake Engi-
neering Knowledge,” Computer-Aided Civil and Infrastructure Engineering, Vol. 5, No. 2,
1990, pp. 95-109.
8. Yu, G. and Adeli, H., "Object-Oriented Finite Element Analysis Using EER Model", Jour-
nal of Structural Engineering, ASCE, Vol. 119, No. 9, 1993, pp. 2763-2781.
9. Chuang, L.C. and Adeli, H., "Design-Independent CAD Window System Using the Ob-
ject-Oriented Paradigm and HP X Widget Environment", Computers and Structures, Vol.
48, No. 3, 1993, pp. 433-440.
10. Hung, S.L. and Adeli, H., "Object-Oriented Back Propagation and Its Application to Struc-
tural Design", Neurocomputing, Vol. 6, No. 1, 1994, pp. 45-55.
11. Adeli, H. and Yu, G., "An Integrated Computing Environment for Solution of Complex
Engineering Problems Using the Object-Oriented Programming Paradigm and a Black-
board Architecture", Computers and Structures, Vol. 54, No. 2, 1995, pp. 255-265.
12. Adeli, H. and Kao, W.-M., "Object-Oriented Blackboard Models for Integrated Design of
Steel Structures", Computers and Structures, Vol. 61, No. 3, 1996, pp. 545-561.
13. Jiang, X. and Adeli, H.,”Object-Oriented Model for Freeway Work Zone Capacity and
Queue Delay Estimation", Computer-Aided Civil and Infrastructure Engineering, Vol. 19,
No. 2, 2004, pp. 144-156.
8 H. Adeli

14. Adeli, H. and Hung, S.L., "A Production System and Relational Database Model for Pro-
cessing Knowledge of Earthquake-Resistant Design", Engineering Applications of Artifi-
cial Intelligence, Vol. 3, No. 4, 1990, pp. 313-323.
15. Saleh, A. and Adeli, H., "Solid Modeling of Space Stations in a Concurrent Processing
Environment", International Journal of Imaging Systems and Technology, Vol. 3, No. 1,
1991, pp. 58-63.
16. Adeli, H. and Kamal, O., "Parallel Structural Analysis Using Threads,” Computer-Aided
Civil and Infrastructure Engineering, Vol. 4, No. 2, 1989, pp. 133-147.
17. Adeli, H. and Park, H.S., "Fully Automated Design of Superhighrise Building Structure by
a Hybrid AI Model on a Massively Parallel Machine", AI Magazine, Vol. 17, No. 3, 1996,
pp. 87-93.
18. Adeli, H. and Kamal, O., Parallel Processing in Structural Engineering, Elsevier Applied
Science, London, 1993.
19. Adeli, H. and Kumar, S., "Distributed Finite Element Analysis on a Network of Work-
stations - Algorithms", Journal of Structural Engineering, ASCE, Vol. 121, No. 10, 1995,
pp. 1448-1455.
20. Adeli, H. and Kumar, S., Distributed Computer-Aided Engineering for Analysis, Design,
and Visualization, CRC Press, Boca Raton, Florida, 1999.
21. Adeli, H. and Yeh, C., "Perceptron Learning in Engineering Design,” Computer-Aided
Civil and Infrastructure Engineering, Vol. 4, No. 4, 1989, pp. 247-256.
22. Hung, S.L. and Adeli, H., "A Model of Perceptron Learning with a Hidden Layer for En-
gineering Design", Neurocomputing, Vol. 3, No. 1, 1991, pp. 3-14.
23. Adeli, H. and Zhang, J., "An Improved Perceptron Learning Algorithm", Neural, Parallel,
and Scientific Computations, Vol. 1, No. 2, 1993, pp. 141-152.
24. Adeli, H. and Cheng, N.-T., "Integrated Genetic Algorithm for Optimization of Space
Structures", Journal of Aerospace Engineering, ASCE, Vol. 6, No. 4, 1993, pp. 315-328.
25. Adeli, H. and Cheng, N.-T., "Augmented Lagrangian Genetic Algorithm for Structural Op-
timization", Journal of Aerospace Engineering, ASCE, Vol. 7, No. 1, 1994, pp. 104-118.
26. Kociecki, M. and Adeli, H., “Shape Optimization of Free-Form Steel Space-Frame Roof
Structures with Complex Geometries Using Evolutionary Computing,” Engineering Appli-
cations of Artificial Intelligence, Vol. 38, 2015, pp. 168-182.
27. Waheed, A. and Adeli, H., "Case-Based Reasoning in Steel Bridge Engineering",
Knowledge--Based Systems, Vol. 18, No. 1, 2005, pp. 37-46, 2005.
28. Sirca, G.F., Jr. and Adeli, H., "Case-Based Reasoning for Converting Working Stress De-
sign-Based Bridge Ratings to Load Factor Design-Based Ratings", Journal of Bridge En-
gineering, Vol. 10, No. 4, 2005, pp. 450-459.
29. Karim, A. and Adeli, H., “CBR Model for Freeway Work Zone Traffic Management”,
Journal of Transportation Engineering," Journal of Transportation Engineering, ASCE,
Vol. 129, No. 2, 2003, pp. 134-145.
30. Adeli, H. and Yeh, C., "Explanation-Based Machine Learning in Engineering Design",
Engineering Applications of Artificial Intelligence, Vol. 3, No. 2, 1990, pp. 127-137.
31. Jiang, X. and Adeli, H., "Fuzzy Clustering Approach for Accurate Embedding Dimension
Identification in Chaotic Time Series", Integrated Computer-Aided Engineering, Vol. 10,
No. 3, 2003, pp. 287-302.
32. Samant, A. and Adeli, H., "Feature Extraction for Traffic Incident Detection using Wavelet
Transform and Linear Discriminant Analysis", Computer-Aided Civil and Infrastructure
Engineering, Vol. 15, No. 4, 2000, pp. 241-250.
Four Decades of Computing in Civil Engineering 9

33. Zhou, Z. and Adeli, H., “Time-Frequency Signal Analysis of Earthquake Records Using
Mexican Hat Wavelets”, Computer-Aided Civil and Infrastructure Engineering, Vol. 18,
No. 5, 2003, pp. 379-389.
34. Wang, N. and Adeli, H., “Web-based Tutor for Interactive Design of Single-Span and
Continuous Steel Beams” Computer Applications in Engineering Education, 20:2, June
2012, pp. 383-389.
35. Adeli, H., “Neural Networks in Civil Engineering: 1989-2000”, Computer-Aided Civil and
Infrastructure Engineering, Vol. 16, No. 2, 2001, pp. 126-142.
36. Amezquita-Sanchez, J.P., Valtierra-Rodriguez, M., Aldwaik, M., and Adeli, H., “Neuro-
computing in Civil Infrastructure,” Scientia Iranica A, 23:6, 2016, 2417-2428.
37. Adeli, H. and Hung, S.L., Machine Learning - Neural Networks, Genetic Algorithms, and
Fuzzy Systems, John Wiley and Sons, New York, 1995.
38. Adeli, H. and Park, H.S., "A Neural Dynamics Model for Structural Optimization - Theo-
ry", Computers and Structures, Vol. 57, No. 3, 1995, pp. 383-390.
39. Park, H.S. and Adeli, H., "A Neural Dynamics Model for Structural Optimization - Appli-
cation to Plastic Design of Structures", Computers and Structures, Vol. 57, No. 3, 1995,
pp. 391-399.
40. Adeli, H. and Park, H.S., Neurocomputing for Design Automation, CRC Press, Boca Ra-
ton, Florida, 1998.
41. Aldwaik, M. and Adeli, H., “Advances in Optimization of Highrise Building structures,”
Structural and Multidisciplinary Optimization, 50:6, December 2014, 899-919.
42. Adeli, H. and Karim, A., "Scheduling/Cost Optimization and Neural Dynamics Model for
Construction", Journal of Construction Management and Engineering, ASCE, Vol. 123,
No. 4, 1997, pp. 450-458.
43. Karim, A. and Adeli, H., "CONSCOM: An OO Construction Scheduling and Change
Management System", Journal of Construction Engineering and Management, ASCE, Vol.
125, No. 5, 1999, pp. 368-376.
44. Adeli, H. and Karim, A., Construction Scheduling, Cost Optimization, and Management –
A New Model Based on Neurocomputing and Object Technologies, Spon Press, London,
2001.
45. Adeli, H. and Wu, M., "Regularization Neural Network for Construction Cost Estimation",
Journal of Construction Engineering and Management, ASCE, Vol. 124, No. 1, 1998, pp.
18-24.
46. Senouci, A.B. and Adeli, H., "Resource Scheduling using Neural Dynamics Model of
Adeli and Park", Journal of Construction Engineering and Management, ASCE, Vol. 127,
No. 1, 2001, pp. 28-34.
47. Adeli, H. and Samant, A., "An Adaptive Conjugate Gradient Neural Network - Wavelet
Model for Traffic Incident Detection", Computer-Aided Civil and Infrastructure Engineer-
ing, Vol. 15, No. 4, 2000, pp. 251-260.
48. Adeli, H. and Karim, A., “Fuzzy-Wavelet RBFNN Model for Freeway Incident Detec-
tion”, Journal of Transportation Engineering, ASCE, Vol. 126, No.6, 2000, pp. 464-471.
49. Jiang, X. and Adeli, H., “Freeway Work Zone Traffic Delay and Cost Optimization Mod-
el”, Journal of Transportation Engineering, ASCE, Vol. 129, No. 3, 2003, pp. 230-241.
50. Jiang, X. and Adeli, H., "Clustering-Neural Network Models for Freeway Work Zone Ca-
pacity Estimation", International Journal of Neural Systems, Vol. 14, No. 3, 2004, pp. 147-
163.
51. Adeli, H. and Saleh, A., "Optimal Control of Adaptive/Smart Bridge Structures", Journal
of Structural Engineering, ASCE, Vol. 123, No. 2, 1997, pp. 218-226.\
10 H. Adeli

52. Adeli, H. and Kim, H., "Wavelet-Hybrid Feedback-Least Mean Square Algorithm for Ro-
bust Control of Structures", Journal of Structural Engineering, ASCE, Vol. 130, No. 1,
2004, pp. 128-137.
53. Jiang, X. and Adeli, H., “Dynamic Fuzzy Wavelet Neuroemulator for Nonlinear Control of
Irregular Highrise Building Structures, International Journal for Numerical Methods in
Engineering, Vol. 74, No. 7, 2008, pp. 1045-1066.
54. Wang, N. and Adeli, H." Self-Constructing Wavelet Neural Network Algorithm for Non-
linear Control of Smart Structures," Engineering Applications of Artificial Intelligence, 41,
May 2015, 249-258.
55. Gutierrez Soto, M. and Adeli, H., “Many-Objective Control Optimization of Highrise
Building Structures using Replicator Dynamics and Neural Dynamics Model,” Structural
and Multidisciplinary Optimization, 56:6, 2017, 1521-1537.
56. Park, H.S. Lee, H.M., Adeli, H. and Lee, I., “A New Approach for Health Monitoring of
Structures: Terrestrial Laser Scanning,” Computer-Aided Civil and Infrastructure Engi-
neering, Vol. 22, No. 1, 2007, pp. 19-30.
57. Jiang, X. and Adeli, H., "Pseudospectra, MUSIC, and Dynamic Wavelet Neural Network
for Damage Detection of Highrise Buildings", International Journal for Numerical Meth-
ods in Engineering, Vol. 71, No. 5, July 2007, 606-629.
58. Rafiei, M.H., and Adeli, H., “A Novel Unsupervised Deep Learning Model for Global and
Local Health Condition Assessment of Structures,” Engineering Structures, 156:1, Febru-
ary 2018, 598-607.
59. Adeli, H. and Saleh, A., Control, Optimization, and Smart Structures - High-Performance
Bridges and Buildings of the Future, John Wiley and Sons, New York, 1999.
60. Adeli, H. and Jiang, X., Intelligent Infrastructure – Neural Networks, Wavelets, and Chaos
Theory for Intelligent Transportation Systems and Smart Structures, CRC Press, Taylor &
Francis, Boca Raton, Florida, 2009.
61. Adeli, H. and Kim, H., Wavelet-Based Vibration Control of Smart Buildings and Bridges,
CRC Press, Taylor & Francis, Boca Raton, Florida, 2009.
62. Amezquita-Sanchez, J.P. and H. Adeli, H., “Synchrosqueezed Wavelet Transform-
Fractality Model for Locating, Detecting, and Quantifying Damage in Smart Highrise
Building Structures,” Smart Materials and Structures, 24, 2015, 065034 (14pp).
63. Adeli, H. and Hung, S.L., "An Adaptive Conjugate Gradient Learning Algorithm for Ef-
fective Training of Multilayer Neural Networks", Applied Mathematics and Computation,
Vol. 62, No. 1, 1994, pp. 81-102.
64. Ahmadlou, M. and Adeli, H., “Enhanced Probabilistic Neural Network with Local Deci-
sion Circles: A Robust Classifier,” Integrated Computer-Aided Engineering, 17:3, 2010,
pp. 197-210.
65. Hirschauer, T., Adeli, H., and Buford, T., “Computer-Aided Diagnosis of Parkinson’s Dis-
ease using an Enhanced Probabilistic Neural Network,” Journal of Medical Systems,
39:179, November 2015 (12 pages).
66. Rafiei, M.H. and Adeli, H. “A New Neural Dynamic Classification Algorithm,” IEEE
Transactions on Neural Networks and Learning Systems, 28:12, 2017, 3074-3083
(10.1109/TNNLS.2017.2682102).
67. Rafiei, M.H. and Adeli, H., “NEEWS: A Novel Earthquake Early Warning System Using
Neural Dynamic Classification and Neural Dynamic Optimization Model,” Soil Dynamics
and Earthquake Engineering, Vol. 100, 2017, pp. 417-427.
68. Siddique, N. and Adeli, H., Nature Inspired computing – Physics- and Chemistry-based
Algorithms, CRC Press, Taylor & Francis, Boca Raton, Florida, 2017.
Four Decades of Computing in Civil Engineering 11

69. Siddique, N. and Adeli, H., “Simulated Annealing, Its Variants and Engineering Applica-
tions” International Journal on Artificial Intelligence Tools, 25:6, 2016, 1630001 (24 pag-
es)
70. Imran Hossain, S., Akhand, M.A.H., Shuvo, M.I.R., Siddique, N., and Adeli, H., “Optimi-
zation of University Course Scheduling Problem using Particle Swarm Optimization with
Selective Search,” Expert Systems with Applications, 127, 2019, 9-24.
71. Siddique, N.H. and Adeli, H., “Harmony Search Algorithm and Its Variants,” International
Journal of Pattern Recognition and Artificial Intelligence, 29:8, 2015, 1539001 (22 pages).
72. Siddique, N.H. and Adeli, H. “Spiral Dynamics Algorithm,” International Journal on Arti-
ficial Intelligence Tools¸ 23:6, 2014, 1430001 (24 pages).
73. Siddique, N.H. and Adeli, H. “Water Drop Algorithms,” International Journal on Artificial
Intelligence Tools, 23:6, 2014, 1430002 (22 pages).
74. Siddique, N.H. and Adeli, H., “Gravitational Search Algorithm and Its Variants,” Interna-
tional Journal of Pattern Recognition and Artificial Intelligence, 30:8, 2016. 1639001 (22
pages).
75. Siddique, N.H. and Adeli, H., “Central Force Metaheuristic Optimization,” Scientia Irani-
ca, - Transaction A: Civil Engineering, 22:6, 2015, 1941-1953 (Invited Paper).
76. Siddique, N. and Adeli, H., “Nature Inspired Chemical Reaction Optimization Algo-
rithms,” Cognitive Computation, 9:4, 2017, Pages: 411-422 (DOI 10.1007/s12559-017-
9485-1).
77. Wang, J., Zhong, D., Wang, D., Liu, H., and Adeli, H., “Smart Bacteria Foraging Algo-
rithm-Based Customized Kernel Support Vector Regression and EPNN for Compaction
Quality Assessment and Control of Earth-Rock Dam,” Expert Systems, 35:6, 2018 (DOI:
10.1111/exsy.12357).
78. Rafiei, M.H., Khushefati, W.H., Demirboga, R., and Adeli, H., “Novel Approach for Con-
crete Mixture Design Using Neural Dynamics Model and the Virtual Lab Concept,” ACI
Materials Journal, 114:1, January-February 2017, 117-127 (MS No. M-2016-104.R1, doi:
10.14359/51689485).
 Industrialized construction of medium span concrete
bridges using movable false work

Jose TURMO1
1
Universitat Politècnica de Catalunya, 08034 Barcelona, Spain
jose.turmo@upc.edu

Abstract. Modern geometrical requirements for high capacity ground infra-

structure, as highways or high-speed train lines, imply increasing numbers of
long tunnels and viaducts in mountainous countries. Moreover, for the sake of
the environment, only some selected construction methods are eligible for sen-
sitive areas. Long viaducts are usually solved by means of medium span con-
crete viaducts. For the sake of quality and health and safety, industrialized con-
struction methods, as launching girders or precast construction, is usually pre-
ferred. The paper will describe the pros and cons of some construction methods
using movable trusses to hold in place either the formwork or precast segments,
emphasizing the possibilities of reducing the critical path. New construction se-
quences, non-standardized structural details, static and dynamic tests, reinforc-
ing criteria, structural response, finite element model analysis and recent al-
ready built examples will be described to illustrate the possibilities of the meth-
ods.

Keywords: Movable Scaffolding System, Industrialized Construction, Medium

Span Bridges,

1 Construction of in situ bridges with the movable scaffolding

system

Modern geometrical requirements for high-capacity ground infrastructure, as

highways or high-speed train lines, imply increasing number of long tunnels and via-
ducts in mountainous countries. Moreover, for the sake of the environment, only some
selected construction methods are eligible for sensitive areas. Among them, a very
popular one for medium span concrete viaducts is the span-by-span construction with
movable scaffolding system (MSS). The method can be easily adapted to any geome-
try and, from an environmental point of view, only the shadow of the deck is spoilt
during construction periods. Another advantage is the fact that the method is inde-
pendent of the height of the piers, allowing separating the construction of the deck of
the ground conditions. Moreover, form the health and safety point of view, the fact
that we are working with a moving platform eases the incorporation of collective
safety measures to the system. Since its first application in the Kettiger Hangbrücke,
(Germany) in 1959 [Däbritz (2011)], this construction method has been applied ex-

© Springer Nature Singapore Pte Ltd. 2020 13

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_2
14 J. Turmo

tensively all around the world. The method became very popular in Spain and it has
been applied there extensively [Seopan (2016)]. It is claimed that spans up to 90 m
can be built with such a procedure [Pacheco et al. (2011)], even though spans up to 60
m are more common for highway bridges. This construction method can adapt to any
longitudinal geometry
Concreting of span-by-span construction with travelling scaffolding (Fig 1) is typi-
cally made in two phases. In the first phase, webs and bottom flange of the box girder
are concreted (and usually diaphragms). In order to do so, steel internal and external
formwork is used. In the second phase, top flange is concreted. Once the internal steel
formwork is removed, overhang top slab is concreted being supported on the external
steel formwork and top slab between the flanges is concreted either with traditional
scaffolding or with the help of ribbed precast slabs used as lost formwork. Once the
cross section is self-supporting, this is to say, once the concrete of the top flange is
hardened, the segment can be prestressed and the movable scaffolding system can be
shifted to the next span.

Fig. 1. Movable scaffolding system to build span-by-span viaducts.

Fig 2 shows a sketch of the two concreting phases of the cross section of a box
girder bridge for the traditional concreting of the deck. This procedure takes usually
two weeks per span. It is true that it is possible to reduce the construction cycle dura-
tion, but it requires a third shift. Moreover, all activities belong to the critical path,
being then critical all of them. This implies that any hitch in any of them might cause
a delay in the construction plan.

2 New Casting Method

In order to shorten the critical path, it is advisable to look for self-supporting cross
sections after the concreting of the first phase. This allows to prestress and to move
the scaffolding subsequently. In this way, the pouring of the second phase can be
done out of the critical path. The first phase would include the concreting of webs,
bottom flange and the overhung top flanges (Fig 3). The second phase will imply the
concreting of the inner part of the top flange between webs.
Industrialized construction of medium span … 15

The first documented bridge built with this system is the Viaduct over the Genil
River, in Spain [Diaz (2016a)]. In this simply supported structure, the overlapping
between reinforcement belonging to different phases was made by straight prolonga-
tion; impeding therefore the easy movement of the inner formwork. Hence, to pro-
mote the use of this construction method, the problem of the structural connection
between phases has to be solved, since overlapping is not appropriate. The use of
transverse prestressing or the use of mechanical couplers would be technically perfect
options, but their cost would impede their use as a general solution. Welding is also
widely avoided, as ductility of steel reinforcement is compromised. More expensive
quality control and specialized manpower is also required. Therefore, this connection
must be guaranteed by means of conventional reinforcement, where the length of
overlapping and the form of these are critical for the success of the procedure.

Fig. 2. Traditional casting method. Phase 1 in orange. Phase 2 in blue. [3]

Fig. 3. New casting method. Phase 1 in orange. Phase 2 in blue. [3]

The search of joints between phases of compact geometry leaded to loop joints as
the ones shown in Fig 4. These solutions are innovative and are not covered by stand-
ards whatsoever. Moreover, they are not included in the exhaustive document pub-
lished by the FIB [FIB (2000)]. Hence, their effectiveness had to be investigated ex-
perimentally [Diaz et al (2012), Contreras (2012), Diaz et al (2013), Villalba et al
(2013), Villalba and Casas (2016)].

3 Structural challenges for the movable scaffolding system

The position of the joint in the new casting sequence makes this to be subjected
to different normal and shear efforts. On the one hand, transverse flexure and vertical
shear will load the joint. Moreover, horizontal shear, due to differential bending,
torque, and differential shrinkage and creep; has also to be transferred through the
16 J. Turmo

joint. As the movable scaffolding system construction method aims to be applicable

not only to highway bridges but also to high-speed railway bridges, the fatigue
strength of the loop joints has to be carefully studied.

Fig. 4. Loop joints protruding from the first phase (Courtesy of Mr. R. Soto, from
Mecanotubo)

Villalba and Casas (2016) tested a hook stiffened by transverse reinforcement

bars. The studies concentrated on the strength, stiffness and serviceability of such
joint under combined shear and bending (simple flexure). The research work com-
prised the static bending behaviour of reinforced concrete slabs with loop joints of
different geometries. The results were compared with control slabs. In the static tests,
crack widths and cracking pattern were observed at service load levels, and the ulti-
mate behaviour was evaluated by means of tests to failure. Service performance of
loop joints was confirmed from test results to be similar to slabs without joints. Static
loading tests confirmed the good performance and effectiveness of this loop joint type
under static loads. Loop joint detail design criteria were also proposed. Villalba et al
(2013) studied experimentally the fatigue behavior and strength of loop joints with
regard to loop bar diameter, loop joint width and applied load ranges for simple flex-
ure. Results of slabs subjected to 2.106 cycles were compared with the ones from
control slabs without joints. The experimental loop joint design was found to perform
correctly under fatigue loads.
Whereas Villalba and Casas (2016) and Villalba et al (2013) focused on simple
flexure, Contreras (2012) focused on the experimental behavior of joints under pure
bending (no shear in the joints). Slabs were also subjected to static and cyclic loading.
Both test types focused on studying the longitudinal and transverse reinforcement
strain, deflection, joint opening and ultimate strength. For the dynamic loading tests,
different load levels and load variation was used. Strain, deflection and joint opening
were recorded after 2.106 cycles. After dynamic tests, the remaining static strength of
the specimens was also tested.
Diaz et al (2012) studied experimentally a fixed loop joint geometry under static
pure flexure for different concrete types. Normal strength concrete, high strength
concrete and self-compacting concrete slabs were evaluated. Loop joint specimens
and control ones performed similarly regarding the joint opening. It was deduced that
Industrialized construction of medium span … 17

joint behaviour was mostly influenced by the construction joint. A modified formula
for calculation of joint opening was proposed and the joint opening did not imply any
concern under service loads. Ultimate strengths of loop joint and control specimens
were very similar. Diaz et al (2013) studied experimentally a fixed loop joint geome-
try under static shear for different concrete types. Tests were performed on normal
strength concrete, high strength concrete and self-compacting shear-off panels to
check the ability of the joint to withstand horizontal shear. Half of the panels were
previously cracked under bending, in order to simulate the cracking induced by trans-
verse flexure. It was checked that transverse bending cracks hardly affected the shear
strength of the joints. Ultimate shear strength of the joint was much higher than poten-
tial ultimate loads in actual bridges.
On the other hand, another great structural problem to solve is the control of
transverse bending moment and distortion of the cross section while the second phase
is still open. Practical solutions that have employed by different designers are partial
concreting of the top slab between webs, leaving some open windows to remove the
inner formwork, or the disposition of several ties between both webs to avoid the
distortion of the transverse cross section. None of these solutions is totally satisfacto-
ry, since the total elimination of any element that impedes the free movement of the
inner formwork should be aimed. Moreover, analyses made for designing these bridg-
es do not usually involve three-dimensional models. Control of the transverse flexure
is made by abacuses and frame models simulating the structural behaviour of the
cross-section. With these models it is difficult to extract conclusions about the sizing
of auxiliary elements to control the transverse flexure. The structural validity of the
method was assessed with a three-dimensional analysis [Diaz et al (2016)]. The effec-
tiveness of different structural systems to control the transverse flexure was also eval-
uated, measuring the distortion of the open section and the longitudinal and transverse
stresses

4 Prefabricated segmental concrete bridges with external

prestressing and dry joints

Prefabricated segmental concrete bridges with external prestressing and dry

joints are associated with a span-by-span construction process that is thought to be the
fastest among this type of construction process. For the construction of each of the
spans, the segments are placed one next to the other, suspended from a beam or ar-
ranged in a mobile falsework, and are assembled by means of external prestressing. In
Europe, where usually a waterproofing layer to prevent leakage is applied on the top
of the deck, it is not necessary to apply any epoxy resin between the joint-faces of the
segments. It was also a common practice in some states of the USA some time ago.
Nevertheless, epoxy-free joints are presently forbidden there due to the durability
problems that aroused in some bridges built in the past. It is precisely the subject mat-
ter of the present work to study externally prestressed segmental bridges with resin-
free dry joints. Its most significant characteristic is the nonexistence of bond rein-
18 J. Turmo

forcement crossing the joints, neither active nor passive. The first example of applica-
tion constructed was the Long Key Bridge.
For the Serviceability Limit State (SLS), these bridges are designed considering
that the limit state of decompression must not be reached; hence a minimum compres-
sion ın = 0.5 MPa is maintained in all sections and the joints remain closed. When the
overload increases up to the Ultimate Limit State (ULS), joints open up significantly,
and the structure rapidly loses stiffness and reaches a considerable deflection. The fact
that there is no passive reinforcement crossing the joints means that bending moments
must be carried by more active reinforcement or by higher initial stresses in the steel
tendons. In simply-supported bridges, it is not the SLS of decompression which
guides the design but the ULS of normal stresses.
The shear transmission through open joints is a more complex subject. The uni-
versally accepted theory proposes that the shear forces flow across the joint, through
web and flanges, by two qualitatively and quantitatively different mechanisms. The
first mechanism takes into account the support effect of the interlocking shear keys
usually provided at the joint. Just the shear keys remaining in contact are able to
transfer shear across an open joint. The second represents the friction force that arises
when two flat and compressed surfaces tend to slip one against the other, which is
proportional to the actual compression.
One issue that causes controversy among researchers is the reinforcement of the
segment in the proximity of the open joint. Some authors support the necessity of
providing extra shear reinforcement to hang shear in the zone next to the open joint.

5 Theoretical analysis

It is known that in any beam, the shear load, V, acting at a section x, is mathe-
matically related with the exterior bending moment, M, by (1).

wM(x)
V(x) (1)
wx

After flexural cracking, the exterior bending moment, M, in each section x of the
beam is compensated by a pair of axial forces separated by a lever arm, z; including a
compressive and a tensile axial force, Nc and Ns, respectively (2).

M(x) N s·z N c·z N·z (2)

Then, Equation 1 can be transformed in the following manner:

wM(x) wN·z  wN  wz 
V(x) z·  N. (3)
wx wx wx wx
In conventional concrete beams, the lever arm, z, remains approximately constant
between two contiguous sections, at least during the initial loading stages. Then,
Industrialized construction of medium span … 19

w z  w N 
0 and V(x) z· (4)
wx wx
Equation (4) represents what is traditionally known as the beam effect, and leads to
a distribution of tangential stresses along the length and width of the transversal sec-
tion after cracking. Hence, it is necessary to place stirrups to carry the tensile stresses
when the web of the beam cracks (strut-and-tie analogy). If due to any circumstance
the bond of the longitudinal reinforcement with the surrounding concrete is lost, the
reinforcement is unable to vary its stress from one section to another of the beam,
which implies:

w N  w z 
0 and V(x) N. (5)
wx wx
Equation (5) is known as the arch effect, and means that the shear force is resisted
by an inclination of the compressive axial force. Generally, these two mechanisms are
superimposed before the beam fails by shear.
In a segmental structure with external prestressing, where there is no passive rein-
forcement connecting the segments, and where the prestressing tendons contact the
concrete only at anchor blocks and deviators, the axial force remains essentially con-
stant in every cross section of the beam. Thus, the transmission of the shear force
relies on the arch effect. In this manner, an inseparable association of longitudinal
compressive stresses and tangential stresses takes place.

6 Experimental program

Shear tests involved a total of four segmental beams with an I-shape cross section
of 0.60-m height and 7.60-m length. Two of the beams were cast of normal conven-
tional concrete (PC) and the other two of steel fiber reinforced concrete (SFRC). The
beams were provided with interlocking dry joints, with three shear keys of 90-mm
height in each joint (Fig. 5).

Fig. 5. Test configuration (dimensions in mm) [10]

The tests involved four simply-supported beams of 7.20-m span consisting of three
segments of varying length assembled together with external prestressing, to which an
external load, Q1, was applied up to failure. The following nomenclature was used in
20 J. Turmo

these test series: 1- V1-PC-35: PC beam of mean cylinder compressive strength (fcm)
of 30 MPa with an axial prestressing force of 350 kN. 2- V1-PC-70: PC beam of fcm =
30 MPa with an axial prestressing force of 700 kN. 3- V1-SFRC-35: SFRC beam of
fcm = 30 MPa with an axial prestressing force of 350 kN. 4- V1-SFRC-70. SFRC beam
of fcm = 30 MPa with an axial prestressing force of 700 kN.
The objective of the tests was: (i) to verify the behavior of an open joint when sub-
jected to shear loading; (ii) to quantify the influence of the axial prestressing force on
the capacity of the joint and the structure; (iii) and to study the efficiency of the shear
reinforcement near the joint to determine if it is necessary to include the hanger rein-
forcement that is proposed by some authors. With V1-SFRC tests, which complement
V1-PC tests, the objective was to study the possibility of replacing the conventional
shear reinforcement by steel fibers when using SFRC. With this aim, the conventional
reinforcement placed in V1-PC EHDPV ĭ VWLUUXSV HYHU\ 30 cm) was completely
removed in these beams (except for the reinforcement at anchorage and deviator
zones, and a few longitudinal bars placed to avoid the premature flexural failure of the
longest segment).
From the pre-failure phase, it is worth noting that the apparent position of the neu-
tral axis, as well as the number of active shear keys and cracking development, are
intimately linked to the axial prestressing force. Note that only the PC beams were
taken up to failure, since the conventional reinforcement was lacking in beams V1-
SFRC. There was a fear that a brittle failure could occur, which could cause harm to
both people and equipment. Thus, tests were interrupted when an imminent failure
was intuited.
In beam V1-PC-35, cracking only affected the upper keys (Fig. 6.b). Subsequently
to the diagonal cracking of segment D3, that initiates at the base of the keys and ori-
entates at 40º towards the loading point, a vertical crack propagates following the
position of the stirrup. Finally, a horizontal crack appears in the compressed zone, in
the joint area, due to excessive normal stresses, and develops towards the loading
point losing horizontality due to the combination of normal and tangential stresses.
The beam fails through the flange under a shear force of Vu,exp = 0.081 MN. In beam
V1-PC-70, as a consequence of the deeper position of the neutral axis, cracking af-
fects the three keys (Fig. 6.d). Again, the reinforcement of the segment induces a
bifurcation of the crack that develops following the layout of the shear reinforcement,
perpendicular to the axis of the segment. This crack, with an inclination of 70º to the
horizontal, finally concentrates the entire crack opening. Failure occurs when the
shear crack reaches the compression zone, corresponding to a shear force of Vu,exp =
0.136 MN. In beam V1-SFRC-35, only one crack appeared, at the base of the central
key. It developed with a 45º-inclination up to the bottom of the top flange, continuing
its development along a horizontal plane at the web-flange interface, achieving a sig-
nificant opening, as it can be seen in Fig. 6.a. The test was interrupted before the fail-
ure of the beam, with the joint subjected to a shear force Vu,exp = 0.074 MN. In beam
V1-SFRC-70, a large crack originated at the base of the lower key, which developed
towards the loading point with an angle of 30º to the horizontal (Fig. 6.c). The open-
ing of the joint fully extended throughout the web of the segment. The deformations
Industrialized construction of medium span … 21

of the beams were significant, visually appreciable. The test was terminated before
the failure of the beam, with the joint subjected to a shear force Vu,exp = 0.126 MN.

a b

c d

Fig. 6. Crack patterns obtained in beam tests

V1-SFRC-35 (a); V1-PC-35 (b); V1-SFRC-70 (c); V1-PC-70 (d). [10]
One of the apparently more surprising results obtained in the tests is the fact that
beams with very different reinforcement arrangements present very similar load-
carrying capacity. V1-PC beams, reinforced with conventional shear stirrups, showed
an ultimate shear capacity very similar to V1-SFRC, with no shear stirrups. The shear
response mechanism that assures the flow of tangential stresses in beams without
shear reinforcement is the arch effect. Only the arch effect allows justification that
beams with and without shear stirrups to have so similar a shear load-carrying capaci-
ty. The shear reinforcement does not seem to be effective in these types of structures.

7 Conclusions

The oral presentation will focus on the already built examples, on the evolution
of the method and its construction details, on the different experimental campaigns
done and their results and on the finite element model analysis of the construction
22 J. Turmo

sequence of the MSS. It will also show a comparison with the segmental construction,
highlighting the totally different structural behavior.

8 Acknowledgement

The author is indebted to the Spanish Ministry of Economy and Competitive-ness for
the funding provided through the research project BIA2013-47290-R and BIA 2017-
86811-C2 founded with FEDER funds and directed by Jose Turmo. Author is also
LQGHEWHGWRWKH6HFUHWDULDG¶8QLYHUVLWDWVL5HFHUFDGHOD*HQHUDOLWDWGH&DWDOXQ\DIRU
the funding provided through Agaur (2017 SGR 1481)

References
1. &RQWUHUDV&3³9HULILFDFLyQH[SHULPHQWDOGHXQLRQHVPHGLDQWHOD]RVGHDUPDGXUD
\VXDSOLFDFLyQHQODFRQVWUXFFLyQHYROXWLYDGHSXHQWHV´3K''LVVHUWDWLRQ8QLYHUVLWDW3o-
litècnica de Catalunya, BarcelonaTech http://www.tdx.cat/handle/10803/81691
2. 'lEULW] 0  ³0RYDEOH 6FDIIROGLQJ 6\VWHPV´ 6WUXFWXUDO (QJLQHHULQJ ,QWHUQDWLRQDO
4/2011 DOI: 10.2749/101686611X13131377725442
3. 'LD]-57XUPR--RUTXHUD--%DUUDJDQ%5DPRV*$SDULFLR$&³2p-
timization of in situ construction of concrete decks: flexure tests of compact splices of re-
LQIRUFHPHQW EHWZHHQ SKDVHV´ &RQVWUXFWLRQ  %XLOGLQJ 0DWHULDOV (OVHYLHU 9RO  SS
191-203, DOI: 10.1016/j.conbuildmat.2012.11.085
4. Diaz, J.R.; Turmo, J.; Jorquera, J.J.; Barragan, B.; Ramos, G.; Aparicio, A.C. (2013)
³6KHDU-off strength of compact reinforcement splicing for improved construction of in situ
FRQFUHWH VWUXFWXUHV´  &RQVWUXFWLRQ  %XLOGLQJ 0DWHULDOV 9RO  SS -207, DOI:
10.1016/j.conbuildmat.2013.04.052
5. 'LD]-5+DDFK9*7XUPR--RUTXHUD--D³,PSURYHG&RQVWUXFWLRQRI&Rn-
FUHWH9LDGXFWVZLWK0RYDEOH6FDIIROGLQJ6\VWHPLQ6SDLQ´-RXUQDORI%ULGJH(QJLQHHULQJ
ASCE, Vol 21, Iss 9, article number 04016050, DOI: 10.1061/(ASCE)BE.1943-
5592.0000831
6. 'LD]-5+DDFK9*7XUPR--RUTXHUD--E³1XPHULFDOVWXG\RIDQHZFRn-
VWUXFWLYHVHTXHQFHIRUPRYDEOHVFDIIROGLQJV\VWHP066DSSOLFDWLRQ´$GYDQFHVLQ&Rn-
crete Construction. 01/09/2016. 4.3. Pags: 173
7. FIB (2000) Bond of reinforcement in concrete Bulletin 10 Fédération International du Bé-
ton
8. 3DFKHFR 3 HW DO  ³7HFKQLFDO &KDOOHQJHV RI /DUJH 0RYDEOH 6FDIIROGLQJ 6\VWHPV´
Structural Engineering International 4/2011 DOI: 10.2749/101686611X13131377725640
9. Seopan ³0DQXDOGHFLPEUDVDXWRODQ]DEOHV´6HRSDQ
10. 7XUPR - 5DPRV * $SDULFLR $&  ³6KHDU %HKDYLRXU 2I 8QERQGHG 3RVW-
7HQVLRQHG6HJPHQWDO%HDPV:LWK'U\-RLQWV´ACI-Structural Journal, American Concrete
Institute, Vol 103, Iss 3, pp. 409-417,
11. VillDOED6&DVDV-5$SDULFLR$&9LOODOED9³1HZVWUXFWXUDOMRLQWE\UHEDU
ORRSLQJDSSOLHGWRVHJPHQWDOEULGJHFRQVWUXFWLRQIDWLJXHVWUHQJWKWHVWV´$6&(-RXUQDORI
Bridge Engineering, 2013, 18, No. 11, pp. 1174±1188
12. Villalba, S., Casas, J.R. ³1HZ VWUXFWXUDO MRLQWE\ UHEDU ORRSLQJ DSSOLHG WR VWDJHG
ER[JLUGHUEULGJHFRQVWUXFWLRQ6WDWLFWHVWV´6WUXFWXUDO&RQFUHWH1R
 Using Simulation to Estimate and Forecast
Transportation Metrics: Lessons Learned
L. R. Rilett, PhD, P.E.

University of Nebraska ± Lincoln, Lincoln, NE 68583, USA

lrilett2@unl.edu

Abstract. In recent years transportation planners and engineers have begun to

utilize traffic simulation models to estimate and forecast new transportation op-
erations and reliability metrics. For example, the Highway Capacity Manual,
Sixth Edition: A Guide for Multimodal Mobility Analysis (HCM-6) has recent-
ly adopted 1) passenger car estimation methods that are based on the mi-
crosimulation model VISSIM, and 2) urban arterial reliability estimation meth-
ods that are based on a Monte Carlos simulation technique. The advantage to
simulation methods is that the metrics, which may be based on central tendency
(e.g. mean, median), dispersion (variance, percentile), or even a combination of
other metrics (e.g. reliability index), may be easily calculated and/or estimated.
For this reason, the number of metrics developed and used has continued to in-
crease. As one example, many researchers over the past decade have focused
on developing and estimating metrics related to network reliability and resili-
ence. However, it is an open research question on when and where these simu-
lation approaches are appropriate to use. This paper will discuss a number of is-
sues related to using simulation for estimating transportation metrics with a fo-
cus on model assumptions and model calibration. Specific examples from real-
world test beds will be provided. Lastly, the paper will provide an overview of
lessons learned and areas of future research.

Keywords: Calibration, Validation, Simulation, Transportation Metrics

1 Introduction

In recent years transportation planners and engineers have begun to utilize traffic
simulation models to estimate and forecast various transportation operations and reli-
ability metrics [1-6]. For example, the Highway Capacity Manual, Sixth Edition: A
Guide for Multimodal Mobility Analysis (HCM-6) has recently adopted 1) passenger
car estimation (PCE) methods that are based on the microsimulation model VISSIM,
and 2) urban arterial reliability estimation methods that are based on a Monte Carlos
simulation technique [6]. The use of simulation models is particularly noteworthy
because for most of its existence the Highway Capacity Model relied on empirical
data for developing its planning and design values.
The advantage to microsimulation models is that the metrics, which may be based
on central tendency (e.g. mean, median), dispersion (variance, percentile), or even a
combination of metrics (e.g. reliability index) may easily be calculated/estimated

© Springer Nature Singapore Pte Ltd. 2020 23

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_3
24 L. R. Rilett

from the simulation model output. For this reason, the number of metrics used by
transportation operators and users has continued to increase. To cite but one example,
many researchers have focused on developing and estimating metrics related to relia-
bility and resilience. However, it is an open research question on when and where
these simulation approaches are appropriate to use.
This paper will discuss a number of issues related to using simulation techniques
for estimating transportation metrics with a special focus on model assumptions and
model calibration. The paper adopts the use of three case studies. The first is related
to general calibration issues, the second examines the use of simulation models to
estimate transportation metrics, and the third is related to using simulation approaches
for estimating system reliability. All examples are from real-world test beds. Lastly,
an overview of lessons learned and areas of future research will be provided.

2 Case Study: Microsimulation Calibration

Traffic microsimulation models simulate the behavior of individual drivers/vehicles

and the roadway environment [7]. As shown in Figure 1, there are two main inputs.
The first is the supply which consists of infrastructure (e.g. roadways, lanes, lane
width, link length, etc.) and operating strategies (e.g. traffic signal timing, signs, etc.).
Supply is typically under the control of the decision maker. Demand represents the
LQGLYLGXDO GULYHU¶V XVH RI WKH V\VWHP DQG LV W\SLFDOO\ UHSUHVHQWHG E\ DQ RULJLQ-
destination matrix. Demand is typically not under the control of the decision maker.
The model simulates the interaction between the demand and supply at set intervals
(e.g. 0.1 seconds) over the entire simulation time (e.g. one hour). Note that the mod-
els can have a large number of input parameters (e.g. car following distance, truck
acceleration, etc.). The user may choose to use the default values or input their own.
The output consists of intersection, link, corridor and/or system performance infor-
mation such as travel time, level of service, etc. Because of the complex, non-linear
relationships between demand and supply, microsimulation models have become
increasingly popular because they can readily model most transportation systems.
While relatively easy to implement and use, it is unclear whether the results from
the simulation model are realistic or correct. A typical user will simulate a known
condition (e.g. the current system) and compare the simulation output to existing em-
SLULFDO GDWD  ,I WKH VLPXODWLRQ PRGHO ³PDWFKHV¶ WKH HPSLULFDO GDWD WKHQ WKH model
might be considered appropriate. There are any number of output measures that can
be compared (e.g. travel time, delay, link volume) and many non-statistical (e.g. mean
absolute error) and statistical methods (e.g. t-tests, f-tests, non-parameter tests) that
can be used to measure how ³JRRG´WKe empirical data is replicated [7, 8]. For exam-
ple, a user might decide that if there is no statistical difference between the observed
mean link travel times and the simulated mean link travel times then the model is
deemed acceptable. Alternatively, if the means absolute error between all links in the
network are within ten percent the user as the decision criteria of whether a given
model is acceptable.
Using Simulation to Estimate and Forecast … 25

Fig. 1. Microsimulation Model Calibration Procedure (Source: [7])

If the empirical data and the simulated data do not ³match´ WKHQ the user may
choose to calibrate the model. Calibration is the process of modifying the model pa-
UDPHWHU YDOXHV VR WKDW WKH PRGHO ³DFFXUDWHO\´ replicates the observed local traffic
conditions. As shown in Figure 1, if the comparison is not favorable the simulation
parameter values are changed and the process repeats. The optimization step can use
any number of algorithms including genetic algorithm, simplex methods, etc. At some
point the user will decide the model is acceptable or decide that the model cannot be
calibrated. Note that the calibration process can be automated and when this is done
the process may involve tens of thousands of simulation runs before the model is
considered calibrated. In addition, every parameter set may be simulated a set num-
ber of times (e.g. ten) so that statistical inference techniques (e.g. t-test) may be uti-
lized for the comparison.
Figure 2 shows the results of a calibration of an urban arterial in Houston, Texas
[8]. There were two tests used. The first was the mean absolute error ratio (MAER)
that was used to identify how close the average simulated travel times on the links
compared to the actual travel times. The lower the MAER value the closer the link
travel times were to each other. The second was a Kolmogorov±Smirnov (KS) test
that compared the distribution of the simulated link travel times to the distribution of
the empirical times. If the p value of the test was greater than 0.05 then there was no
statistical difference between the two distributions. Figure 2 shows the MAER versus
the p results for a subset of the parameter sets tested. There are a number of im-
portant points related to this figure. First, the MAER value for the default parameters
26 L. R. Rilett

set was 0.21, which indicates that using the default values would give an average
error of 21 percent. Secondly, there are a number of parameter sets that have a low
MAER values but where the simulated travel time distribution did not match, statisti-
cally, the empirical travel time distribution. Lastly, there were a number of parameter
sets where the MAER was low and which also had statistically similar travel time
distributions (e.g. black squares).

Fig. 2. Mean Absolute Error Ratio versus P Value (Source: [8])

There are two important points to the above case study.

1) It is critical to calibrate simulation models. Simply using the default values,
while very easy to do, rarely results in satisfactory performance. It is not uncommon
to achieve MAER values of over 100 percent using default values.
2) When calibrating a model it is important to consider the ultimate application. If
the user is interested in link travel time distributions then a non-parameter statistical
test should be used. While a low MAER value may result in the appropriate travel
time distribution there is also a good chance it may not. This is not surprising as the
MAER only considers measures of central tendency and not measures of dispersion.

3 Case Study: Passenger Car Equivalent Estimation

The passenger car equivalent (PCE) of a truck relates the impact of a truck on a traffic
stream to an equivalent number of passenger vehicles that would have the same im-
pact. The HCM-6 employs an equivalency capacity methodology to calculate the
PCE. In this case the PCE is the value would make the capacity of mixed flow traffic
equal to the capacity of auto-only flow traffic [6, 9, 10]. A series of PCE estimation
models were developed for mixed flow traffic that had different truck percentages,
different truck composition, grades, and grade lengths [6, 11-14]. These models were
Using Simulation to Estimate and Forecast … 27

used to calculate the PCE values in the HCM-6. The published tables in the HCM-6
PCE are disaggregated based on the percentage of single unit trucks and heavy trucks,
the total truck percentage, the grade, and the grade length [6]. The underlying data
used to develop these models were obtained from over 3,680 VISSIM microsimula-
tion model runs
The HCM-6 PCE model developers changed four VISSIM parameters related to
highway geometrics (e.g. lane width, grade, etc.), nine parameters related to vehicle
characteristics (weight and power distribution, acceleration rates, etc.), and four relat-
ed to traffic flow (e.g. truck percentage, desired speed, etc.). All other VISSIM pa-
rameters were set to the default values. While not explicitly stated in the HCM-6 the
models were developed for US east coast conditions. There are a number of key dif-
ferences between these conditions and the underlying characteristics of western US
rural highways including:
1. High truck percentages. US rural values are typically greater than 25%;
2. Differences in percentage of Single Unit Trucks and Tractor Trailers. Rural ar-
eas typically experience a much higher percentage of Heavy Trucks compared to
SUT;
3. Differences in operating characteristics between passenger cars and trucks.
Many heavy trucks in the western U.S. are equipped with speed limiters so that their
speeds are considerably less than the speed limit. In addition, the speed limits in the
western US tend to be higher than urban areas;
4. Most US rural interstate roadways have only two lanes per direction;
5. Different heavy vehicle characteristics (e.g. length distributions).
Based on the above differences it was hypothesized that the HCM-6 PCE values
might not be appropriate for western US rural interstate highways. The HCM-6 pro-
cedure was repeated using a calibrated VISSIM model based on empirical data from
I-80 in western Nebraska. The PCEs were estimated using the HCM-6 equal-capacity
method and VISSIM 9.0 simulation data under 1) the HCM-6 conditions and 2) the
Nebraska empirical conditions. It was found that the PCEs recommended in HCM-6
underestimated the effects of trucks on four-lane level freeway segments which have
1) high truck percentages, 2) large differences in free-flow speed distributions, and 3)
different truck lengths. Figure 3 shows this relationship between estimate PCE and
truck percentage. It can be seen that, the PCE values calculated for Western US con-
ditions are, on average, 54% higher. This difference could have profound impacts on
the estimation of many metrics including freeway Level of Service, delay, travel time,
etc. For example, the HCM-6 is often used to make funding decisions. It is easy to
hypothesize an example where an analysis, using HCM-6 assumptions, would indi-
cate that a given freeway is operating much better than it actually is.
There are two important points to this case study.
1. If microsimulation models are to be used to develop design and/or planning
models, such as the HCM-6, all the information related to the microsimulation model
development needs to be stated up front. This includes what model version was used,
what microsimulation parameter values were used, what seed values were used, how
the microsimulation model was calibrated and validated, and what type of output was
chosen for analysis.
28 L. R. Rilett

2. The assumptions underlying the model need to be identified explicitly. One of

the major reasons for the discrepancy between the HCM-6 results and the revised
model was the difference in free flow speeds between passenger cars and heavy
trucks. Without knowing these assumptions it would be impossible for a user to as-
certain whether the model estimates (e.g. HCM-6 PCE values) are appropriate for a
given application.
As the example showed, if the underlying model development and the underlying
assumptions are not appropriate for local conditions the model could provide errone-
ous planning information.

Fig. 3. Passenger Car Equivalent versus Truck Percentage (Source: [11])

4 Case Study: Urban Travel Time Model

The 6th edition of the Highway Capacity Manual (HCM-6) developed a methodology
for predicting travel time reliability (TTR) which is a measure of the distribution of
WULSWUDYHOWLPHVRYHUDQH[WHQGHGSHULRGRIWLPH775LVDQLQKHUHQWSDUWRIWUDYHOHUV¶
route choice decisions and is used by traffic managers to better quantify operations
rather than on simply using average travel times. The HCM-6 methodology uses as
input historical data (e.g., weather and volume fluctuations) and empirical traffic data
(e.g., 1-day volume count). The output is average travel time and reliability predic-
tions over an extended period (e.g., a year).
The HCM-6 methodology for estimating travel time distributions and the reliability
performance metrics is illustrated in Figure 4. The HCM-6 methodology assumed
that the day-to-day variability in travel time can be estimated by adjusting demand
and predicting weather and incident occurrences over an extended period of time. The
method is based on a Monte Carlo simulation approach which models the effect of
changes in demand, incidents, and weather over time. The result is an estimate of the
mean-travel time and a measure of reliability as determined by day-to-day variability
Using Simulation to Estimate and Forecast … 29

over an extended period. A complete description of the process may be found else-
where [6, 16]. In brief the modeler needs the
(a) functional class and geometric features of the subject urban street facility,
(b) weather data,
(c) basic traffic volumes and count date and time under clear weather conditions,
(d) segment and intersection annual crash frequency, and
(e) duration of the analysis period (typically 15-mins.), study period (e.g., P.M.
peak 4.5 h to 5.5 h), and reliability reporting period (e.g., all weekdays in a year)
The HCM-6 STREETVAL engine, is used to adjust the base dataset demand and
capacity to reflect their variations under the weather, incident, and time of day infor-
mation. This information is then used in the HCM-6 urban street facility evaluation
methodology (i.e., Chapter 16 of HCM-6) to estimate the 15 minute average travel
time on the testbed for each of the analysis period. These estimates are then used to
calculate the travel time distribution of the testbed and, based on this travel time, the
associated reliability metrics. For example, once the travel time distribution is

e
Fig. 4. HCM-6 Method for Estimating Travel Time Distribution (Source: [6] see Exhibit 17-2).

obtained it is very easy to calculate a number of reliability metrics including the:

a) Mean Travel Time Index (TTI) (e.g. ratio of mean to free flow travel time,
b) Planning Time Index (PTI) (e.g. 95th percentile time divided by the free flow
time), and
c) Buffer Index (e.g. 95th percentile travel time which is the quotient of i) the dif-
ference between the 95th percentile travel time, and ii) the mean travel time and the
mean travel time).
Interestingly, there does not appear to be any existing literature on validating the
HCM-6 methodology with empirical data other than the research described in this
30 L. R. Rilett

section [16]. Archived Bluetooth (BT) data from a 3.6-mile urban arterial in Lincoln,
Nebraska that was collected from January 2015 to May 2017 were analyzed [15, 16].
Figure 5 shows the distribution of travel times along a half mile section of the corridor
that were estimated using the HCM-6 approach and the corresponding empirical BT
data. It was found that there were statistically significant differences, but minimal
practical differences, between the mean-travel times for the HCM-6 predictions and
the Bluetooth data. However, the HCM-6 predictions had considerably less variation.
Not surprisingly, the methodology underestimated the buffer index and the PTI by
approximately 62% and 9%, respectively.

Fig. 5. Blue Tooth (BT) and HCM-6 Travel Time Distributions (Source: [16])

The research on estimating reliability on this corridor is on-going. The next step in
the research is the calibration of the HCM-6 Monte Carlo simulation model to local
conditions using a similar calibration process to that shown in Figure 1. In addition, a
longer corridor is being analyzed. There are two important points to this case study.
1. Reliability metrics are becoming increasingly popular in transportation planning
and operations. This was the impetus behind the reliability methodology being in-
cluded in the HCM-6. Implicit in these approaches is the need to accurately predict
not only mean travel times but also the entire travel time distribution. Needless to say
this is a much more difficult task. Fortunately, with the recent advances in Intelligent
Transportation Systems and computer technology there is now substantially more
empirical data available to develop these sophisticated prediction models.
2. It is critical that the simulation models not only be calibrated to local conditions
but that they be validated for the conditions on which they were developed. While
this may have been done for the HCM-6 methodology there are no examples in the
literature describing how or when this was done. While the approach is very sophisti-
cated and a very important step forward in this area it is up to the user to decide
whether it is appropriate for their application.
Using Simulation to Estimate and Forecast … 31

5 Concluding remarks

In recent years transportation planners and engineers have begun to utilize traffic
simulation models and simulate to estimate and forecast various transportation opera-
tions and reliability metrics. This is in response to the general complexity of trans-
portation networks in that simple, deterministic models do not provide the type of
information required by decision makers. As the case studies showed there are a
number of issues related to using simulation models that users should be aware of.
These include:
1) Assumptions should follow local conditions. As shown, in case study 2 if the
assumptions underlying the model are violated there is a good chance the model re-
sults will be inappropriate. Following on this it is also critical that the model devel-
opers explicitly state what assumptions are used. For example, in the HCM-6 there is
no mention of the underlying assumptions used to develop the PCE values.
2) It is important to calibrate and validate the model. As case study 3 showed im-
plementing a model that has not been validated may lead to issues in accuracy. There
is also the danger that users may decide that the new metrics that are being developed,
such as reliability, are not worth pursuing because of inaccurate results.
3) Related to point 2 is that model developers and model users should calibrate to
the metrics they are interested in estimating and/or forecasting. Case 1 showed an
example whereby calibrating to mean link travel time may result in erroneous esti-
mates of link travel time distribution. If the ultimate use of the model is predicting
link travel time distribution then the model should be calibrated to this metric.
4) It is important for the users to understand the underlying logic of the models and
the calibration procedures. This is particularly true for simulation models that can be
categorized as having many parameters and are stochastic in nature. The users are
ultimately responsible for any errors that the models produce and it is there responsi-
bility to have a deep understanding of the literature on the subject. As this paper
demonstrated simply running a model using default values may lead to incorrect re-
sults.
5) If, after a thorough literature review, it is unclear what the assumptions of the
model are, how it was calibrated, how the underlying methodology works, etc. then
the user should reach out to the model developers. The novel and impactful models
discussed in this paper are an exciting improvement on current techniques. If there
are issues the model developers are often more than happy to discuss these with users.

6 Acknowledgements

The work discussed in this paper was based on the work of a number of former and
current graduate students including Seung Jun Kim, Wonho Kim, Jianan Zhou, Ernest
Tufuor, Justice Appiah, and Bhaven Naik and a number of colleagues including Cliff
Spiegelman, Eun Sug Park, and Elizabeth Jones. I am particularly grateful to George
List, Nagui Rouphail, and Jim Bonneson, who took lead roles in developing the inno-
vative HCM-6 methodologies described in this paper and who were very gracious in
32 L. R. Rilett

sharing their expertise on how the HCM-6 models were developed, calibrated, and
validated.

References
1. &&DUULRQDQG'/HYLQVRQ³9DOXDWLRQRIWUDYHOWLPHUHOLDELOLW\IURPD*36-based exper-
LPHQWDOGHVLJQ´Transportation Research Part C, vol. 35, pp. 305-323, 2013.
2. +*DQ;<HDQG4:DQJ³,QYHVWLJDWLQJWKHHIIHFWRIWUDYHOWLPHYDULDELOLW\RQGULYHUV

URXWHFKRLFHGHFLVLRQVLQ6KDQJKDL&KLQD´7UDQVSRUWDWLRQ3ODQQLQJDQG7HFKQRORJ\YRO
33, no. 8, pp. 657-669, 2010.
3. -:& YDQ /LQW ³0RQLWRULQJ DQG SUHGLFWLQJ IUHHZD\ WUDYHO WLPH UHOLDELOLW\ XVLQJ ZLGWK
and skew of the day-to-day-WUDYHOWLPHGLVWULEXWLRQ´3UHVHQWHGDWWKHWK7UDQVSRUWDWLRQ
Research Board Annual Meeting, Washington, DC. 2005.
4. W. L. (LVHOH%1DLNDQG/55LOHWW³(VWLPDWLQJURXWHWUDYHOWLPHUHOLDELOLW\IURPVLPXl-
WDQHRXVO\ FROOHFWHG OLQN DQG URXWH YHKLFOH SUREH GDWD DQG URDGZD\ VHQVRU GDWD´ ,QWHUQa-
tional Journal of Urban Sciences, vol. 19, no. 3, pp. 286-304, 2015.
5. FHWA. Travel time reliability: Making it there on time, all the time. Federal Highway
Administration, US Department of Transportation, Office of Operations Report 2006,
https://ops.fhwa.dot.gov/publications/tt_reliability/TTR_Report.htm Accessed Feb. 5,
2018
6. Highway Capacity Manual, 2016. Transportation Research Board, National Research
Council, Washington, D.C., 2016.
7. Spiegelman, Clifford, Eun Sug Park, and Laurence R. Rilett. Transportation Statistics and
Microsimulation. Chapman and Hall/CRC, 2011.
8. Kim, S.K., W. Kim, DQG/55LOHWW³&DOLEUDWLRQRI0LFUR-Simulation Models using Non-
3DUDPHWULF 6WDWLVWLFDO 7HFKQLTXHV´ Transportation Research Record, no. 1935 (2005):
111-119.
9. Dowling, R., List, G., Yang, B., Witzke, E. and Flannery, A. Incorporating Truck Analysis
into the Highway Capacity Manual (No. Project NCFRP-41), 2014.
10. Yang, B. On the HCM's Treatment of Trucks on Freeways. Master thesis of North
Carolina State University, 2013.
11. =KRX-LDQDQ/55LOHWWDQG(-RQHV³Estimating Passenger Car Equivalent Using the
HCM-6 PCE Methodology on Four-Lane Level Freeway Segments in Western U.S.´
Recommended for publication in the Transportation Research Record: Journal of the
Transportation Research Board.
12. Zhou, J. / 5 5LOHWW ( -RQHVDQG < &KHQ ³(VWLPDWLQJ3DVVHnger Car Equivalents on
Level Freeway Segments Experiencing High Truck Percentages and Differential Average
6SHHGV´Transportation Research Record: Journal of the Transportation Research Board,
Vol. 2672 (15) (2018): 44-54: DOI: 10.1177/0361198118798237.
13. List, G., Rouphail, N., Yang, B. PCE Value for Single Grades, Working Paper #9,
Continuation of NCFRP 41: Incorporating Truck Analysis into the Highway Capacity
Manual, under NCHRP 3-115, 2014.
14. Zegeer, John, et al. Incorporating travel time reliability into the Highway Capacity Manu-
al.[supporting datasets]. No. SHRP 2 Report S2-L08-RW-1. United States. National
Transportation Library [distributor], 2013.
15. Tufuor, Ernest OA, and Laurence R. Rilett. Analysis of Low Cost Bluetooth-Plus-WiFi
Device for Travel Time Research. No. 18-05992. 2018.
Using Simulation to Estimate and Forecast … 33

16. 7XIXRU(UQHVWDQG/55LOHWW³Validation of the HCM Urban Street Travel Time Relia-

bility Methodology using Empirical Data´5HFRPPHQGHGIRUSXEOLFDWLRQLQWKHTranspor-
tation Research Record: Journal of the Transportation Research Board.
 eHighway ± An Infrastructure
for Sustainable Road Freight Transport

Manfred Boltze

Technische Universität Darmstadt, Institute of Transport Planning and Traffic Engineering,

Otto-Berndt-Str. 2, 64287 Darmstadt, Germany.
boltze@verkehr.tu-darmstadt.de

Abstract. In all countries, road transport with heavy vehicles will have a high
share in future freight transport. When using fossil fuels for those vehicles, the
resulting emissions of carbon dioxide and pollutants such as particulate matter
and nitrogen dioxide put a risk on the environment and on human health. Since
heavy vehicles are already contributing significantly to greenhouse gas emis-
VLRQVWKHH+LJKZD\´V\VWHPZDVGHYHORSHGDQGLWZDVLGHQWLILHGDVDSRVVLEOH
solution to counteract these problems.
The eHighway system allows trucks to be fed with electric energy from a ca-
tenary which is mounted above the road. An efficient implementation of such
system requires studies in different fields to understand the impacts of the
eHighway system on today¶VURDGLQIUDVWUXFWXUH)RUWKDWSXUSRVHWKH*HUPDQ
Federal Ministry for the Environment, Nature Conservation and Nuclear Safety
is funding three field trials, and the ELISA project is the most advanced of
them. In ELISA, a 5 kilometer section of the motorway A5 in the German Fed-
eral State of Hessen is equipped with the eHighway system in both directions.
The project is led by the respective road authority Hessen Mobil, and Tech-
nische Universität Darmstadt is responsible for the accompanying research.
This contribution explains the contribution of road freight transport to
greenhouse gas emissions and discusses approaches to make this transport sec-
tor more sustainable. The basics of the eHighway technology are briefly pre-
sented. The layout of the ELISA test track on motorway A5, the arrangements
for the test operation, and the research program for the field test are described.

Keywords: Road infrastructure, road freight transport, electric road systems.

1 Introduction

1.1 Contribution of Road Freight Transport to GHG Emissions

In Germany, the transport sector is crucial to achieve national climate targets, and it
bears serious challenges. Transport stands for about 30 % of the total final energy
consumed in Germany, and 82 % thereof is used by road transport [1]. Due to growth

© Springer Nature Singapore Pte Ltd. 2020 35

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_4
36 M. Boltze

in transport demand and despite all political intentions to reduce greenhouse gas
emissions, since 1990 no significant reduction could be achieved, but nevertheless, by
2050 carbon emissions from transport shall be nearly completely avoided.
It comes along that local emissions of pollutants from combustion engines are seri-
ously affecting the health of road users and roadside residents. In Germany, it is al-
ready foreseeable that no later than after tackling the nitrogen oxide problems we will
turn again to particulate matter problems, and this time to the hazardous health im-
pacts from ultrafine particles. Therefore, the speedy conversion towards zero-
emission engines is urgently needed for multiple reasons.
Besides all efforts to disseminate electric vehicles for passenger transport, also
road freight transport needs effective measures towards zero emissions. Heavy vehi-
cles cause about 35 % of road transport related greenhouse gas emissions and more
than 6 % of the total CO2-emissions in Germany [2]. A larger modal shift from con-
tinuously growing road freight traffic to rail seems not realistic due to limited capacity
in the German rail network and due to the demand of companies for direct transport.
Current forecasts [3] estimate further growing freight transport demand and a share of
rail transport which will not significantly exceed the current share of 18 % (based on
tkm). Therefore, also in future, road freight transport will have to bear more than 70
% of the freight transport performance. Since trucks in Germany are nearly complete-
ly driven by diesel engines, serious need for action exists despite all desirable effort
for modal shift.

1.2 Approaches to Sustainable Road Freight Transport

There are certainly many approaches to achieve sustainability in road freight
transport. Freight transport demand management is a key concept, and it includes
influencing production concepts, supply chains, logistic concepts, mode choice, load
rates and many other factors [4]. However, in this article, the focus is on the types of
energy and drives used for trucks.
Requirements for the form of energy and for the type of engine used in trucks de-
pend strongly on the operational conditions, e.g. the use in regional transport or in
long-distance transport. Therefore, to achieve zero emissions from freight transport, it
will be necessary to follow several technical approaches. For example, this includes
also solely battery-electric trucks or trucks with fuel-cells. Recent studies, e.g. [5],
have elaborated the strengths and weaknesses of such technical approaches, and they
have clearly indicated that feeding trucks with electricity by using a road-side catena-
ry can be a solution which is also economically sustainable.
While it seems well feasible to apply solely battery-electric drives for passenger
cars, light-duty commercial vehicles, and even heavy trucks in regional traffic, such
solution is quite unsuitable for heavy vehicles in long-distance traffic due to the
weight of the battery. Driving a 40t-truck for 800 km from Hamburg to Munich would
need about 1,.700 kWh (assuming an energy consumption of 1.,5 kWh/km and a dis-
charge depth of max. 70 % to prevent the battery from damage). Even for a lithium-
ion battery with an energy density of 200 Wh/kg that would lead to a battery weight
eHighway – An Infrastructure for Sustainable Road … 37

of 8.,5 t. For trucks with such heavy batteries the reduced payload and battery costs
are as critical as the availability of raw materials for the batteries. To improve that
situation significantly, it needs radical innovations in energy storage which are not
predictable.
To get along with smaller and lighter batteries, these have to be charged frequently
on the ride ± as possible in motion. In the so-called eHighway system, the catenary
therefore is serving two purposes: to feed the electric engine of the truck with energy
and to charge the small on-board battery again and again. Currently it is assumed that
after a motorway section with catenary of e.g. 20 km at least the same distance could
be driven on battery. Therefore, it is not necessary to build catenaries at sections with
constructional difficulties (e.g. complex junctions or tunnels). By using the battery,
heavy trucks may also use the electric drive outside motorways, e.g. in urban areas.
And finally, the eHighway infrastructure on the motorways could also be used by
trucks for regional transport and by coaches.
Alternatives to catenaries for supplying the energy to the vehicles, e.g. via a power
rail or by inductive energy supply mounted under the road surface, would certainly
have the advantage that they could be used by passenger cars, as well. However, on
the other hand, such constructional intervention in the road surface, the robustness
under settings of the roadway, and the grip of the road surface are examples for poten-
tial problems. In contrast, the catenary is a robust technology with comprehensive
experience from railway, tram and trolleybus operation.
After a series of preceding studies [8] [9] and a first field test on a public road in
Sweden (Region Gävleborg), there are now three field trials in Germany which are
financed by the German Federal Ministry for the Environment, Nature Conservation
and Nuclear Safety. They shall investigate the practical applicability of the catenary
system. The first of these projects is ELISA ± Elektrifizierter, innovativer
Schwerverkehr auf Autobahnen (Electrified, innovative heavy vehicle traffic on mo-
torways). The ELISA test track is located in the German Federal State of Hessen, on
motorway A5, between Darmstadt and Frankfurt am Main. The other two projects are
FESH, implemented on motorway A1 near Lübeck in the Northern Federal State of
Schleswig-Holstein, and eWayBW, implemented on the federal road B462 near Kup-
penheim in the State of Baden-Württemberg.

2 Basics of the eHighway technology

The main elements of the eHighway system are the vehicles with pantograph, battery
and electric drive, and the road-side infrastructure with the catenary system and the
power supply.
The eHighwa\WUXFNVDUHHTXLSSHGZLWKDSDQWRJUDSKZKLFK³FDQFRQQHFWDQGGLs-
connect the vehicle to the contact lines at all speeds. The pantograph transfers energy
directly from the overhead contact line to the electric motor of the eHighway adapted
truck. A specially designed sensor technology allows the movable pantograph to au-
38 M. Boltze

tomatically adjust its position under the contact line to compensate for lateral move-
PHQWVRIWKHWUXFNLQWKHODQH´>@
The electric drive of the vehicle is served with power via the pantograph and the
catenary system. The same power connection is used to charge an on-board battery
(e.g. with 30 or 60 kWh capacity). When not connected to external power supply, the
vehicle will use the energy from the battery for the electric drive. As no electric ener-
gy is available ± neither from the catenary system nor from the battery ± a conven-
WLRQDOGLHVHOHQJLQHZLOOGULYHWKHYHKLFOH³'XULQJEUDNLQJWKHHOHFWULFPDFKLQHVDFW
as generators, which allows otherwise unused braking energy to be fed into the on-
board energy storage or back into the grid to be used as needed by other trucks on the
H+LJKZD\´>@7KHQHZHVWJHQHUDWLRQRIH+LJKZD\WUXFNVLVDVHPLWUDLOHUWUDFWRUDV
shown in figure 1.

Fig. 1. The New Generation of eHighway Trucks (Photo: © Scania 2018)

Main road-side element of the power supply for the vehicles is the two-pole con-
tact line which is mounted in the catenary system (figure 2). The masts are mounted
along the road in a distance of 50 to 60 meters to each other. The energy is fed by
substations along the eHighway which are connected to the medium-voltage power
grid. The voltage at the catenary is 670V direct current (DC).
eHighway – An Infrastructure for Sustainable Road … 39

Fig. 2. Detail of the Catenary System (Photo: © IVV 2019)

3 The ELISA Field Test

3.1 ELISA Test Track

ELISA is a project under the lead of the Road Authority Hessen Mobil. The Institute
of Transport Planning and Traffic Engineering at Technische Universität Darmstadt is
coordinating the scientific research program. Further project partners are Entega AG
(electricity supplier), Siemens AG, and several transport partners. The acronym
ELISA stands for Elektrifizierter, innovativer Schwerverkehr auf Autobahnen (Elec-
trified, innvoative heavy vehicle traffic on motorways). As the first of the three Ger-
man eHighway field tests, ELISA officially started the test operation on 7 May 2019.
The project will end by December 2022.
The location of the ELISA test track is on motorway A5 between Frankfurt am
Main and Darmstadt (figure 3). This section is a most frequented German federal
highway with 4 lanes per direction, bearing an Average Daily Traffic Volume of 135
.000 vehicles per day, with about 10 % heavy vehicles. After the planning, approval,
and bidding processes undergone in the year 2017, in 2018 the road-side eHighway
infrastructure was built on both sides of the motorway. The equipped section is 5 km
long, has 229 masts, and 2 power supply substations (figure 4).
40 M. Boltze

Fig. 3. Location of the ELISA Test Track

Many new aspects were coming up until the test track operation could be finally
started in May 2019. For example, the approval process needed careful communica-
tions and decisions. Specific constructional issues needed to be handled, e.g. the lack
of space around a rest area was solved by pylons on the centre island (figure 5). The
monitoring and control of the eHighway infrastructure operation needed to be inte-
grated into the Hessen Traffic Control Centre. It was also necessary to have a close
exchange with emergency and rescue service operators, and to develop adapted strat-
egies for incidents and accidents.

Fig. 4. Motorway A5 Equipped with the eHighway Infrastructure (Photo: © IVV 2019)
eHighway – An Infrastructure for Sustainable Road … 41

Fig. 5. eHighway Infrastructure with Pylons on the Center Island (Photo: © IVV 2019)

3.2 ELISA Test Vehicles

As for the truck operation, there was a lot of interest and positive perception of the
eHighway field test among transport companies. Looking for diversity, five compa-
nies with quite different profiles have been selected (table 1), each of them to use one
eHighway test vehicle in daily operation. All test vehicles are semitrailer tractors, as
shown in figure 1. All selected companies are offering a minimum of three daily tours
with the test vehicle forth and back along the test track. These five transport partners
and also some others which cannot get a suitable vehicle for the moment (like Hegro
Eichler or Heldmann), are helping a lot to understand the practical issues of using
eHighway trucks. In addition, research trucks from Scania, Volkswagen, and Siemens
will be using the ELISA test track.

Table 1. ELISA Transport Partners

42 M. Boltze

3.3 ELISA Research Program

As for the roles and tasks of the eHighway field tests, it is helpful to have a look at the
so-called Technology Readiness Levels (TRL) [7], as shown in figure 6. They were
originally conceived at NASA in 1974 for space technology development. Later, in
1989, they were formally defined. Taking the eHighway system as an example, we
can say that the ENUBA project [8] brought it on level 3, from basic principles and
technology concept formulation to experimental proof of concept.

Fig. 6. Technology Readiness Levels of the eHighway System (based on [7])

The most obvious result of ENUBA 2 [9] was the test track in Großdölln (Germa-
ny), which was a full-scale test field, but not on public road space and therefore under
lab conditions. It achieved TRL 5. The current eHighway field tests are starting at
level 6. The technology is demonstrated in the relevant environment, under realistic
conditions, and this is continued for a period of several years. This is where we can
clearly allocate the Swedish eHighway demonstration projects in Borlänge-Gävle and
the earlier test at the Port of Los Angeles. And the three German test tracks, including
the ELISA field test, are now going several steps further.
From such systematic view on the TRLs, we can identify a major role of the cur-
rent eHighway field tests as validators and demonstrators. For that purpose, a detailed
research program and a comprehensive concept for data collection, processing and
analysis was set up [10].
In this contribution, only a few sample research questions can be mentioned. It is
certainly one of the main tasks of ELISA to investigate the real savings in CO 2 emis-
sions when operating the eHighway trucks in practice. ELISA will also investigate the
impacts of the system on local levels of noise and air pollution. Several research ques-
tions are regarding traffic flow, e.g. the impacts of the eHighway system on the visi-
bility of traffic signs, on driving behaviour and lane choice, on the speed choice of
eHighway truck drivers and on their readiness to leave the catenary lane for overtak-
eHighway – An Infrastructure for Sustainable Road … 43

ing slower vehicles. Regarding road operations, it needs to be investigated if there are
potential complications in cleaning traffic signs or cutting the green, or how a de-
icing of the overhead wires could be done safely and efficiently.
Of course, potential impacts on traffic safety are also carefully observed. The per-
ception and acceptance of the eHighway system by different stakeholder groups is
another research issue. Which factors are influencing the acceptance rate, and how are
these acceptance rates changing over time? Sample research questions related to the
transport partners are addressing the specific requirements of different types of
transport companies on using the eHighway system, the integration of eHighway
trucks into daily operation, and the robustness of the eHighway technology and vehi-
cle components under frequent use.
For some of these questions, we are getting first preliminary answers, already. For
example, traffic counts from loop detectors are indicating that there is no or very little
influence of the eHighway infrastructure on lane choice. Consistently, in a question-
naire pre-test, 96 % of all car drivers stated that the eHighway would not influence
their driving behaviour and that they will use the right lane in the same manner they
did before. However, also 96 % said that they are feeling somewhat insecure driving
under the catenary (which might be in contrast to their feelings under the tram´s cate-
nary in urban areas). In the same pre-test, about 87 % of car drives stated that they
observed no reduction in the visibility of traffic signs, while 4 % stated that they felt
strongly disturbed.
Field tests are also about real incidents and even accidents. We certainly cannot
imagine all potential problems that might come up in practice. For example, it was not
expected that a camper van might burn directly under the catenary. But it happened
already, and fortunately, after careful investigation no significant damage was found.
Anyway, for the field tests clear routines must be defined for all cases of incidents
and accidents.

4 Outlook

The scientific research program of the ELISA field test for the eHighway system has
just started. We can expect comprehensive results from this experiment which will
support decisions on the large-scale implementation of the eHighway system. The
other field tests in Germany, several accompanying projects on socio-economic issues
and roll-out strategies, the projects on electric road systems in Sweden and potential
future projects in other European countries, such as Italy and France, will all help to
shape the best solution. In addition, the applicability to other types of vehicles (such
as coaches) or to other countries (such as Vietnam with its dedicated North-South
road transport corridor) should be investigated.
The future of the eHighway system will depend on positive results from these pro-
jects. But a large-scale implementation will also need clear political decisions, inter-
national cooperation and standardization, and contributions from the automotive in-
dustry and many other stakeholders.
44 M. Boltze

There is certainly an urgent need for making road freight transport more sustaina-
ble. As we do not have so many alternatives, the eHighway system seems to be an
option very worthwhile to follow. Without any doubt, this will create some costs.
However, what we cannot afford at all is doing nothing.

References
1. Bundesministerium für Verkehr und digitale Infrastruktur: Verkehr in Zahlen ± Ausgabe
2018/19. Berlin/Germany (2018)
2. Bundesministerium für Umwelt, Naturschutz und nukleare Sicherheit: Klimaschutz in
Zahlen ± Ausgabe 2018. Berlin/Germany (2018)
3. Schubert et al.: Verkehrsverflechtungsprognose 2030 ± Zusammenfassung der Ergebnisse.
Research Report FE-Nr.:96.0981/2011. Freiburg, München, Aachen, Essen/Germany
(2014).
4. Rühl, F., Boltze, M.: Freight Transport Demand Management: Influencing the Freight
Transport Demand with Traffic Management. In: Abele, E., Boltze, M. Pfohl, H.-C. (eds.):
Dynamic and Seamless Integration of Production, Logistic and Traffic. pp. 163-184.
Springer, Cham/Switzerland (2017)
5. Kühnel, S., Hacker, F., Görz, W.: Oberleitungs-Lkw im Kontext weiterer Antriebs- und
Energieversorgungsoptionen für den Straßengüterfernverkehr. Editor: Öko-Institut, Ber-
lin/Germany (2018)
6. Siemens AG: eHighway ± Innovative electric road freight transport. Information brochure,
Munich/Germany (2017)
7. Deutsches Institut für Normung (DIN): Raumfahrtsysteme ± Definition des Technologie-
Reifegrades (TRL) und der Beurteilungskriterien. Norm-Entwurf (ISO 16290:2013). Ber-
lin/Germany (2014)
8. Siemens AG: ENUBA ± Abschlussbericht der Siemens AG. Erlangen/Germany (2011).
9. Siemens AG: ENUBA II ± Abschlussbericht zum Förderprojekt ENUBA II. Erlan-
gen/Germany (2016).
10. Boltze, M., Wauri, D.: Evaluation Concept for the ELISA eHighway System Field Trial.
Contribution to the 2nd Electric Road Systems Conference. Arlanda/Sweden (2018).
 Innovation Reduces Risk for Sustainable Infrastructure

Suzanne Lacasse1
1 Norwegian Geotechnical Institute (NGI), Oslo, NORWAY
sl@ngi.no

Abstract. Society and standards require more and more "risk-informed" decisions.
The paper demonstrates the potential of reducing risk by implementing reliability
and risk concepts as a complement to conventional analyses. Reliability evalua-
tions can range from qualitative estimates, simple statistical evaluations to full
quantitative probabilistic modelling of the hazards and consequences. The paper
first introduced recent innovative developments that help reduce risk. Risk assess-
ment and risk management are briefly touched upon. An example of the applica-
tion of the new stress testing method is given. The usefulness of the seminal (1969)
Observational Method is discussed. The need for developing sustainable and ho-
listic civil engineering solutions is also briefly mentioned. The paper concludes
that reliability-based approaches provide useful complementary information, and
enable the analysis of complex uncertainties in a systematic and more complete
manner than deterministic analyses alone. There is today a cultural shift in the
approach for design and risk reduction in our profession. Reliability and risk-based
approaches will assist preparing sustainable engineering recommendations and
making risk-informed decisions.

Keywords: Risk Assessment, Risk Management, Remote Sensing, Stress Test-

ing.

1 Introduction

Science is often seen as a pure search for truth and knowledge. Engineering is the im-
plementation of science to serve society. To make science and engineering meaningful,
"at some point, you really need to save a life" (trial lawyer Fran Visco at the US Capitol,
2007).
In a survey on the state of practice and preparedness for the impact of climate change
on landslide risk [1], 21 countries/regions were concerned with four issues: (1) Wide-
spread landslides, including massive landslides, with increased propensity, scale and
mobility; (2) Widely affected populations, together with breakdown of infrastructure;
(3) Lack of understanding of extreme landslide scenarios, partly due to the scarcity of
their earlier occurrence; and (4) Lack of adequate preparedness for extreme landslides,
with aggravating consequences. The major needs included: (1) Quantifying the uncer-
tainties; (2) Insights from scenario-based assessments; (3) Application of improved tech-

© Springer Nature Singapore Pte Ltd. 2020 45

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_5
46 S. Lacasse

nology and methods; (4) Enhanced communication; (5) Addressing policy gaps, includ-
ing national and international cooperation; and (6) Formulating strategies for a way for-
ward.
This paper looks into risk reduction for a sustainable infrastructure, and reviews re-
cent technological advances and how they will improve risk reduction and thereby con-
tribute to geosciences serving society.

2 Innovation for Saving Lives

2.1 Risk Reduction

Risk reduction, for example sue to landslide hazard, requires an understanding of the
physical system (the unstable slope) and the context associated with the landslide (i.e.
who is endangered, what are the consequences). There are innovative technologies that
can address the needs for information (data) and understanding the methodology re-
quired for interpreting the data (models and other tools). In addition to ‘engineering’,
innovative approaches also comprise the ‘softer sciences’, such as perception and com-
munication. For effective risk reduction (Fig. 1), communication and education are pre-
requisites. Techniques for measuring and sensing variables to provide the data for mod-
elling and forecasting have a variable degree of success. To understand and predict the
behaviour of the physical system, theoretical analyses, experience and engineering judg-
ment are key for parameter selection and result interpretation.

Physical System Context

Commu
Communication
Education
Human
influences Exposure
Modelling and
Geo- Risk evaluation Conse-
Variables forecasting
tools quence
tools
Other
Perception
variables

Mitigation measures
Fig. 1. Risk reduction system.

Modelling and forecasting. Tools for modelling and forecasting depend on natural var-
iables and human factors:

x Anthropogenic impact, e.g. land use, construction, deforestation, climate change.

x Natural variables, e.g. precipitation, infiltration/runoff, time, soil properties.
x "Non-variables" (invariables), e.g. topography, geology, hydrogeology, etc.
Innovation Reduces Risk for Sustainable Infrastructure 47

Risk evaluation. Risk evaluation covers a range of issues. Both quantitative and quali-
tative methods are used, e.g. risk matrices, event trees, indicator-based methods and the
quantitative reliability-based methods. Risk evaluation and risk reduction need to con-
sider (1) exposure for population, infrastructure, economic values, coinciding events,
changes in land use patterns, development works, demography, as well as political
trends; (2) consequences, including loss of life, injury, property and economical losses,
disruptions, environmental losses and losses of essential social services; and (3) percep-
tion by stakeholders, authorities and the general public.

Mitigation measures, communication and education. Mitigation is the implementa-

tion of actions to reduce probability of occurrence (hazard) and/or reduce exposure and
hence reduce the risk. Mitigation measures include civil works, early warning systems,
influencing societal trends, land use planning, political decisions, avoidance, and so on.
Physical and non-physical (e.g. insurance) mitigation measures are both important.
Communication is usually the element that fails in major disasters. Yet, together with
education, communication should be the essential and unifying factor of a risk reduction
strategy. Technology plays an important role in facilitating education and communica-
tion associated with landslide risk. A significant ingredient of successful risk manage-
ment is the recognition that the risk is never zero.

2.2 Technological advances

Sensing technology. Sensing technology is deployed to, e.g., warn of danger, follow the
evolution of a hazard, monitor the effect of mitigation measures and provide feedback
to modify and adjust measures. The techniques for measuring variables (Fig. 2) fall into
two categories: (1) point sensing (measuring one variable in a single location or over a
relatively small area) and (2) wide area sensing (measuring parameters over a large area
or volume often producing an image), including information from satellites (Fig. 3). The
early trends of few dedicated monitoring with high quality expensive electronics and
high accuracy measurements have moved to vast quantity of sensors, ad hoc data
sources, low quality-low cost-low accuracy, but very many, data, sophisticated pro-
cessing and interpretation tools based on statistics, machine learning and artificial intel-
ligence. Satellites cover wide areas, repeatedly and at regular intervals, and can access
even the most remote areas. Resolution (pixel size) is continuously improving and new
sensors are constantly being developed. The data are easily accessible, usually at a cost,
but often at reduced fee for academic purposes.
Remote sensing techniques (Fig. 4) include drones. The June 11, 2018 issue of the
newsmagazine TIME devoted its cover page story to "The Drone Age". The flying tech-
nology is transforming the world, from military strategy to life-saving innovations in
health care to creating new works of art. In these respect, drones are ideal for reducing
landslide risk. When Hurricane Maria hit Puerto Rico in September 2017, it ravaged the
island's electrical grid and communications systems, and no communications were pos-
sible with mainland U.S. AT&T, with its drone COW ("cell on wheels"), re-established
Internet connection and communication.
48 S. Lacasse

Fig. 2. Single point (left) and areal sensing/imaging with remote radar technique (right).

Fig. 3. Network of satellites and drone in flight (inset) (images by NASA and DJI.com).

Remote sensing
InRAR
InSAR
LiDAR
Optical satellite images
Drone
LiDAR
Gigapan
Aerial photos

Fig. 4. Newer remote sensing technologies (Hefre, H.H.. NGI. Personal communication 2018).
Innovation Reduces Risk for Sustainable Infrastructure 49

With any technology, there is an inflection point where it goes from something in the
near future to being a part of everyday life (thing mobile phone!). The time has come for
drones. Farmers use drones to monitor and spray crops, first responders use drones to
coordinate operations and search for missing hikers. Facebook works with drones to
beam Internet connectivity to isolated corners of the world, Shell uses drones to keep its
network of offshore rigs running smoothly. Artificial intelligence, automation and inter-
connectivity will enable the simultaneous operation of massive drone fleets, increasing
efficiency and greatly expanding their capabilities.

Big Data, Machine Learning and IoT. There is a paradigm shift from dedicated, high
quality small scale monitoring systems towards large, distributed systems with vast num-
bers of data. Big Data are "the three Vs": Volume (high quantities of data), Velocity
(often real time), and Variety (text, images, audio, video, data fusion), necessitating new
tools, new data management and new system architecture for enhanced insight, decision-
making and automation. The focus has shifted from designing a system (e.g. specific
sensors at specific locations) to observing a response (e.g. streams of data with direct
and indirect relationships to the underlying physical processes).Artificial intelligence
(AI) creates technology that allows computers and machines to function in an intelligent
manner. Will AI/ML defeat/replace professional engineers in the near future? Machine
Learning (ML) works [2], but it will not replace engineering judgment. It helps with
added knowledge and may well become standard procedure in a not too distant future.
ML is already used to predict several geotechnical aspects, including:

x Ground displacement due to rainfall and reservoir level, Three Gorges area in China.
x Landslide susceptibility and depth to new stratum, e.g. bedrock or sensitive clay layer.
x Cone resistance, pore pressure and side friction from cone penetration tests.
x Soil resistance under installation of suction caisson installation offshore.
x Interpretation of geophysical data and tunnel defect recognition.
x Leak-off pressure (hydraulic fracturing and horizontal stress) in rock formation.

Forthcoming interconnectivity will enable using the position, movement and use of
mobile phones to help managing disaster response. The position information for mobile
phones can help assess the quantity of traffic on roadways. Trending information on
social media and advanced processing using big data and/or machine learning can help
predict exposure by identifying where people are likely to be at any given time. How-
ever, as data quantity increases, the quality of the data probably decreases, requiring
application of big data and machine learning techniques and advanced statistical tech-
niques. The DESR laboratory at the Hong Kong University of Science and Technology
[http://ceyhwang.people.ust.hk/Lab/DESR_lab.htm] specializes in the applications of
the "Geotechnical Internet of Things" (Geo-IoT), Deep Learning and Big Data Analytics
for sustainable city development, e.g. critical infrastructure, slope monitoring and open
platforms for geotechnical industries to collaborate and share resources.
In a single Internet minute (2013), 360,000 tweets, 3.4 million searches, 500 hours of
video, 2.4 million Gigabytes of data, 300,000 photographs, 4 million "shares" and 1.8
billion users have been recorded. In one day, Facebook transfers 25 Terabytes of data.
50 S. Lacasse

The abundance of data available leads to increased use of statistical methods and ma-
chine learning techniques and less use of the "pure science" approach. The IoT allows
objects to be sensed remotely across existing networks, resulting in improved efficiency,
accuracy and embedded computing system. Experts estimate that the IoT will consist of
30 to 50 billion objects by 2020.

2.3 A Changing Climate

Climate is defined in terms of variations in atmospheric conditions (temperature, precip-
itation, wind velocity and direction, cloudiness, atmospheric pressure, air humidity and
chemistry of the air and contained dust). Landslide and rock slide hazards increase with
increased intensity and frequency of rainfall [1]. In a context of (1) more extreme
weather and more rainfall, (2) human population increasing at an exponential rate, and
thereby increasing vulnerability, (3) human population moving towards coastal areas
(www. Livesciences.com) and (4) increase in news reporting on natural disasters, espe-
cially since the invention of the Internet, the risk associated with climate-driven natural
hazards is expected to increase. The survey from 21 countries/regions [1], and especially
in permafrost and ice areas, revealed that most of the countries and regions expect in-
creasing landslide hazards and increasing risk to population. The concerns varied
greatly, depending on weather patterns, soil types, past landslide history and the nature
of the triggers (earthquake, floods, rainfall, etc.). The countries/regions were, however,
at different stages of preparedness for climate change impact. Hongkong has come fur-
thest in terms of holistic and systematic assessment of impact of extreme weather events
on slope safety through stress testing. All acknowledged the difficulty in preparing real-
istic scenarios for improved preparedness, but early warning systems and improved com-
munication were seen as the most effective landslide risk reduction measures.

3 Risk Assessment and Management

3.1 Hazard assessment

At the completion of the European SafeLand project, Corominas et al. [3] illustrated a
scenario-based risk assessment framework (Fig. 5), where the aforementioned compo-
nents, with enhanced technology, can be included. To modernize hazard and risk map-
ping and assessment, one should: (1) Improve the methodologies for hazard mapping;
(2) Increase the efficiency and quality of data collection; (3) Include the available data-
bases (e.g. drone technology and satellite observations) in the information sources; (4)
Facilitate the application of new technology, including AI/ML; and (5) Include climate
and extreme weather impacts.

3.2 Risk Management

Landslide risk management refers to the coordinated activities to assess, direct and con-
trol the risks posed to society by gravity mass flows and landslides. Risk management
Innovation Reduces Risk for Sustainable Infrastructure 51

integrates the recognition of risk and the development of appropriate strategies for the
reduction of the risk. Risk management needs to be a systematic application of manage-
ment policies, procedures and practices, including context definition identification, anal-
ysis, evaluation, communication, consultation, implementation and monitoring of miti-
gation measures.

Fig. 5. Landslide risk analysis by scenarios [modified after [3]).

A "modern" landslide risk reduction strategy would encompass:

x Avoidance, with land-use planning, warning or alert systems, and public education.
x Prevention, e.g. enforcing slope investigation, design, supervision and maintenance.
x Mitigation, with implementing engineering measures to reduce the impact.
x Preparedness, resource management, emergency systems and community training.
x Response, with humanitarian relief, inspections, safe re-settlement of people etc.
x Recovery to bring the affected area back to the normal and do mitigation works.
x Working in a network for climate adaptation.
The newest within landslide risk mitigation is nature-based solutions (NBSs). NBS is
an umbrella term for 'ecosystem services', 'green–blue infrastructure' and 'natural capital'
[4]. The premise for NBSs is that nature itself is a source of ideas and solutions for
mitigating risk from climate-driven natural hazards. Nature's designs are elegant, effec-
tive and frugal and adapted to their environment. Water is usually the main culprit in
triggering of landslides and debris flows. Planting trees and adapted vegetation to control
surface erosion and strengthen the surficial soil layers exposed to excessive amounts of
runoff water during extreme precipitation events is one of the few established NBSs for
landslide risk mitigation. Several large ongoing research projects in EU's H2020 frame-
work programme, e.g. PHUSICOS [www.phusicos.eu], focus on developing innovative
52 S. Lacasse

NBSs for landslide control, including comprehensive frameworks for monitoring and
verifying the performance of NBSs, i.e. a direct application of Peck's observational
method [5]. The monitoring would enable to identify the intensity of extreme events that
the NBSs could withstand.

Cascading Events. The experience from the events following the 2008 Wenchuan earth-
quake in China led to a complex hazard-response chain of mass movement processes
[6], describing dynamic transitions among six mass movement processes (landslide, rock
fall, debris flow, landslide dam, flood and aggraded riverbed) (Fig. 6). The natural geo-
phenomena form the outer periphery of the chain. The arrows illustrate the processes
and transition that led to increased hazards. Earthquakes, heavy rainfall, can reactivate
movements and create landslides or debris flows. Landslide dams can be formed. The
evolution will not stop before the entire system reaches a new equilibrium (which will
also have a finite risk level).

Fig. 6. Hazard-response of mass movement processes following a strong earthquake (after [6]).

Shortly after the Wenchuan earthquake, slides and rock falls were dominant. With the
hillslope deposits evolving into channel sediment deposits, the rain-induced debris flows
became dominant 2 to 5 years after the earthquake. As repeated debris flows occurred,
the fine particles in the soils eroded due to surface flow. Over time, large particles be-
came exposed, leading to decreased debris flow activity. At present, the debris flow haz-
ards are still active, and debris flows can outbreak in a large storm as a tremendous
Innovation Reduces Risk for Sustainable Infrastructure 53

amount of loose material is suspended on the hill slopes, ready to be eroded and trans-
ported. The flood hazards then become dominant. The landform reflects the changes due
to the many factors that have occurred over time.
How can one develop a system to keep account of all these events and the cascading
and interrelationship among the hazards and the risks posed by each of the events? Stress
testing presents an innovative approach for dealing with such complex situations.

Extreme Events and Stress Testing. Stress testing is a new solution for situations with
low probability of occurrence and very high consequences such as extreme events. Stress
testing is a targeted reassessment of safety margins involving testing a system beyond
normal operational capacity, often to a breaking point. The approach was used to evalu-
ate and improve the slope safety system in Hong Kong under extreme rainstorms [7].
The framework included: (1) Identifying future critical rainstorm scenarios under the
changing climate; (2) Evaluating the response of the slope safety system to the critical
rainstorm scenarios; (3) Assessing the risks posed by the multi-hazard processes; (4)
Evaluating the bottlenecks of the slope safety system; and (5) Proposing strategies for
improving system performance. The analyses used correlations between landslides and
rainfall in Hong Kong, layers of digital information (e.g. rainfall intensity, surface geol-
ogy, slope gradient) and integrated geo-information in a cell-based analysis. A stress test
of the response of Hong Kong Island to four storms corresponding to 29%, 44%, 65%
and 85% of the 24-hour probable maximum precipitation (PMP) was done (Fig. 9 [7]).

Fig. 7. Landslides, debris flows and floods on Hong Kong Island, 85% PMP storm [6].
54 S. Lacasse

The number of slope failures and debris flows increased abruptly from 44 to 65% PMP.
From 65 to 85% PMP, the impact was less distinct, probably because there was less
mobile material left. The testing showed that the extreme storms require an exceptional
level of response and new strategies for multi-hazard risk management.
Stress testing is becoming a central tool for the future. In the afWHUPDWKRIWKH7ǀKRNX
earthquake and Fukushima Dai-ichi accident, stress tests were imposed by WENRA
(West European Nuclear Regulation Association) on all nuclear power plants in Europe
in 2011 and 2012.

The Observational Method. One recurring factor in geotechnical failures is that the
construction did not follow the original script. Examples include the Aznalcóllar and the
Mount Polley tailings dam failures where, among several factors, the downstream slopes
were steeper than intended. These failures reinforce the need for the "Observational
Method", a seminal deterministic method in geotechnics [5]. Even in this "Innovation
paper", it is recommended that this "old" seminal method be "modernized " with the new
technological developments and adapted for statistical analysis with the Bayesian updat-
ing method.
The Observational Method enables using monitoring to adjust a design during and
after construction. Unnecessary or costly overdesign can be avoided without compro-
mising on safety or the environment. The approach offers a number of advantages for
geotechnical projects: (1) the method uses the "most probable" design parameters, as
opposed to conservative parameters, and assesses the range of probable behaviours; (2)
it sets out the modification in construction or warning to be implemented if the parame-
ters or the behaviour turn out to be less favourable than assumed; (3) it monitors the
behaviour of the structure, prescribing the frequency of the monitoring and the trigger
levels at which corrective action is required; and (4) it analyses the data and triggers the
implementation of contingency plans.
The Observational Method could be "modernized" with Bayesian updating. Bayes'
theorem allows updating prior estimates with new information. Coupling the Observa-
tional Method to risk management, with dynamic updating of the risk and response sce-
narios would be a viable strategy to make slopes safer, and provide support for risk-
informed decision-making. With today's computer and modelling techniques, with real-
time internet access to display site data and probabilistic evaluations, the Observational
Method reflects the geotechnical engineer's ambition: a coherent approach combining
observation, analysis, judgment and risk-informed decision-making, for optimising civil
engineering from both cost and safety points of view.

4 Sustainable Civil Engineering Assessments

The demand for sustainability assessments of design will increase in the near future
within the building, construction and transportation sectors. There is an increasing need
to provide viable, bearable and equitable civil engineering solutions for environmental,
economic and social sustainability. The profession needs to:
Innovation Reduces Risk for Sustainable Infrastructure 55

x Raise awareness
x Incorporate sustainability in design
x Develop guidance on assessment of sustainability
x Estimate impacts
x Develop new tools oriented to sustainability

Such steps will contribute to deliver sustainable infrastructure solutions, provide docu-
mentation to help sustainable (and risk-informed) decision-making, reduce negative or
non-sustainable impacts in the building, construction and transportation sector, and re-
duce contributions to climate change

5 Conclusions

New challenges reside in integrating remote sensing, geotechnical engineering and risk
assessment and management into innovative and practical risk management tools to re-
duce landslide risk associated with natural and man-made geo-hazards. The result will
be a risk reduction strategy, with robust, documented and practical risk management
tools. Remote sensing techniques can identify and quantify geo-hazards and intercon-
nectivity will allow to rapidly update hazard, vulnerability and risk maps. The Observa-
tional Method, coupled with Bayesian updating will focus on risk on the basis of obser-
vations and scenarios. Stress testing will provide improved preparedness, especially for
extreme events. A risk management strategy should integrate all aspects, with special
focus on communication, education and "lessons learned" from earlier events.
Are science and engineering, and landslide risk reduction, candidates for the growing
list of failing institutions that often characterise society (e.g. democratic politics, crimi-
nal justice, health care, public education)? On the contrary! Science and engineering
have innumerable niches and ways to improve their service to society by taking on board
technological innovations. The most valuable "science and engineering" in the future
will be linked to the people needing help. This should incentivize scientists to focus on
the solutions rather than abstract knowledge. Research agendas will search for improved
technological solutions.
There is today a cultural shift in the approach for design and risk reduction in our
profession (Table 1).

Table 1. Cultural shift required for reducing risk.

From To
Hazard Consequence
Response Preparedness and Risk Reduction
Reactive Proactive
Science-driven Multi-disciplinary
Response management Risk Management
Single agencies "Everyone's business"
Planning for communities Planning with communities
Decision-making Risk-informed Decision-making
56 S. Lacasse

As a closing perspective, science and engineering helps us to predict hazards. Know-

ing the hazards and the risk helps us make risk-informed decisions, and the consequences
of hazards can be reduced with innovative technology. Human factors can turn unex-
pected events into catastrophes, and tomorrow’s disasters are exponentially building on
the disasters of today. Focus on disaster risk reduction is required because the social
costs tend to persist over a person’s lifetime, whereas material costs are a one-off. The
time to innovate for improved landslide risk reduction is now. In the future, we need to
show that we make «risk-informed» decisions (ISO2394:2015 [8]).

References
1. Ho, K.K.S., Lacasse, S. and Picarelli, L. (2017). Slope Safety Preparedness for Climate
Change Effects. Taylor & Francis. London. 572 pp.
2. Silver, D., Schrittwieser, J. and Simonyan, K. (2017). Mastering the game of Go without
human knowledge. Nature. 550. ௅2FW
3. Corominas, J., van Westen, C. Frattini, P., Cascini, L., Malet, J.P., Fotopoulou, S , Catani, F.,
Van Den Eeckhaut, M., Mavrouli, O., Agliardi, F., Pitilakis, K., Winter, M.G., Pastor, M.,
Ferlisi, S., Tofani, V., Hervás, J. and Smith, J.T. (2014). Recommendations for the quantita-
tive analysis of landslide risk. Bull. Engineering Geology and the Environment. 73 (2). 209௅
263. May 2014.
4. Nature (2017): Natural language. Nature. Editorial. 541 (12 Jan. 2017) 133௅134.
5. Peck, R.B. (1969). Advantages and Limitations of the Observational Method in Applied Soil
Mechanics. Géotechnique௅
6. Zhang, S., Zhang, L., Lacasse, S., & Nadim, F. (2016). Evolution of Mass Movements near
Epicentre of Wenchuan Earthquake, the First Eight Years. Nature. Scientific Reports. 6.
36154. doi:10.1038/srep36154.
7. Zhang, L., Gao, L., Zhou, S., Cheung, R.W.M and Lacasse, S (2017). Stress Testing Frame-
work for Managing Landslide Risks under Extreme Storms. 4th World Landslide Forum.
Ljubljana. Slovenia. pp. 17-32. http://hdl.handle.net/1783.1/85575.
8. ISO2394:2015. General principles on reliability for structures. ISO/TC 98/SC 2. 111 pp.
 Applying a national BIM model to Vietnam's National
Implementation of BIM: Lessons learned from the UK-
Vietnam collaboration for the Industry

Adam Matthews1 and Binh Ta2

1
Centre for Digital Built Britain, University of Cambridge, Cambridge CB2 1TN, UK
2
Institute of Construction Economics, Ministry of Construction Vietnam, Hanoi, Vietnam

Abstract. This paper applies the national BIM strategy method as described by
the EU BIM Task Group and the UK's Centre for Digital Built Britain to Vi-
etnam's development of a national digitalization vision. It will describe UK ex-
perience in promoting digital construction through BIM application in construc-
tion and facility management activities toward Smart city based on 4 strategic
areas: Foundation of Public leadership; Communicate vision and foster commu-
nities; Build a common, collaborative framework; and Grow Industry Capacity.
International collaboration on the national introduction of BIM is growing be-
tween countries and between governments. In 2018, the UK and Vietnam signed
an MOU for collaborating on the introduction of BIM to the national plan of
Vietnam started under the Prime Minister's decision no. 2500/QĐ-TTg dated De-
cember 22nd, 2016. This paper applies the national BIM method to identify (1)
the experience of the UK's national BIM strategy, (2) Vietnam's lessons learned
of its national implementation of BIM and (3) proposals and recommendations
for Vietnam's Construction Industry.

Keywords: BIM, UK-VN collaboration, Vietnam BIM Program, GIP.

1 Introduction

Governments across Europe and around the world are promoting the value of BIM as a
strategic enabler for cost, quality and policy goals. More and more countries had a reg-
ulatory requirement for BIM or are planning on implementing one in the near future.
According to EU BIM Handbook [1], BIM is a digital form of construction and asset
operations. It brings together technology, process improvements and digital infor-
mation to radically improve client and project outcomes and asset operations. How-
ever, BIM can also be described as “a digital representation of physical and func-
tional characteristics of a facility” [2] or “a modeling technology and associated set
of processes to produce, communicate, and analyze building models” [3].
As is often the case when significant change is introduced, it is easier to accelerate
adoption by applying a specific term. In this case, BIM is defined as a term to compare
to CAD in order for people to focus on the differences, and perhaps measure differ-
ences. BIM can be applied not only to new build projects but also the renovation,
© Springer Nature Singapore Pte Ltd. 2020 57
C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_6
58 A. Matthews and B. Ta

refurbishment and maintenance of the built environment [4]. Also, BIM concerns all
stakeholders include real estate; ownership; finance; all areas of architecture, engineer-
ing, and construction (AEC); manufacturing and fabrication; facility maintenance, op-
erations, and planning; regulatory compliance; asset management; sustainability; and
disposal within the facility lifecycle.
Nowadays, Building Information Modelling (BIM) is at the centre of a digital transfor-
mation of the construction sector and the built environment [5]. Several leading coun-
tries have finished introducing BIM and taken the steps toward digital transformation
in construction industry.

1.1 UK-Vietnam collaboration

The UK Government's Global Infrastructure Programme (GIP) aims to improve eco-
nomic prosperity in the world's fastest growing communities. This program is sup-
ported by three UK Government partners: Foreign & Commonwealth Office (FCO),
the Department for Business, Energy & Industrial Strategy (BEIS) and the Infrastruc-
ture and Projects Authority (IPA) with delivery partners.
By improving infrastructure delivery by the adoption of UK best practice methodolo-
gies, Vietnam is able to benefit from increased productivity which means better value
for money - and more resources are available to build more infrastructure.
The work is carried out over three phases (Fig. 1).
- During Phases 1 and 2 of the programme, CDBB established relationships with Vi-
etnam's government, Ministry of Construction (MoC). This involved workshops and
in-bound delegations to the UK to develop shared understanding and a collaboration
plan. This culminated in the signing in July 2018 of a Memorandum of Understanding
(MoU).
- Phase 2 in Vietnam can be summarized as the adapt and train stage. National BIM
Strategy group is formed and trained on policy, standards and methods of working to
shift towards collaboration and BIM. Training occurs at differing levels, in Phase 2 the
focus is at the national and organizational levels, enabling the decision makers to un-
derstand BIM benefits and its application. Potential pilot projects to showcase the use
of BIM will be reviewed and selected for implementation of Phase 3.

Fig. 1. Implementation phases of GIP program.

Applying a national BIM model to Vietnam's National Implementation... 59

- The third and final phase is the rollout of BIM on the selected pilot project and in-
volves support in planning and delivery of the project. The aim for this pilot project is
to become Vietnam's case study for BIM adoption, from which to build from. Learning
from this pilot project will also be crucial, to review and iterate current policy and
standards as required.

1.2 Research method

This paper references the national BIM method as described by the EU BIM Task
Group, in its EU BIM Handbook [1], and the UK Centre for Digital Built Britain.

Fig. 2. Strategic framework for public sector BIM programmes.

This strategic framework provides a common approach for the introduction of BIM by
the European public sector [1]. The framework identifies the following four strategic
areas of action that are important when developing BIM initiatives (Fig. 2): Establish-
ing public leadership; Communicating vision and foster communities; Developing a
collaborative framework; Growing client and industry capability and capacity.

2 UK experience in promoting digital construction

through BIM application in construction and facility
management activities toward Smart city.

2.1 Foundation of Public leadership

The UK Government’s Construction Strategy 2011 has set a clear vision by mandating
the use of “collaborative BIM” on all centrally procured built assets across all govern-
ment departments by 2016 [6]. This mandate was later supported across parliamentary
terms by the Construction 2025 policy and the Construction Strategy 2016–2020. The
goal is to save money by digitally transforming the construction industry and add value
by contributing to 2025 UK construction industry objectives: 50% reduction in gas
emissions, 50% increase in exports, 33% reduction in cost and 50% reduction in sched-
ule.
60 A. Matthews and B. Ta

The UK Government’s Cabinet Office are responsible for coordinating the Govern-
ment’s drive to the development of standards enabling all members of the supply chain
to work collaboratively through Building Information Modelling (BIM).
Following the publish of government’s report on Data for public good at the end of
2017, Centre for Digital Built Britain (CDBB) was assigned with ensuring The UK can
exploit new technologies and digital connections to transform the construction industry
and bring practical and social benefits. CDBB is responsible for managing, coordinat-
ing national programs on promoting digital transformation of UK construction industry.
CDBB now is organizing three main programs:
∞ Policy: Working with the Government to develop proper policy frameworks based
on specific pilot results, encouraging knowledge sharing as well as support through
bidding, procuring in a consistent way. CDBB connects directly with ministries, re-
search centers and clients to ensure that reasonable proposals could be made and
applied.
∞ Research: Creating an active research network; actively develop new technologies,
new ideas and promote multidisciplinary cooperation with many universities. Engi-
neers, architects, scientists in the field of information technology, mathematics, eco-
nomics, psychology, sociology and anthropology, ... other related fields are gathered
to shape the future of the construction industry. Cambridge university and other
training and research institutions in the network are conducting researches to enable
digital transformation the construction industry.
∞ Encouraging knowledge sharing: CDBB works with agencies and units in the indus-
try to test new ideas, put the best ideas into practice; practical sharing, support for
business transformation; build common standards, certificates, education and train-
ing.

2.2 Communicate vision and foster communities

The alliance was originally pulled together to form two main roles (Fig. 3).

Fig. 3. BIM Alliance Forum.

Firstly, it was going to be the group who supported BIM Level 2 becoming business as
usual in the UK after the mandate when the focus elsewhere was moving onto BIM
Applying a national BIM model to Vietnam's National Implementation... 61

Level 3. Secondly it was going to provide a home for the already established and run-
ning BIM Regions and BIM groups which up until that point had all been independent
and self-sufficient.
The alliance made up of lots of members (Fig. 4) as well as “patron” organizations who
sponsor the alliance, this is the groups primary source of income.

Fig. 5. Regional BIM groups and other specific BIM groups.

The alliance uses mass communication tools, such as online media, events, web, social
media website, email newsletter, Twitter, LinkedIn, etc. to reach audiences. But the
problem with all of these have been that they get traffic from people who are already
aware of BIM and not those they need to be targeting who don’t know about BIM.

2.3 Build a common, collaborative framework

The UK develops a set of standards to support BIM implementation as follows:
∞ Capital Delivery: Publicly Available Specification (PAS) 1192-2:2013 is the speci-
fication for information management for the capital/delivery phase of construction
projects using a BIM process including the requirements for Employer’s Information
and the BIM Execution Plan.
∞ Operational Delivery: PAS 1192-3:2014 is to act as a partner to PAS 1192:2 and
support the Level 2 BIM objective by setting out a framework for information man-
agement for the whole life cycle of asset management.
∞ COBie / FM: BS 1192-4:2014 defines expectations for the exchange of non-graph-
ical project information fulfilling employer’s information exchange requirements us-
ing COBie – Code of practice. COBie enables the exchange of structured infor-
mation for the commissioning, operation, and maintenance of a project in a neutral
spreadsheet format.
∞ Cyber Security: PAS 1192-5:2015 is designed to provide a framework to assist asset
owners in understanding the key vulnerability issues and the nature of controls re-
quired to enable the trustworthiness and security of digitally built assets.
∞ BIM Protocol: The Construction Industry Council (CIC) BIM Protocol is a legal
addendum to design and construction contracts that allows parties to share data
within a contract when working to BIM Level 2.
62 A. Matthews and B. Ta

∞ Government Soft Landings: The Government Soft Landings (GSL) is a form of soft
handover which incorporates the operational team into design decisions. The design
team are also on hand after completion to assist with the handover and learn lessons
from the real everyday use of the asset.
∞ Digital Plan of Works (Levels of Detail) and Validation Tool: The Digital Plan of
Work (DPoW) outlines the information requirements for any constructed asset, new
or existing.
∞ Classification: The UK government commissioned a unified classification system
(Uniclass) for team members. A single classification system enables electronic pro-
ject data to be indexed and structured to be accessible and searchable.

2.4 Grow Industry Capacity

One of the most important decisions taken by the UK BIM programme was the recog-
nition that project requirements must change first if industry as a whole were to change.
A standards framework development alone was insufficient to drive a whole sector
transformation. Communications actions created crucial engagement with industry,
however, use of public procurement was the strategic force at the centre of the UK BIM
programme.
In 2011, the UK BIM programme communicated a target that all centrally funded public
construction projects would require the use of BIM by 2016. During 2012 to 2015 the
pipeline of publicly funded projects adopting BIM Level 2 requirements increased sig-
nificantly from approx. £100m to over £9bn.
This progressive increase in the number of public projects requiring BIM Level 2 was
essential to steadily build capability of the supply chain and for the public client. It
allowed time for skills development and learning to occur within the UK BIM Task
Group, the public clients and within industry.
Currently, the UK has 6 key Ministries must apply BIM in their projects, the others
encourage BIM application for large projects. This working group is supported by the
Prime Minister's Office.
Related to training, the UK BIM Programme developed a Learning Outcomes Frame-
work (LOF) for BIM. The LOF provides consistent information on BIM Level 2 to
institutions, academia, training providers and private educators developing and deliv-
ering training courses to professionals in the sector. This aims to build capacity devel-
opment in the industry.

3 Vietnam BIM implementation since 2014

Since 2014, Ministry of Construction of Vietnam (MOC) has assigned to Institute of

Construction Economics to conduct a research project on “Building Information Mod-
eling (BIM) roadmap for improving efficiency of design, construction and facility man-
agement activities in Vietnam”. In the beginning, some main contractors such as Hoa
Binh, Coteccons, Cofico take the lead as they saw the benefits that BIM application
could bring to the construction. Several design consultants such as VNCC, CDC…. and
Applying a national BIM model to Vietnam's National Implementation... 63

some BIM consultants specialized in modeling use BIM tools for their projects. BIM-
Related topics have been mentioned in many seminars organized by state management
agencies (Ministry of Construction, Ministry of Transport, Ho Chi Minh City Depart-
ment of Transportation ...), research institutions, universities, and consulting organiza-
tions [7]. Then, the government begins to take part in facilitating BIM implementation
by introducing several BIM-related contents in construction law No. 50/QH13/2014
and other regulations.
However, several gaps and challenges are identified for Vietnam Construction Industry
such as: lack of skilled BIM workforce, organizational structure and working style do
not facilitate collaboration, low labor productivity… [7]
This research project was finished and submitted to the Ministry in 2015 which led to
MOC’s proposition to develop a national BIM program for the Government. The na-
tional BIM program was drafted in 2016 and was approved by the Prime Minister in
Decision no. 2500/QĐ-TTg dated December 22nd, 2016. Since 2017, the national BIM
program has been carried out by Vietnam BIM Steering Committee.

3.1 Foundation of Public leadership

Although the Construction Law has some regulations and principles for using BIM in
construction investment and project management, it is still missing specific guidelines,
so investors and project management units are quite embarrassing in studying and ap-
plying BIM to their projects.
With the task assigned by the Prime Minister, the Ministry of Construction has chaired
and coordinated with related ministries and agencies to implement the scheme and is-
sued decisions to establish the Steering Committee for the implementation of the
Scheme, including representatives of Ministry of Construction, Ministry of Planning
and Investment, Ministry of Transport, Ministry of Industry and Trade, Ministry of Ag-
riculture and Rural Development; Promulgated the Master Plan for implementation
from 2017 to the end of 2020 (Decision No. 203/QD-BXD, No. 204/QD-BXD dated
March 21st, 2017 of the Minister of Construction). The BIM Steering Committee -
Ministry of Construction has established BIM Technical Committee to advise the Steer-
ing Committee to implement the program, including experts with experience in training
and BIM implementation in the construction sector. Every year, the BIM Steering Com-
mittee, based on the Master Plan issued by the Minister of Construction, issued detailed
operation plans.

3.2 Communicate vision and foster communities

Understanding BIM is still a lot of confusion due to lack of information. Overall, there
is still much confusion between BIM and software. In order to raise awareness on BIM,
Steering Committee use website as official mean to communicate (www.bim.gov.vn).
In addition, many workshops were organized by BIM Steering Committee to dissemi-
nate the national BIM program, BIM benefits and experiences on BIM implementation
in 2017, 2018 and 2019. Many other stakeholders such as universities, associations,
64 A. Matthews and B. Ta

clients, project management boards, communities also organize conferences in many

cities across the country to raise BIM awareness.
Also, coordination with media agencies to share information and experience of BIM
application is also focused on implementation. Activities promoting BIM in confer-
ences and seminars are promoted through media. In particular, Vietnam Television 1
has released 02 documentaries about BIM with the participation of many BIM practi-
tioners, clients and government officials.
The connection is also focused on deploying in the region and some countries in the
world. It can be seen that the work of raising awareness and encouraging research to
apply BIM has been basically implemented according to the objectives and plans. Many
construction industry owners have learned more about BIM, the benefits and difficul-
ties in implementing and applying BIM.

3.3 Build a common, collaborative framework.

Policy-related issues are a major challenge for BIM application: contracting process,
dependence on technology, ownership and responsibility for models, insurance issues.
In order to solve those problems, the Ministry of Construction has announced the Guid-
ance on temporary BIM application for pilot projects in Decision No. 1057/QD-BXD
dated 11/10/2017 of the Minister of Construction. In addition, detailed instructions on
BIM application for urban technical infrastructure (urban transport, water supply and
drainage) as well as civil engineering (architecture, structure, electromechanical sub-
jects), being urgently built to complete the report to the Ministry of Construction.
Basically, the temporary BIM application guidelines have been applied for pilot pro-
jects. However, according to the survey, some contents need more detailed guidance
(Process of modeling for each discipline; Coordination between disciplines in the im-
plementation process).

3.4 Grow Industry Capacity

The BIM training framework for pilot projects has been published in Decision No.
1056/QD-BXD dated October 11th, 2017 by the Minister of Construction. Detailed
BIM training documents are being developed for providing extensive training to rele-
vant individuals and organizations.
Many training activities have been implemented for pilot projects such as Viettel Mili-
tary Industry and Telecoms Group at Viettel Academy from March 28-29th, 2018; BIM
pilot projects’ clients in April, 2018 and July 2018.
In addition, BIM tools training courses have been implemented in some universities
such as Hanoi Architecture, Transport, Transport Technology, Electricity Power uni-
versities to promote and improve training materials. A BIM Academic Forum has been
established with following objectives:

∞ To create a sustainable network connecting higher education institutions in Vietnam;

∞ To create an effective and efficient communication channel in order to promote
knowledge exchange between institutions, institutions and the AEC industry;
Applying a national BIM model to Vietnam's National Implementation... 65

∞ To propose and implement BIM research and delivery projects, to deliver training
courses for developing BIM human resources for Vietnam construction industry;
∞ To promote social criticism and support in policies development for the Government
in terms of BIM and digital transformation in construction industry.
As BIM application requires training and the use of software to create models with the
cost for purchasing software and training to use software. Employer Information Re-
quirements and BIM fee have been developed with the help of BIM Technical Com-
mittee for pilot projects in order to select BIM-related suppliers.
Currently, some projects still have difficulties in establishing a common data environ-
ment; coordinating and managing BIM coordination process between BIM consultants
and design consultants, contractors; assuring BIM model quality control; BIM applica-
tion in reviewing design, resolving clashes; data organization or BIM for facility man-
agement.
Preliminary assessments of BIM benefit for a number of pilot projects has been carried
out and followed closely. Detailed evaluation of BIM benefit will be carried out next
year for the review and assessment of the implementation of BIM application in every
quarter.

4 Recommendations for Vietnam Construction Industry

With the UK bringing expertise and consultant expertise to support, Vietnam can ben-
efit by learning more quickly, and by adopting international standards in order to
achieve productivity goals and build more infrastructure. Also, it will help accelerate
Vietnam’s national introduction of BIM, developed a more impactful programme and
increase the benefits to Vietnam from international standards and best practice aligned
with national strategy and change management models. BIM has organically developed
in private industry as geometry and design improvements – the UK-Vietnam collabo-
ration shows how to implement BIM as a collaborative data sharing process across the
project lifecycle. The UK has learned how to adapt technical and legal framework from
its work in the UK and Europe to Vietnam. This has helped improve the UK’s govern-
ment to government collaboration.
This collaboration enables better quality infrastructure delivery with more delivered,
higher skilled workforce. More importantly, it will help the UK working together across
borders with other countries and exporting to different markets using the same concepts
and principles for digital construction.
Vietnam BIM program according to Decision 2500/QD-TTg dated December 22nd,
2016 has been able to promote BIM adoption in construction industry. However, in
order to have more impacts and enable the use of new technologies, BIM-related con-
tents need to be included in legal documents, namely construction law or other relevant
regulations concerning design approval, appraisal, quantity take-off, cost estimation,
document submission and facility maintenance.
Digital technologies can be applied not only in the construction investment stage but
also facility management process. Also, construction system (especially the infrastruc-
ture system) need to be connected in order to bring more benefits. Therefore, in order
66 A. Matthews and B. Ta

to promote digital transformation in construction industry (including BIM), the govern-

ment needs to set clearly objectives and play a leading role for the whole industry (in-
cluding the private sector) by creating the environment for BIM adoption such as de-
veloping standards, guidelines, contract forms... An implementation team need to be
created to plan, connect, monitor and support in collaboration with other agencies, min-
istries, sectors.
In addition, a vision needs to be set for the construction industry by issuing a Strategy
or Roadmap for digital transformation with objectives, priority orders and responsibil-
ities of different government agencies. Ministry of Construction needs to play a leading
role in promoting digital transformation in the construction sector. It needs to issue
regulations for e-submission, data exchange standards, data storage structure…, de-
velop digital database for construction projects that local governments and other min-
istries can use for facility management and maintenance.
Since some leading countries have gone down this path of promoting digital transfor-
mation in construction industry, international cooperation to promote experience shar-
ing, conduct researches and referring to the roadmap and plan in other countries will be
very important.

5 References
1. EU BIM Task Group: Handbook for the introduction of Building Information Modelling by
the European Public Sector – Strategic action for construction sector performance: driving
value, innovation and growth (2017).
2. NIBS (2008). United States National Building Information Modeling Standard, Version 1,
Part 1: Overview, Principles, and Methodologies. http://nbimsdoc.opengeospatial.org/
(2009).
3. Sacks, R., Eastman, C., Lee, G., & Teicholz, P.: BIM handbook: a guide to building infor-
mation modeling for owners, designers, engineers, contractors, and facility managers 3rd
edn. John Wiley & Sons (2018).
4. AGC.: An Introduction to Building Information Modeling (BIM) 3rd edn (2017)
5. NBC: Roadmap to Digital Transition for Ireland’s Construction Industry 2018-2021 (2017)
6. BIM Industry Working Group: A report for the government construction client group build-
ing information modelling (BIM) working party strategy paper. Communications. London,
UK (2011).
7. B. TA, et al.: Research on establishing Building Information Modeling (BIM) implementa-
tion roadmap to improve design, construction and facility management efficiency in Vi-
etnam, Final report (2015).
 Part II
Advanced Modeling and Characterization
of Structures
 Advanced Analysis Software for Steel Frames

Phu-Cuong Nguyen*, My Ngoc-Tra Lam, Duc-Duy Pham,

Trong-Nghia Nguyen, Phong Thanh Nguyen

Faculty of Civil Engineering, Ho Chi Minh City Open University,

Ho Chi Minh City, Vietnam
*
cuong.pn@ou.edu.vn ; henycuong@gmail.com

Abstract. The manuscript shows an advanced analysis software of steel frames

under static loadings. Input and output processes are programed using C++ and
FORTRAN. A visual graphic interface for the advanced analysis software is
coded for modeling easily and showing the result performance of analyzing
structures. The proposed software uses stability functions (SF) to predict the
second-order effects on the structures. The plastic hinge approach (PHA) and
the fiber plasticity approach (FPA) are integrated in the advanced analysis
software. The Generalized Displacement Control Algorithm (GDCA) is utilized
for capturing the behavior of frames. A steel 3D frame is analyzed aim to vali-
date the accuracy and time efficiency of developed software.

Keywords: Steel framed structures; advanced analysis; geometric nonlinearity;

nonlinear analysis algorithm; plasticity.

1 Introduction

For the practical engineering design, effective length factors are used for taking into
accounting the interaction between whole structure and elements in it. Using effective
length factor gives a not bad design for steel frames. But, this method has its disad-
vantages. It does not give a correct indication of failure phenomena, because this
method does not take into accounting the interaction between stability and strength of
elements and whole structure in a direct way. The next perhaps the worst disad-
vantage might not logical in design procedure: elastic structural analysis is utilized to
find out internal forces at each element of whole structure, while inelastic structural
analysis is utilized to predict the ultimate strength of each isolated member. As a re-
sult, there is no explicit guarantee that elements of the structure will sustain design
loads due to the configuration of the structure affecting on the steel frame.
Nowadays, the local and global stability of the structure might be considered care-
fully aim to determine the ultimate strength of structures. Such design method is
drawn in Fig. 1 like a direct design and analysis approach. “Advanced Analysis” is
also called to be direct design method, Nguyen et al. [1-10] developed several ad-
vanced analysis methods for steel frames. In this method, we do not use effective
length factors because element strength calculated by equations in design standard is
not necessary. Therefore, it may feasible to utilize advanced analysis tools for daily

© Springer Nature Singapore Pte Ltd. 2020 69

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_7
70 P.-C. Nguyen et al.

engineering design. Fig. 1 compares the conventional indirect approach using ASD,
PD, or LRFD and the direct approach using advanced analysis.

Fig. 1. Scheme for structural design.

An Advanced Analysis Software (AAS) using two beam-column approaches is

coded for estimating the second-order inelastic behavior of 3-D frames. Second-order
effects (P-delta effects, large and small) are predicted by SF to optimize modeling and
time efficiency, whilst inelasticity of steel is considered by employing the PHA or the
FPA. The proposed software is developed based on Nguyen et al.‘s work [1-10]. The
GDCA [11] is adopted to solve equilibrium equations for static problems. A visual
graphic tool of the proposing program is coded for easily modeling the structure and
performing results.

2 Advanced analysis software

The input and output procedures of the software are developed using the program-
ming language of C++ and FORTRAN. Graphic user interface is friendly made to
simplify step by step working. Analysis procedure using the software is illustrated
through Fig. 2.
The element library of the coding software includes two nonlinear elements of
beam-column types such as: the PHA element and the FPA element. The detail for-
mulation derivation of the two elements can be found in [1, 3, 4, 6, 8-10].
Advanced Analysis Software for Steel Frames 71

Fig. 2. Analysis procedures of the software.

2.1 PHA element

To monitor the effects of the interaction due to moments and axial force affecting on
an element, the SFs are employed aim to optimize for modeling and time efficiency.
The effects of P-delta are correctly traced by using only one element for a beam or a
column [1, 3, 10].
The inelasticity of material is considered by utilizing of plastic hinges due to ge-
ometry imperfections, residual stresses, axial force and bending. The gradual yielding
of an element occurred by residual stresses is taken into accounting by employing the
tangent elastic modulus concept, whilst plasticity occurred by bending is shown
through using a mathematical function [1, 3, 10]. For more information, the readers
are recommended seeing the papers in the Reference.

2.2 FPA element

Monitoring gradual yielding along the length of elements, a fiber plastic approach is
used from Nguyen’s studies [4, 6]. FPA elements are meshed into a discrete number
of watched sections represented through monitoring points on the length of elements.
Watched sections are meshed into i small separate fibers, and these separate fibers are
represented by their area, geometric configuration, and their location. For hot-rolled
steel sections, residual stresses are assigned directly to fibers as initial stresses. For
more information, the readers are recommended seeing the papers in the Reference.
72 P.-C. Nguyen et al.

3 Nonlinear analysis algorithm

In several algorithms, the GDCA invented by Yang and Shieh [11] like the best effec-
tive algorithm to solve equilibrium equations for nonlinear problems including several
limit points since its numerical efficiency and stability. The GDCA is adopted in this
program.

Fig. 3. A steel 3D frame.

4 Numerical example
A steel 3D frame shown in Fig. 3 was investigated by Professor Orbison [12]. The
elastic modulus, shear modulus, and yield stress of steel was 206,850 MPa, 79,293
MPa, and 250 (MPa), respectively. Wind forces were acted through nodes on the
column tops (53.376 kN) in the Y-direction. Uniform loadings (9.6 kN/m2) on the
floor were converted into equivalent forces at the column tops. For the PHA, one
element for one beam-column was used to model the structure for nonlinear inelastic
analysis. For the FPA, beam-column elements with seven integration points on the
length were employed to simulate the structure. Cross sections of all elements were
divided to become 90 fibers (34 at top and bottom flange, 20 at the web).
Fig. 4 compares displacement response curves under static loadings at Y-direction
node A of the structure generated by the proposed software using both the refined
Advanced Analysis Software for Steel Frames 73

plastic hinge approach and the distributed plasticity approach, and Chiorean’s result
[13] using the distributed plasticity approach. The ultimate load generated by the pro-
posed software using the plastic hinge element is higher than the result using the dis-
tributed plasticity element, however, it is approximate to Chiorean’s result. As a re-
sult, it can be concluded that the PHA correctly estimates the ultimate strength of the
structure. Fig. 4 shows positions of plastic hinges of the frame at the ultimate loads
using the plastic hinge approach.

Fig. 4. Displacement curves at node A and plastic hinges at the ultimate load.

5 Conclusion
A second-order inelastic software for steel frames are being programed carefully. The
software is using SF for taking into accounting the P-delta effects, PHA and FHA for
taking into accounting the effects of material plasticity for 3D frames to optimize
modeling and time efficiency. As shown in the above example, the coding software
performs the accuracy and the time efficiency in tracing the second-order inelastic
behavior of steel framed 3D structures. As a result, the coding advanced analysis
software performs to be reliable and valuable for civil engineers at Vietnam and all
over the world in daily direct design work.

Acknowledgment
The authors gratefully acknowledge the financial support provided by the Scientific
Research Fund of the Ministry of Education and Training (MOET), Vietnam.
(No. B2019–MBS–01). The authors would also like to thank colleagues at Ho Chi
Minh City Open University for supporting this project.
74 P.-C. Nguyen et al.

References
1. Nguyen, P.C., Kim, S.E.: Nonlinear elastic dynamic analysis of space steel frames with
semi-rigid connections. Journal of Constructional Steel Research 84, 72-81 (2013).
2. Nguyen, P.C., Doan, N.T.N., Kim, S.E.; Nonlinear inelastic response history analysis of
steel frame structures using plastic-zone method. Thin-Walled Structures 85, 220-33
(2014).
3. Nguyen, P.C., Kim, S.E.: Nonlinear inelastic time-history analysis of three-dimensional
semi-rigid steel frames. Journal of Constructional Steel Research 101, 192–206 (2014).
4. Nguyen, P.C., Kim, S.E.: An advanced analysis method for three-dimensional steel frames
with semi-rigid connections. Finite Elements in Analysis and Design 80, 23-32 (2014).
5. Nguyen, P.C., Kim, S.E.: Distributed plasticity approach for time-history analysis of steel
frames including nonlinear connections. Journal of Constructional Steel Research 100, 36-
49 (2014).
6. Nguyen, P.C., Kim, S.E.: Second-order spread-of-plasticity approach for nonlinear time-
history analysis of space semi-rigid steel frames. Finite Elements in Analysis and Design
105, 1–15 (2015).
7. Nguyen, P.C., Kim, S.E.: Advanced analysis for planar steel frames with semi-rigid con-
nections using plastic-zone method. Steel and Composite Structures 21, 1121-1144 (2016).
8. Nguyen, P.C., Kim, S.E.: Investigating effects of various base restraints on the nonlinear
inelastic static and seismic responses of steel frames. International Journal of Non-Linear
Mechanics 89, 151-167 (2017).
9. Truong, V.H., Nguyen, P.C., Kim, S.E.: An efficient method for optimizing space steel
frames with semi-rigid joints using practical advanced analysis and the micro-genetic algo-
rithm. Journal of Constructional Steel Research 128, 416-427 (2017).
10. Nguyen, P.C., Kim, S.E.: A new improved fiber plastic hinge method accounting for lat-
eral-torsional buckling of 3D steel frames. Thin-Walled Structures 127, 666-75 (2018).
11. Yang, Y.B., Shiel, M.S.: Solution method for nonlinear problems with multiple critical
points. AIAA Journal 28, 2110–2116 (1990).
12. Orbison, J.G., McGuire, W., Abel, J.F.: Yield surface applications in nonlinear steel frame
analysis. Computer Methods in Applied Mechanics and Engineering 33, 557–573 (1982).
13. Chiorean, C.G.: A computer method for nonlinear inelastic analysis of 3D semi-rigid steel
frameworks. Engineering Structures 31, 3016–3033 (2009).
 Finite Element Modelling for
Axially Loaded Concrete-Filled Steel Circular Tubes

Duc-Duy Pham and Phu-Cuong Nguyen*

Faculty of Civil Engineering, Ho Chi Minh City Open University,

Ho Chi Minh City, Vietnam
*
cuong.pn@ou.edu.vn ; henycuong@gmail.com

Abstract. This study presents a Finite Element Analysis modelling (FEA) for
Concrete-Filled Steel Circular Tubes (CFSCT) subjected to axial compressive
load using the commercial FEA software ABAQUS. A new stress-strain
relationship for concrete is established using the experimental data collecting
from literature review. Concrete-Damaged Plasticity Model (CDPM) is
employed in this study. A new softening regime is proposed for the behavior of
confined concrete, and this study also introduces new parameters for
constitutive model of concrete when we input into the ABAQUS sofware. The
proposed analysis result is more accurate than the one of previous studies in the
case of high-strength concrete or thin-walled steel tubes.

Keywords: Concrete-filled steel circular tube; Axial compressive load; Finite

element analysis; ABAQUS; Stress-strain relationship; Confinement effect.

1 Introduction

CFST columns are used widely because of its strength and stability. Several research-
ers studied the experimental behavior of CFST columns aims to develop the new for-
mulation of CFST columns for design specifications.
In recent years, finite element (FE) models were developed by Han et al. [1] and
Hu et al. [2] used widely to simulate CFST columns by using the ABAQUS software.
However, Hu¶V model proposes confinement of concrete using the stress-strain rela-
tionship and ignoring the interaction between concrete and steel, ignoring the dilation
angle. With the ratio of D/t > 150 or D/t < 27.1, this model can¶W predict the stress ±
strain relationship for CFST columns. In other hand, Han¶V model using concrete-
damaged plasticity model in ABAQUS and using default parameters such as the ratio
of the second stress invariant on the tensile meridian to that on the compressive me-
ridian (Kc), the dilation angle (\), the flow potential eccentricity (e), the ratio of the
compressive strength under biaxial loading to the uniaxial compressive strength
(fbo/I¶c). The main purpose of the paper is to develop a new stress-strain curve for the
FE model to predict accurately the static behavior of CFSCT columns.

© Springer Nature Singapore Pte Ltd. 2020 75

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_8
76 D.-D. Pham and P.-C. Nguyen

2 Finite Element Modelling

The steel box of CFST columns are modelled using the 4-node integration shell ele-
ments (S4R), which has six degrees of freedom per node. The concrete core of CFST
was modelled using the 8-node solid elements with three degrees of freedom per node
(C3D8R).
The interaction of the steel tube and concrete core uses surface-to-surface contact
in ABAQUS. A contact surface pair comprised of the inner surface of the steel tube
and the outer surface of the concrete core can be defined. Normal direction is used
³+DUGFRQWDFW´E\WKH specified interface, which allows the separation of the interface
in tension and no penetration of that in compression. Tangential direction is used
friction formulation penalty with the friction coefficient by Mohr Coulomb model. In
this study, a friction coefficient of the steel tube and concrete was taken equal to 0.6
as Han et al. [1].

2.1.1 Constitutive modelling for steel

The stress ± strain relationship model V - H proposed by Tao et al. [3] is used in this
study as shown in Fig. 1a.

Fig. 1. a. Stress ± strain curve proposed by Tao et al. for steel, b. Stress ± strain cure pro-
posed by authors for confined concrete.

2.1.2 Concrete damaged plasticity model

The concrete-damaged plasticity model is used to describe confined concrete for

CFST columns and which available in ABAQUS. In this model, material parameters
can be determined through formulas. Include the ratio of the second stress invariant
on the tensile meridian to that on the compressive meridian (Kc), dilation angle (\),
strain hardening ± softening rule, the modulus of elasticity (Ec), ratio of the compres-
sive strength under biaxial loading to uniaxial compressive strength (fboI¶c), and frac-
ture energy (GF). Other parameters include flow potential eccentricity (e) and viscos-
ity parameter used default values 0.1 and 0 in ABAQUS respectively.

2.1.3 Determining Kc - \ - fboI¶c - GF

The ratio of the second stress invariant on the tensile meridian to that on the compres-
sive meridian (Kc) is one of parameters for determining the yield surface of concrete.
Finite Element Modelling for Axially Loaded … 77

According to Yu et al. [4] indicate that the range of Kc varies from 0.5 to 1 and the
following equation can be calculated Kc:
5.5 u fb 0
Kc (1)
3 u f c'  5 u fb 0
The dilation angle (\) is one of parameters required for ABAQUS to define the
yielding potential. In ABAQUS, the dilation angle of \ ranges from 00 to 560. The
following equation is proposed by Tao et al. [5] based on regression analysis to de-
termine \ for circular CFST columns:
­56.3 u 1  [ c  ȟ c d 
°
\ ® 7.4 (2)
°
¯6.672 u e
4.64 [c
ȟ c ! 
where the confinement factor [c is expressed as
As u f y
[c (3)
Ac u f c'
Based on the test data, Papanikolaou and Kappos [6] proposed the following equa-
tion to predict the ratio of fboI¶c:
fb 0 1.5
' 0.075
(4)
fc fc'  
The effects of ligament length, rate of loading and concrete composition on the
specific fracture energy Gf and the strain-softening diagram are investigated. The
following equation is used to define Gf for tensile concrete [7]
0.7
§ f'·
Gf  2
0.0469 u dmax  0.5 u dmax  26 u ¨ c ¸ u 103  N/mm (5)
© 10 ¹
2.1.4 Constitutive modelling for concrete

A new proposed model of the stress-strain relationship of concrete consists of three-

stages is present as shown in Fig. 1b. Three-stages included ascending branch (from
Point O to Point A), plateau branch (from Point A-B), descending branch (from Point
B to Point C). Ascending branch represents the strain hardening rule. In this stage is
very little interaction between the steel tube and concrete. The model proposed by
Samani and Attart [8] is used to describe the ascending curve OA.
2
V Au X  B u X
0  H d H c0 (6)
1   A - 2 u X   B  1 u X
' 2
fc
2

where A
Ec u H c 0
; B
 A  1  1; X
H
'
fc 0.55 H c0
The strain at the peak stress of the unconfined concrete is calculated in Eq. (7) pro-
posed by Tasdemir et al. [9] based on regression analysis of 228 uniaxial compression
test specimens with uniaxial compressive strength values ranging from 6 to 105 MPa.
78 D.-D. Pham and P.-C. Nguyen

§ 0.067 u f ' 2  29.9 u f '  1053 · u106

H c0¨
©
 
c c ¸
¹
(7)

The strain at the point B (Hcc) for the concrete model is determined by the following
equation proposed by Xiao et al. [10], which are found to provide accurate prediction
for both normal and high-strength concrete:
1.06
H cc §f ·
1  17.4 u ¨ B' ¸ (8)
H c0 © fc ¹
where fB is the confining stress provided to the concrete at the point B.
In the elastic stage, there is no confining stress but just before and after the yielding
of the steel, the confining stress increases very fast. Once the ultimate strength
reaches, fr keeps stably or increases very slowly depending on the value [c. On the
basis of regression analysis, Eq (9) is proposed to determine fB for circular columns:
0.03 u f y
fB D
(9)
0.02u
e t

After the OA stage, the peak strength value remains constant range Hc0 to Hcc. This
stage represents increment of peak strain and strength of concrete increases due to
confinement by the interaction between the steel tube and concrete.
An exponential function proposed by Binici [11] is used to represent the descend-
ing branch (BC)
ª § H  H cc · E º
V  
f r  f c'  f r u exp «  ¨ ¸ » H t H cc (10)
¬« © D ¹ ¼»
Parameters in the function include the residual stress (fr), the shape of the softening
branch (Į and ȕ). These three parameters are proposed by Tao et al. [5].

3 Verification of the proposed FE Model

The accuracy of the current FE models is verified through the test data summarized in
Table 1. The predicted ultimate strength (Nuc) from the proposed FE model, Han et al.
[1] and Hu et al. [2] models are compared with the measured ultimate strength (Nue)
from the experimental tests as shown in Fig. 2. Table 2 shows both the mean value
(Mean) and Standard Deviation (STD) of the ratio of Nue/Nuc of more than 66 speci-
mens with different cross-sections.
The mean values of Nue/Nuc are 1.011, 0.975 and 0.980 for the proposed model,
+DQHWDO¶VPRGHO+XHWDO¶VPRGHOUHVSHFWLYHO\,WFDQEHVHHQWKHSURSRVHGPRGHO
predicts the ultimate strength of CFSCT columns more exactly than the results of Han
and Hu.
Based on the comparison shown in Fig. 3, the proposed model predicts accurately
the static behaviour of CFSCT columns in case of circular columns with normal and
high-strength concrete and thin-walled steel tubes than the models of Han et al. and
Hu et al.
Finite Element Modelling for Axially Loaded … 79

Table 1. Test data of circular CFST columns.

D t fy fc ¶
Source N D/t L/D
(mm) (mm) (MPa) (Mpa)
140 2.0 21.54 3.0 265.8 27.15
Hu et al. [2] 06
300 6.5 150 4.3 341.7 31.15
Yu et al. [12] 1 100 1.9 52.63 3.0 404 101.91

Table 2. Comparing ultimate strength of the experimental tests and FE models.

FE model STD Mean

Proposed 0.055 1.011
Han et al. [1] 0.090 0.975
Hu et al. [2] 0.129 0.980

1.3

1.1

0.9

0.7
Proposed model Han et al. Hu et al.

0.5
0.00 0.40 0.80 1.20 1.60 2.00 2.40 2.80 3.20 3.60 4.00

Fig. 2. Comparing ultimate strength of the experimental tests and FE models.

3500 D = 300 mm, t = 3 mm, L = 900 mm 1200 D = 100 mm, t = 2.0 mm,
fy = 232 MPa, I¶c = 27.2 MPa, [ c = 0.35 L = 300 mm
3000 1000 fy = 404 MPa, I¶c = 101.9 MPa,
[ c = 0.32
Axial load N - (kN)

2500
Axial load N - (kN)

800
2000
Test-CU-100 600
1500 Concrete Core
TEST-C30-1
Proposed model
Steel Tube 400 Proposed model Concrete Core
1000
Han et al. Han et al. Steel Tube
500 200
Hsuan et al. Hsuan et al.
0 0
0 10000 20000 30000 0 10000 20000 30000
Axial strain H - (PH) Axial strain H - (PH)
Fig. 3. The axial load ± strain curves of CU-100 and C30-1 specimens.
80 D.-D. Pham and P.-C. Nguyen

4 Conclusion
Authors proposed a new finite element model employing ABAQUS for predicting
accurately the static behavior of concrete-filled steel circular tube column (CFSCT)
subjected axial compression. Some conclusions can be drawn as follows:
Comparing with +DQHWDO¶VDQGHu et al.¶V FE models, the proposed model pre-
dicts more exactly the ultimate strength of CFSCT columns.
The proposed model can capture accurately the static behavior of CFSCT columns
more than previous studies.
The proposed model can apply for CFSCT columns using both normal and high-
strength concrete and/or thin-walled steel tubes.

Acknowledgement
The authors gratefully acknowledge the financial support provided by the Scientific
Research Fund of Ho Chi Minh City Open University, Vietnam. (No. E2019.02.2).

References
1. Han, L.H., Yao, G.H., Tao, Z.: Performance of concrete-filled thin-walled steel tubes un-
der pure torsion. Thin-Walled Structures 45, 24-36 (2007).
2. Hu, H.T., Huang, C.S, Wu M.H, Wu, Y.M.: Nonlinear analysis of axially loaded concrete-
filled tube columns with confinement effect. Journal of Structural Engineering 129, 1322-
1329 (2003).
3. Tao, Z., Wang, X.Q., Uy, B.: Stress-strain curves of structural and reinforcing steels after
exposure to elevated temperatures. Journal of Materials in Civil Engineering 25, 1306-
1316 (2013).
4. Yu, T., Teng, J.G., Wong, Y.L., Dong, S.L.: Finite element modeling of confined concrete-
i: Drucker±prager type plasticity model. Engineering Structures 23, 665-679 (2010).
5. Tao, Z., Wang, Z.B., Yu, Q.: Finite element modelling of concrete-filled steel stub col-
umns under axial compression. Journal of Constructional Steel Research 89, 121-131
(2013).
6. Papanikolaou, V.K, Kappos, A.: Confinement-sensitive plasticity constitutive model for
concrete in triaxial compression. International Journal of Solids and Structures 44, 7021-
7048 (2007).
7. Bazant, Z.P., Giraudon, E.B.: Statistical prediction of fracture parameters of concrete and
implications for choice of testing standard. Cement and Concrete Research 32, 529-556
(2001).
8. Samani, A.K, Attard, M.M.: A stress±strain model for uniaxial and confined concrete un-
der compression. Engineering Structures 41, 335-349 (2012).
9. Tasdemir, M.A, Tasdemir, C, Akyuz, S, Jefferson, A.D, Lydon, F.D, Barr, B.I.G.: Evalua-
tion of strains at peak stresses in concrete: A three-phase composite model approach. Ce-
ment and Concrete Research 20, 301-318 (1998).
10. Xiao, Q.G, Teng, G.J, Yu, T.: Behavior and modeling of confined high-strength concrete.
Journal of Composites for Construction 14, 249-259 (2010).
11. Binici, B.: An analytical model for stress±strain behavior of confined concrete. Engineer-
ing Structures 27, 1040-1051 (2005).
12. Yu, Q, Tao, Z, Wu, Y.X.: Experimental behavior of high performance concrete-filled steel
tubular columns. Thin-Walled Structures 46, 362-370 (2008).
Fracture analysis of crack propagation on the diaphragm
of steel bridge structures

Tran The Truyen1, Nguyen Duc Hieu1, Bui Thanh Tung1, Doan Bao Quoc1,
Nguyen Thuong Anh2
1
University of Transport and Communications (UTC), Viet Nam
2
Geodynamique and Structure, France
tranthetruyen@utc.edu.vn

Abstract: The crack failure occurring on the steel diaphragm of Vam Cong
Bridge during construction stage in Viet Nam has raised issues that it is
necessary to deeply understand the behavior of steel structures in the
construction process along with the boundary conditions of structures and
especially pre-defects within steel materials. Fracture theory should be applied
to analyze the brittle failures in this case which it is totally different from
conventional failure at plasticity state. The investigation analyzes the fracture
mechanism of a typical steel diaphragm structure that is similar to the
diaphragm member of main span in the Vam Cong Bridge. Initial cracks within
steel material being used for the diaphragm will be considered. The result
would contribute to clarify the reasons of the brittle failure in steel bridge
beams as well as crack patterns with different boundary conditions.

Keywords: fracture mechanics, crack propagation, brittle failure, steel bridge,

and diaphragm.

1 Introduction

Vam Cong Bridge has been designed with a total length of 2.97km in which the
main bridge is 870m long including 450m of main middle span with 73 girder
segments [1]. This is the longest steel bridge in the South of Viet Nam. The
superstructure of approach bridge is super-Tee girder type with 1099.7m long span in
the direction to Dong Thap province, and 999.7m span length in the direction to Can
Tho province. The widths of cross-section of main bridge and approach bridge are
24.5m including four lanes for truck with total width of 14m, two lanes for non-
motorized vehicle with total width of 6m, 2m safety lane, 1.5m separating strip and
1m total curbs width.
In November, 2017, the constructor and construction supervision consultant saw a
crack at the diaphragm on P29 pier of Vam Cong Bridge [1], [2] (Fig. 1). That was a
severe problem making many questions need examining.
Three crucial causes were proposed to explain the crack of CB6 diaphragm on P29
pier which could be the concentration of stress, the redundant stress, and the

© Springer Nature Singapore Pte Ltd. 2020 81

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_9
82 T. T. Truyen et al.

qualification of weld during welding process. When parts of girder were combined on
construction site, redundant stress were generated in different parts during welding
process. Cracks were caused due to combination of the three above mentioned factors;
in general, this assumption is also the same as the reasons found in the typical failures
of steel bridges in reality [3]. Some small defects of steel due to welding process
developed during construction causing severe crack in the CB6 diaphragm.

PYLON 1 PYLON 2
Km25+140.760 Km25+590.760
DONG THAP CAN THO

P28 P29

Fig. 1. Fracture in the CB6 Diaphragm on P29 pier of Vam Cong bridge during construction process.

This article aims to explain this crack based on fracture mechanics with hypothesis
on the appearance of initial crack considered a small defect in the CB6 diaphragm.
The propagation of crack in this steel girder was numerically modelled by ANSYS in
combination with RM Bridge software [4], [5] with nonlinear analysis.

2 General analysis of forces impact to CB6 diaphragm on P29

anchor pier

The RM bridge software was used to model calculation of forces impact to CB6
diaphragm on P29 anchor pier.
Calculation scheme of main bridge was illustrated by a girder suspended by cables
and supported by bearings on the pier-caps of pylons, meanwhile cables had two ends
with one connected to the main girder and the other being anchored to the pylons. The
ends of main spans are supported by backward piers. The diagram of main spans is
210m + 450m + 210m.
Fracture analysis of crack propagation … 83

Calculation parameters: Tensile forces in cables and material parameters of cables,

main spans and pylons in calculation model were taken from the design documents of
Vam Cong Bridge.
Calculation results: Based on analysis of calculation model, components of
calculation forces effecting on diaphragm were determined (Fig. 2).
F1 F1

CRACK

BEARING 1 BEARING 2

F2 F3 F3 F2
Fig. 2. Effect forces on the CB6 diaphragm.

Table 1. Effect forces on CB6 diaphragm were calculated from Model on RM Bridge.

Bearing
F1 F2 F3
Order Commination reaction
(kN) (kN) (kN)
(kN)
1 After completing construction 5001.16 2050.21 2050.21 -2909.14
2 With maximum positive moment 3280.35 1680.77 1680.77 -4533.81
3 With maximum axial force 6009.01 2332.73 2332.73 -2458.41
4 With minimum axial force 1324.80 957.435 957.43 -6888.68

3 Local behavior analysis of diaphragm

From force values calculated from general analysis model which CB6 diaphragm
on P29 pier anchor occurring during construction, local behavior analysis of the
diaphragm was conducted by ANSYS software with assumption of appearance of
initial micro-crack on the underside of diaphragm. The initial micro-crack could arise
during manufacturing and handling process.
The physio-mechanical characteristics of steel used to produce main girder
diaphragm are shown in Table 2.

Table 2. The physio-mechanical characteristics of steel.

Thermal
Specific Yield Ultimate Stress
Young Poisson¶V expansion
density strength strength intensity
modulus E ratio X coefficient
J fy fu KC
D
200000 0.0000117 7850 345 450 54
0.3
MPa 1/oC kg/m3 MPa MPa MPa m1/2
84 T. T. Truyen et al.

Forces loaded on the diaphragm were taken from general analysis model of Vam
Cong Bridge by RM bridge software.
The element mesh of diaphragm was solid element type. At the vicinity of crack,
density of element mesh was made increase to ensure accuracy of analysis result.

4 Results and discussions

4.1 Stress analysis result

The detailed descriptions of stress analysis results (Fig. 3 and Fig. 4).

Fig. 3. Stress in diaphragm without initial crack.

Fig. 4. Stress in diaphragm with initial crack.

Fracture analysis of crack propagation … 85

Comment: The case of diaphragm without initial crack: Maximum normal stresses
in accordance with different directions are distributed at different positions such as
the bottom of diaphragm in various sections and vertical web in bearing sections.
Von-Mises stress is VWÿmax = 324.14 MPa in the bottom of diaphragm at various
section of diaphragm.
The case of diaphragm with initial crack: Maximum normal stresses in accordance
with different directions concentrate at the initial crack tip. The stress state at the
crack tip is the three-dimensional state of stresses with Vxmax = 1657.7 MPa, Vymax =
1032.2 MPa, Vzmax = 722.72 MPa. Von ± Mises stress: VWÿmax = 903.5 MPa.
Maximum normal stress in diaphragm with initial crack is seven times as high as
that of in diaphragm without initial crack. Meanwhile, this proportion is three times in
comparison with Von-Mises stress.

4.2 Fracture mode

Analysis results of Stress Intensity Factor related to Mode I, Mode II and Mode III
in the case of diaphragm with initial crack are particularly presented (Fig. 5).
Calculation results of SIFS SIFS degree along crack
79
0.5 70
(MPa.m )
60
50
40
30
20
15
0 0.25 0.50 0.75 1.00 1.39
[m]

30
(MPa.m0.5) 25
20
15
10
5
0
-3.4
0 0.25 0.50 0.75 1.00 1.39
[m]

335
0.5
(MPa.m ) 200
100
0
-100
-200
-330
0 0.25 0.5 0.75 1.00 1.39
[m]

Fig. 5. The results of Stress Intensity Factor.

86 T. T. Truyen et al.

Comment: KImax factor is twice as high as KIImax factor and KIIImax factor. KImax is
higher than KC (78.97 MPa.m1/2 vs 54 MPa.m1/2). KImax is the greatest at intersectant
place of bottom flange and web of diaphragm. Hence, propagation of fracture is
governed by Mode I. The crack develops in longitudinal direction of the diaphragm
and extends up to web of diaphragm.

5 Conclusions

Local analysis results of CB6 diaphragm on P29 backward pier of Vam Cong
Bridge in the two cases of diaphragm without and with initial crack have shown:
As regards to deflection: Deflections in two cases are the same when fractures have
not propagated yet.
As for stress analysis: At the crack tip, stress state is three-dimensional state of
stresses and equivalent stresses in the case of diaphragm with initial crack are much
higher than those in the case of diaphragm without initial crack.
The investigation result has shown that propagation of crack is governed by Mode
I. Development direction of crack width is according to longitudinal direction of the
diaphragm and development of crack reaches above web of diaphragm. This result is
appropriate in comparison with real crack appearing in the diaphragm.
From this early investigation result, one can conclude that appearance of initial
crack could result in crack failure in CB6 diaphragm on P29 backward pier of Vam
Cong Bridge.

References
1. Design Documents of Vam Cong Bridge.
2. http://dtinews.vn/en/news/017/53854/crack-found-in-usd270m-bridge-yet-to-enter-
use.html.
3. Reza Haghani, Mohammad Al-Emrani and Mohsen Heshmati, Fatigue-Prone Details in
Steel Bridges, Buildings 2012, 2, 456-476; doi:10.3390/buildings2040456.
4. Ansys Inc. headquarters building in Canonsburg, Pennsylvania.
5. RM Bridge. Advanced Bridge Modeling, Analysis, Design, and Construction
Engineering, BENTLEY SYSTEMS.
 A Study on Combination of Two Friction Dampers to
Control Stayed-cable Vibration under considering
its Bending Stiffness

Duy-Thao Nguyen1, Duy-Hung Vo2

1,2
The University of DaNang ± University of Science and Technology, 54 Nguyen Luong Bang
Street, Danang city, Vietnam
Email: 1ndthao@dut.udn.vn, 2vdhung@dut.udn.vn

Abstract. The stayed-cable is one of vital component of cable-stayed bridges.

Stayed-cable is often very long with a small diameter and low mass, which can
be considered horizontal flexible structure with very low natural frequency.
Under the influence of cyclic load in specific conditions, stayed-cable can store
the energy and vibrate with large amplitude. This paper focuses on studying the
methods of two-friction damper combination for reducing the cable vibration,
and evaluates the capacity of friction-damper parameters in mitigating the vi-
bration of stayed-cable. The results show the relationship between the damping
factor of stayed-cable and various parameters such as Equivalent viscous con-
stants, friction, spring constant, points attached damper to stayed-cable. For
long-span cable-stayed bridges, cable has relatively large diameter and it is
normally covered by grouting mortar or Epoxy. Consequently, its bending stiff-
ness is considerably increased. Therefore, it is necessary to take into account its
bending stiffness during the vibration analysis process. From these results, de-
signers can assess and choose the attaching point as well as parameters of fric-
tion damper, which are optimal for specific stayed-cable.

Keywords: Stayed-cable, damping ratio, natural vibration frequency, friction

damper.

1 Introduction

Stayed-cable is one of vital component of cable-stayed bridges. During its lifetime,

stayed-cable usually vibrates strongly under some cycle loads such as: wind load, live
loads, interaction of rain and wind loads«WKHQLW ZLOOEHGHVWUR\HGE\IDWLJXHSKe-
nomenon. To mitigate the vibration of cables, some damper devices are used as a
damping countermeasure: Viscous damper (VD) [1], [2], [3]; High damping rubber
damper (HDRD) [4], Friction damper (FD) [5]. However, due to aesthetical reasons,
WKHGDPSHUGLVWDQFHLVFRQVWUDLQHGZLWKLQDIHZSHUFHQWRIWKHFDEOH¶VOHQJWKDQGWKH
damper performance is then quite limited. As a result, in application to long-span
cable-stayed bridges, such as Tatara (Japan) or Stonecutters (Hong Kong), where the
longest cables are over 500m, it may be tough to attain a desired level of modal damp-

© Springer Nature Singapore Pte Ltd. 2020 87

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_10
88 D.-T. Nguyen and D.-H. Vo

ing with a single damper allocation. To increase the modal damping, using two units
of dampers at different locations may constitute a possible solution. In this paper,
authors focus on how to reduce the vibration of stayed-cable using combination of
two friction dampers. In addition, the Matlab program was also generated to design
damper devices. The Matlab program can also optimize parameters of damping de-
vice and the suitable location at which the damping devices should be attached. The
taut cables in long-span Cable-stayed bridges which have relatively big diameter, is
HVSHFLDOO\ FRYHUHG E\ PRUWDU RU HSR[\ DQG WKXV FDEOH¶ EHQGLQJ VWLIIQHVV LQFUHDVHV
significantly. Therefore, it is necessary to take into account the bending stiffness of
cable during the analysis of its vibration.

2 Model of a Stayed-cable attached two friction dampers

Given a cable with tension force (S), length (L) and the mass of damper is Gi. Two
friction dampers are installed at both ends of cable. Each friction of damper is mod-
eled by a viscous force connecting with a spring force as Fig. 1
l1 L-l1-l2 l2

S x EJ,m S

c eq1 u(x,t) ceq2

r(x,t) G1 r(x,t) G2

k1 k2

Fig. 1. A Schematic illustration of a stayed-cable attached to two Friction Dampers

It is assumed that the cable tension force is much larger than its weight and the fric-
tion between cable and the air is negligible. Taking in to account the bending stiffness
of cable and the mass of damper, as Duy-Thao at el [6] and VSL [7] the differential
equation of free vibration of the system cable with damper devices are:
2 4 2
­ wu wu 2 wu w (u  r )2

°° wt  a wx  2 s wx c . wt ( x, t )G ( x)
2 4 2 eqi Xdi
(1)
® 2
° G d r  k r (t )  c d (u ( X )  r ) ( t ) 0 di
(2)
°̄ i dt i 2
dt
eqi

Where u=u(x,t)-WKHGLVSODFHPHQWRIFDEOHDWSRLQW[į-WKH'LUDF¶VIXQFWLRQ; r(x,t)-the

displacement of damper device; Xd-the point attached damper device; Gi-the mass of
ith damper device; ki and ceqi are the spring constant and the equivalent viscous con-
stant of ith damper device.
Use the separation of variables with Fourier series, the solution of Eq. (1) can be
found as:
A Study on Combination of Two Friction Dampers … 89

u( x, t ) X ( x)T (t ) (3)

From Eq. (2), the differential equation of damper device is rewritten as Eq. (4):
iK t
Gi rr((t )  ceqiq .r (t)
( )  ki r (t ) X (li ).ceqi .i .K.e (4)

The solution of Eq. (4) has the form as follow

X (li ).2G i . i .K iK t
r (t ) 2 2
.e (5)
(Zi  2.G i .i.K  K )

From the boundary conditions and the equilibrium in the vertical direction at the
points viscous damping device attached to, we can get the frequency equation of sys-
tem cable and damper device:
2
§ EJ .m 2 · ª m m m m m m º
¨ 2 K  1¸ «cot(K S l1 ).cot[K S ( L  l1  l2 )]  cot(K S l1 ).cot(K S l2 )  cot(K S l2 ).cot[K S ( L  l1  l2 )]  1» 
© S ¹ ¬ ¼
ª m m ceq1 § 2G1 .K · m º (6)
«[ cot[K ( L  l1  l2 )]+ cot(K l2 )] ¨i  2 2 ¸
 [ cot[K ( L  l1  l2 )]+ »
§ EJ .m · S S mS © (Z  2.G .i.K  K ) ¹ S
K  1¸ « »
2 1 1
¨
© S 2
¹« m c § 2G 2 .K · »
« + cot(K l2 )] eq 2 ¨ i  2 2 ¸ »
«¬ S mS © (Z 2
 2.G 2
.i.K  K ) ¹ »
¼
ceq1 § 2G1 .K · ceq 2 § 2G 2 .K ·
 ¨i  ¸ ¨i  ¸
mS © (Z12  2.G1 .i.K  K 2 ) ¹ mS © (Z2 2  2.G 2 .i.K  K 2 ) ¹

where Și- the frequency of cable at ith mode.

Given k as eigenvalue of cable with complex form, k is defined as:

ki V i  j.Mi  [  j.
i
1  [i
2
K / Z
i
0
1
; [i 1  Mi
2
/ V i2 (7)

where Ȧ01- the natural frequency of cable without damping device; ȟi- the damping
ratio of cable at ith mode.
Substitute Eq. (7) into Eq. (6) and separating into real and imaginary part, we can get
the system of equation (8) and (9). Where A '. V 2  I 2  ; B '.V .I ;
a S . 1  l1 / L  l2 / L  .V ; b S . 1  l1 / L  l2 / L  .I ; c S .V .l2 / L ; d S .I.l2 / L ;
2 2
e S .V .l1 / L ; f S .I.l1 / L ; ' S . EJ / ( S . L ) ; 4i ceqi / m.S ; O =Z
0
1
;
4.F / (S .Z. Ai ) ; F | Ath .S .S / L ;
0
ceqi Z1 S / m .S / L ; M1 cosh b  cos a ; 2 2

2 2 4 3 2 2 2 2 2 2 2 2 2
M2 cosh d  cos c ; M 3
2 2
cosh f  cos e ; T (I  V ) O  2O G1I (I  V )  2O Z1 (V  I )  Z1 (Z1  2G1IO ) ;
2 2
1 2 2 2 2 2 2 2
[Z1  2G1OM  O (V  I )]  (2G1OV  O 2MV )
2G1OV (Z12  O 2 (I 2  V 2 )) (I 2  V 2 )2 O 4  2O 3G 2I (I 2  V 2 )  2O 2Z22 (V 2  I 2 )  Z22 (Z22  2G 2IO )
T2 T
[Z1  2G1OM  O 2 (V 2  I 2 )]2  (2G1OV  O 2 2MV )2 3
2
[Z22  2G 2 OM  O 2 (V 2  I 2 )]2  (2G 2 OV  O 2 2MV )2
; ;
2G 2 OV (Z2 2  O 2 (I 2  V 2 ))
T4
[Z2  2G 2 OM  O 2 (V 2  I 2 )]2  (2G 2 OV  O 2 2MV )2
2

.
90 D.-T. Nguyen and D.-H. Vo

ª sin 2e.sin 2a  sinh 2 f .sinh 2b sin 2c.sin 2a  sinh 2 d .sinh 2b º

«  »
M 1 .M 3 M 1 .M 2
2 2 « »
¬ª A  2 A  4 B  1¼º « sin 2e.sin 2c  sinh 2 f .sinh 2d »
«   4 »
¬ M 2 .M 3 ¼
ª sin 2e.sinh 2b  sin 2a.sinh 2 f sin 2e.sinh 2d  sin 2c.sinh 2 f sin 2c.sinh 2b  sin 2a.sinh 2d º
 ¬ª 4 B  A  1 ¼º «   »
¬ M 1 .M 3 M 2 .M 3 M 1 .M 2 ¼ (8)
ª § sin 2a sin 2e · § sin 2a sin 2c · § sinh 2b sinh 2 f · § sinh 2b sinh 2 d · º
4 B «  ¨  ¸ 4 2 .T3  ¨  ¸ 41 .T1  ¨  ¸ 4 2 .T4  ¨  ¸ 41 .T2 »
«¬ © M 1 M3 ¹ © M1 M2 ¹ © M1 M3 ¹ © M1 M2 ¹ »¼
ª§ sinh 2b sinh 2 f · § sinh 2b sinh 2d · § sin 2a sin 2e · § sin 2a sin 2c · º
2  A  1 «¨  ¸ 4 2 .T3  ¨  ¸ 41 .T1  ¨  ¸ 4 2 .T4  ¨  ¸ 41 .T2 »
¬«© M 1 M3 ¹ © M1 M2 ¹ © M1 M3 ¹ © M1 M2 ¹ ¼»
4414 2 (T2T4  T1T3 ) 0

ª sin 2e.sin 2a  sinh 2 f .sinh 2b sin 2c.sin 2a  sinh 2d .sinh 2b sin 2e.sin 2c  sinh 2 f .sinh 2d º
ª¬ 4 B  A  1 º¼ «    4»
¬ M 1 .M 3 M 1 .M 2 M 2 .M 3 ¼
ª sin 2e.sinh 2b  sin 2a.sinh 2 f sin 2e.sinh 2d  sin 2c.sinh 2 f sin 2c.sinh 2b  sin 2a.sinh 2d º
 ª¬ A2  2 A  4 B 2  1º¼ «   »
¬ M 1 .M 3 M 2 .M 3 M 1 .M 2 ¼ (9)
ª§ sin 2a sin 2e · § sin 2a sin 2c · § sinh 2b sinh 2 f · § sinh 2b sinh 2d · º
2  A  1 «¨  ¸ 4 2 .T3  ¨  ¸ 41.T1  ¨  ¸ 4 2 .T4  ¨  ¸ 41.T2 »
«¬© M 1 M3 ¹ © M1 M2 ¹ © M1 M3 ¹ © M1 M2 ¹ »¼
ª§ sinh 2b sinh 2 f · § sinh 2b sinh 2d · § sin 2a sin 2e · § sin 2a sin 2c · º
4 B «¨  ¸ 4 2 .T3  ¨  ¸ 41 .T1  ¨  ¸ 42 .T4  ¨  ¸ 41 .T2 »
«¬© M 1 M 3 ¹ © M 1 M 2 ¹ © 1 M M 3 ¹ © 1 M M 2 ¹ »¼
44142 (T1T4  T2T3 ) 0

The system of Eq. (8) and Eq. (9) are the nonlinear equation. The Matlab program
Cable-BKDN is generated to solve the system by iteration method.

3 Numerical analysis results

In order to evaluate the effects of friction dampers on reducing vibration of cable,

various parameters of damping device such as equivalent viscous constant ceq, spring
constant k, mass of damper device G and location attached to damper have been in-
vestigated. With each parameter of fiction damper changes, the damping ratio of cable
is solved from the system of Eq. (8) and Eq. (9). The following sections present effi-
ciency of friction damper in reducing the vibration of cables.

3.1 Influence of Ceq on mitigating the vibration of cable

The equivalent viscous constant of friction damper is characterized by the non-

dimensional coefficient Ĭi = ceqi/(m.S)0.5. The investigation of variation of the first
GDPSLQJ UDWLR RI FDEOH ³ȟ1´ ZLWK SDUDPHWHUV Ĭ1 = [0÷100] corresponding to Ĭ2 =
3,7,10,23,30 (or Ĭ2 = [0÷100] corresponding to Ĭ1 = 2,8,10,20,30) and points damp-
ing device attached to l1/L= l2/L =0.03, are shown in Fig. 3 and Fig. 4. From the re-
sults in the Fig. 3 and Fig. 4, it shows that WKHILUVWGDPSLQJUDWLR³ȟ1´LQFUHDVHVDORQJ
with the raise of Ĭi +RZHYHU DIWHU UHDFKLQJ WKH PD[LPXP SRLQW ³ȟ1´ VORZO\ Ge-
creases when Ĭi increases. Therefore, it is possible to RSWLPL]H WKH YDOXH ³ceqi´ RI
friction damper and determine the maximum damping ratio of cable.
A Study on Combination of Two Friction Dampers … 91

Furthermore, the optimal value Ĭ1 GRHVQ¶WFKDQJHZKHQĬ2 changes. The same thing

also happens with Ĭ2 when Ĭ1 changes. It means that two friction dampers work in-
dependently of each other and the total damping effect is the sum of the contribution
from single friction dampers

Optimal value
Optimal value

Fig. 2. 5HODWLRQVKLSEHWZHHQȟ1-Ĭ1 Fig.4. 5HODWLRQVKLSEHWZHHQȟ1-Ĭ2

3.2 Influence of the friction force on decreasing cable vibration

Friction of damper can have considerable effect on damping ratio of cable by value:
F=Athʌ6/ 7KH LQYHVWLJDWLRQ RI YDULDWLRQ RI WKH ILUVW damping rDWLR RI FDEOH ³ȟ1´
with parameters F = [0÷30] kN and vibration amplitude A=0.005, 0.01, 0.015, 0.02
(m), gives us the result showed in Fig. 5. From the results in the Fig. 5, corresponding
to the vibration amplitude A and optimal friction F, the damping ratio of cable can
reach the maximum value. When the friction exceeds the optimal friction (F) the
damping ratio will decrease. The reason is that if friction is so high, the cable is
locked at the point attached to damper. The clamping force between damper and cable
will be very high; two components will not slide over each other and leading to not
reduce vibration of cable.

Optimal value

Fig.5. 5HODWLRQVKLSȟ1-F1 (various A1) Fig.6. 5HODWLRQVKLSȟ1- Ĭ1 (various point

attached damper l1/L)

3.3 Influence of the attachment point on reducing vibration of cable

7KHLQYHVWLJDWLRQRIYDULDWLRQRIWKHILUVWGDPSLQJUDWLRRIFDEOH³ȟ1´ZLWKSDUDPHWHUV
Ĭ1 = [0÷100] and the point attached damper l1/L= 0.02, 0.03, 0.04, 0.05, 0.1, gives us
the result showed in Fig. 6. ,W LOOXVWUDWHV WKDW WKH ILUVW GDPSLQJ IDFWRU ³ȟ1´ ZLOO EH
higher when the ratio l/L is larger. However, this value requires lifting up damping
devices to a high position in comparison with the bridge deck. Therefore, this may
bring about difficulties in designing, installing as well as maintaining damping devic-
es and have influence on architectural bridges. Damping devices is usually installed at
the point from 0.02 to 0.05.
92 D.-T. Nguyen and D.-H. Vo

3.4 Influence of the bending stiffness on reducing vibration of cable

The damping ratio of cable depends on cable characteristics as: bending stiffness EJ,
tension force S and the length of cable L. The introduction of non-dimension parame-
WHUǻLVDLPHGDWVLPSOLI\LQJWKHLQYHVWLJDWLRQV7KH SDUDPHWHUǻUHIOHFWVWKHEHQGLQJ
VWLIIQHVV DQG EH GHWHUPLQHG E\ ǻ ʌ2.EJ/(S.L2). The investigation of variation of the
ILUVW GDPSLQJ UDWLR ³ȟ1´ ZLWK SDUDPHWHUV ǻ  >·@ DQG SRLQWV DWWDFKHG damper
li/L=0.02; 0.03; 0.04, it shows that the first damping ratio gradually increases while ǻ
gradually increases. The rate of relative rise of the first damping ratio with l i/L = 0.02
can reach more than 8%. This value is significant when considering the vibration of
the cable.

4 Conclusion

This study focuses on application of two friction dampers in the aim of controlling the
vibration of cable. The research also evaluated the influences of friction damper pa-
rameters in reducing the vibration of cable. The results showed the relationship be-
tween the damping ratio of cable and various parameters of damping device as:
equivalent viscous constant ceq; spring constant K, the relationship between the damp-
ing ratio and many different points damping device attached to, and the effect of
bending stiffness of cable on mitigating its vibration by a non-GLPHQVLRQSDUDPHWHUǻ
. Based on current study, structural engineer can assess and choose the attachment
point as well as other parameters of two friction dampers, which are optimum for
stayed-cable.

References
1. -$0DLQ13-RQHV³)UHH9LEUDWLRQVRI7DXW&DEOHZLWK$WWDFKHG'DPSHU,/LQHDU
9LVFRXV'DPSHU´Journal of Engineering Mechanic, Vol 128, No10, pp 1062-1071 (2002).
2. -$0DLQ13-RQHV³)UHH9LEUDWLRQVRI7DXW&DEOHZLWK$WWDFKHG'DPSHU,,1on-
/LQHDU'DPSHU´-RXUQDORI(QJLQHHULQJ0HFKDQLF9RO1RSS-1081 (2002).
3. <R]R )XMLQR DQG 1DP +RDQJ ³'HVLJQ )RUPXODV IRU 'DPSLQJ RI D 6WD\ &DEOH ZLWK D
'DPSHU´-RXUQDORI6WUXFWXUH(QJLQHHULQJ$6&(SS-278(2008).
4. 9LHW +XQJ &X %LQJ +DQ ³+igh-damping rubber damper for taut cable vibration reduc-
WLRQ´$XVWUDOLDQ-RXUQDORI6WUXFWXUDO(QJLQHHULQJ9ROXPH-Issue 4, pp 283-291(2015).
5. (OVDGH6i&DHWDQR³&DEOH9LEUDWLRQVLQ&DEOH-6WD\HG%ULGJHV´6WUXFWXUDO(QJLQHHULQJ
Documents, (2007).
6. Duy-Thao Nguyen and Van-0\1JX\HQ³9LEUDWLRQFRQWURORI&DEOHXVLQJYLVFRXVOLQHDU
GDPSHUVLQFRQVLGHUDWLRQRIEHQGLQJVWLIIQHVV´7KH:RUOG&RQJUHVVRQ$GYDQFHVLQ
6WUXFWXUDO(QJLQHHULQJDQG0HFKDQLFV$6(0¶6HRXO-Korea, p 2673-2772 (2011).
7. VSL Technical Centre Asia/ Australia, Friction damper calculation, Dec-2012.
 Design proposal and behavior simulation of prestressed
concrete slab track at highway-railway grade crossings

Pham Van Hung1, Tran The Truyen2, Tran Anh Dung2, Doan Bao Quoc2,

Le Hai Ha2, Nguyen Hong Phong2

1 University of Paris-Est Marne-la-Vallee, France
2
University of Transport and Communications, Hanoi, Vietnam.
trananhdung@utc.edu.vn

Abstract. This paper presents the design proposal of the prestressed concrete slab
(PSCS) track used for highway-railroad grade crossings in Vietnam. A new type
of highway-railroad grade crossings is being proposed to replace the traditional
panel crossings made by reinforced concrete, asphalt concrete and rubber.
Numerical simulation was carried out to analyze the structural behavior of the
PSCS. The results show that the structural proposal of the PSCS meets the
requirements of stability and strength under the standard loads of truck and train
engines recommended in Vietnamese specifications.

Keywords: Highway-railway grade crossings, Slab track, Prestressed concrete

slab.

1 Introduction

In the field of transport, the construction of railroad crossings between railways and
roads has been a problem for the railway system in many countries [1-4]. In Vietnam,
the system of railroad crossings with sectional reinforced concrete slabs [5], asphalt
concrete [5] or rubber panels [5-6] are three popular forms that are being applied in
reality (Fig. 1). However, this system is severely degrading, and there are many safety
risks. As evidence suggests, it is necessary for us to develop a system of railroad
crossings to ensure that all components are smooth, durable, and easy to construct and
repair, which helps to reduce maintenance costs and prolong service life while
maintaining the train’s speed requirements through the railroad crossings.
This paper is one of the studies which proposes and designs monolithic highway-
railway grade crossings by prestressed concrete slab (1000mm gauge [7]). The purpose
of this article is to construct a reliable numerical model for PSCS widely used for the
highway-railway grade crossings in Vietnam. This model is established using Midas
FEA software [8] to simulate the slab which we had designed and proposed.

© Springer Nature Singapore Pte Ltd. 2020 93

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_11
94 P. V. Hung et al.

a. Sectional reinforced b. Asphalt concrete c. Rubber panels

concrete slabs
Fig. 1. Three typical crossing surfaces at highway-railway grade crossings in Vietnam.

2 Proposal design of prestressed concrete slab

The result of this research will confirm and clarify the effectiveness of using PSCS
structure compared to normal concrete in the construction of highway-railway grade
crossings. Moreover, PSCS structure is also considered as the first application of
material and new structure of highway-railway grade crossings in Vietnam. In this
study, we give some proposals for PSCS with some main features. PSCS with length
2.2m and width 1.0m, rebar with diameter 10mm and prestressed cable with diameter
8mm were used in the experiment. PSCS is linked with the platform. The PSCS design
is used to replace for two concrete sleepers in the railway. Besides, there are other items
and rails, as described in Fig. 2.
PRESTRESSED CONCRETE SLAB ARRAGEMENT
(SCALE:1/10)
f1-d10 a1-D8 a1-D8 f1-d10
f2-d10 a1-D8
30
150
330

45 3x35=105

CABLE d8 c1-D8 a2-D8 c1-D8

75 125 18x100=1800 125 75

2200

Fig. 2. The typical cross section of PSCS.

3 Simulating the behavior of prestressed concrete slab

3.1 Numerical simulation process

Nonlinear models were assigned to all concrete elements and steel reinforcements to
take into account damaged zones of the slab. The simulation process was carried out
with control of the load. PSCS is defined as a nonlinear material with concrete
mechanical parameters including compressive strength of 58.0 MPa, tensile strength of
5.71 MPa, Young modulus of 37 MPa, Poisson coefficient of 0.2 and density of 1,700
kg/m3. The longitudinal and transversal reinforcement consists of φ8 and φ10 steel rebar
with mechanical properties: Young modulus of 210,000 MPa. The prestressed cable of
 Design proposal and behavior simulation of prestressed … 95

φ8 with mechanical properties: Young modulus of 200,000 MPa, effective prestress

pre-wire of 1,290 MPa. Smeared 3D crack model is considered to simulate in these
examples. Slab model consists of concrete, reinforcing steel, and rail. The authors have
experimented with two mesh sizes. Firstly, PSCS is divided into 44262 solid elements.
Secondly, PSCS is divided into 88524 solid elements. Both ways of meshing give the
same results. However, the second way of meshing takes four times more time to
calculate the mesh. Therefore, considering the effectiveness in terms of time and
accuracy, the authors used the meshing method with 44262 solid elements and 10-node
tetrahedrons. Finite element meshes are constructed so as to be fine in unfavorable
zones and sparse inelastic zones to reduce calculation time. Nonlinear models were
assigned for all concrete elements based on Hordijk and Thorenfeldt models
implemented in Midas FEA; linear elastic model was assigned for steel reinforcements.
Vertical load is up to the value of 22 tons/axle (T22) within dynamic load allowance,
which is equal to 20,163kN. Finite Element Method (FEM) mesh is shown on Fig. 3.

Fig. 3. Finite element mesh.

3.2 Results and discussion of static analysis

The standard axle load of a train used in Vietnam is 22 tons/axle. Considering the
coefficients of shock effect is 1+IM = 1.6. After applying, modeling increased from 0
kN to P= 400kN>Pmax (1+IM) = 352 kN. The values of stress fields in concrete and
axial force in steel rebar are extracted as the output of this simulation.
7
Mid-slab deflection (μm)

6 Mesh: 25mm/one element

Mesh: 50mm/one element
5
4
3
2
1
0
0 100 200 300 400
Load (kN)

Fig. 4. Load versus mid-slab deflection and stress field at cross section of PSCS.

In Fig. 4, a high local compression at the location of load is shown, which is up to 21 –

22 MPa. This value is smaller than the maximum compression strength in concrete
which is equal to 58 MPa. It means that under the equivalent static load of T22, the
96 P. V. Hung et al.

design PSCS does not reach a failure point of the material. As a result of the FEM
analysis, we detected the complete fracture of some solid elements. The crack can occur
with complicated progress due to the interaction effect between the slab and the
subgrade. This further strengthens the need for reinforcement by steel rebar at this
compression concentration zone.
3.3 Dynamic analysis

Modal analysis
As the most important part of the dynamic analysis, the modal analysis of the structure
is carried out. In the modal, the six first eigen-vectors of the PSCS and their
corresponding fundamental frequencies are 28.584 Hz, 28.673 Hz, 37.796 Hz, 387.998
Hz, 424.760 Hz, and 432.657 Hz respectively. These frequencies are well above that of
normal trains, which is estimated to be about 1 Hz at the most.
Analysis of train load as moving mass at a constant velocity

Fig. 5. Model of dynamic load.

In this experiment, the team used a train of 22 tons/axle. The train moves in two
directions:
The horizontal (Y-direction) with moving v is v, velocity v = 40 km / h.
Since the velocity is straight, study should be done at t = 0.09s (from t = 0). The shaft
has a mass of 22 tons/axle reaching the end point of PSCS (1.0m in length). Nodes 1766
and 1869 are the locations where the force begins to move. Node 1858 and Node 1961
are the position of the force that ends the motion.
The vertical method (Z-direction) is described in the following diagram:

    
      (1)


The coefficients are chosen as follows:

A = 10000 (N) D = 0.5 Damping Factor
C = 5000 (N/sec) f=1 Frequency (cps)
t = 10 (sec) P = 0.5
The above waveform is called “a shock absorber's oscillator”, in other words, it is the
inverse diagram of the shock waveform of the shaft acting on the rail [9]. Gravitational
acceleration is g = 9.80665 m/s², ignoring the friction between the train and the rail.
Consider the frequency of the mode N°1. f1 = 28.584 Hz. This frequency is the easiest
of the five modes when passing through a train. In other words, frequency f1 in the
frequency range of the train's oscillation. Points on PSCS’ boundary that connect with
Design proposal and behavior simulation of prestressed … 97

the highway are always dangerous points (Fig. 5). According to the analysis results,
displacement, velocity, and acceleration at those points are the largest. We need to
improve the materials and structures of PSCS at those locations. However,
displacement and acceleration at those points are not too large. The acceleration
changes quite smoothly. There are three points where acceleration equal to 0 per one
meter of rail. Therefore, PSCS is a good shock absorber. Analysis results include
displacement, velocity, stress and acceleration which are generated using a moving
mass model of T22 trains (Fig. 6-Fig. 9).

Table 1. The maximum value when using mass model.

Displacement(mm) Velocity(mm/sec) Acceleration(mm/sec2) Stress(N/mm2)
1.83060E-001 2.70611E+000 8.922E-3 1.17909E+002
The speed of 40 km/h is the flat speed of axle T22 on the PSCS. Through calculation
analysis, this study yields the following charts.

Displacement Velocity
0.16 2.4

0.15
Displacement (mm)

Velocity (mm/sec)

2.2
0.14
2.0
0.13
1.8
0.12

0.11 1.6

0.10 1.4
0 200 400 600 800 1000 0 200 400 600 800 1000
Length PSCS (mm) Length PSCS (mm)
Fig. 6. Displacement of PSCS from the Fig. 7. Velocity of PSCS from the dynamic
dynamic loads caused by the moving mass. loads caused by the moving mass.

Acceleration Von mise stress

0.010
25
0.008
Acceleration (mm/s2)

0.006
20
Stress (N/mm )

0.004
2

0.002 15
0.000
-0.002 10
-0.004
-0.006 5
-0.008
-0.010 0
0 200 400 600 800 1000 0 200 400 600 800 1000
Length PSCS (mm) Length PSCS (mm)
Fig. 8. Acceleration of PSCS from the Fig. 9. Stress of PSCS from the dynamic
dynamic loads caused by the moving mass. loads caused by the moving mass.

The railway sector in Vietnam has speed limits for each segment. The speed of research
results is completely safe for PSCS since the trains run through the highway-railway
grade crossings in Vietnam at approximately 20km/h. The analysis results compared
with L. Björklund’ results [10] are suitable.
Analysis of train load in an earthquake scenario
98 P. V. Hung et al.

The team performed calculations in an earthquake disaster scenario. The earthquake

signal data is taken in historical earthquakes. In this study, the data are collected from
the battle of T2-III-3 (1995, Hyogoken South, Ew) [11] as follows: Displacement:
8.756E-007 m; Velocity: 1.425E-005 (m/sec); Acceleration: 3.042E-001 (m/sec2) and
Stress: 3.457E+002 (kN/m2). From these results, we can see that natural disasters do
not pose significant threats against this structure. Through the above theoretical
calculations and simulations, we can assert that the proposed structure has met all the
required conditions.

4 Conclusions and perspectives

In this paper, the design of the PSCS is examined by analytical and numerical
calculations. Through calculation examples, the practicability of using PSCS instead of
conventional concrete is clearly shown. Hence, PSCS material is applied successfully
for many constructions over the world in civil engineering, as well as bridges and roads.
Numerical study is essential to validate the research results and get in-depth insight into
the practical capacity of PSCS on the highway-railway grade crossings in Vietnam.
From the perspectives of this study, authors want to use the nonlinear model of train
T22 for a special simulation to understand the modification in special behavior of
PSCS.

References
1. Esveld, C.: Slab track-a competitive solution, In Rail International, Schienen der Welt
(1999).
2. Galvín, P et al.: Vibrations induced by HST passage on ballast and non-ballast tracks,
Journal of Soil Dynamics and Earthquake Engineering, 30, pp. 862–873 (2010).
3. Zeinz, T.: Highway-Rail Grade Crossing Safety Course, Dept. Eng, Univ., Wisconsin,
Madison, WI (2007).
4. Rose, G.: Highway-railway at grade crossing structures: Long-term settlement
measurements and assessments, Research ReportKTC-09-06/FR136-04-3F (2009).
5. Circular No.15/2018/TT-BGTVT of Transport Ministry.: Regulations on crossroads and
licensing of construction of essential works within the land area for railways (2018).
6. National standard TCVN 12268:2018.: Specifications and test methods (2018).
7. National standard TCVN 11793:2017.: Design Requirements for Track Work (2017).
8. MIDAS Information Technology Co., Ltd.: Midas FEA software (2018).
9. Komen, B.: Assessment of passenger comfort according to Eurocode in Dutch trains
travelling over tiedarch railway bridges, Master of Science Thesis, Faculty of Civil
Engineering and Geosciences, Delft University of Technology, (2016).
10. Björklund, L.: Dynamic Analysis of a Railway Bridge subjected to High Speed Trains,
Master of Science Thesis, Department of Civil and Architectural Structural Engineering, the
division of Structural Design and Bridges, at the Royal Institute of Technology in
Stockholm, (2004).
11. Geographical Survey Institute of Japan.: Report of Emergency Damage Survey of the 1995
Hyogoken Nanbu Earthquake (1995).
 Effects of Random Road Roughness on Dynamic Impact
Factor of multi-span Super T Girder Bridge with Link
Slab due to Moving Vehicles

Xuan-Toan Nguyen1, Duy-Thao Nguyen2

1,2
The University of Danang -University of Science and Technology, 54 Nguyen Luong Bang
Street, Danang city, Vietnam
Email: 1nxtoan@dut.udn.vn; 2ndthao@dut.udn.vn

Abstract. The purpose of paper analyzes influence of random road roughness

on dynamic impact factor of Bridge subjected to a moving vehicle. The road
roughness is assumed by a stationary random process. The bridge has three
spans which is modeled by FEM. The moving vehicle has three axles and is
idealized by 6-DOFs. Monte-Carlo method is used to create the random road
roughness input. The governing equation of dynamic vehicle-bridge interaction
is established by means of d’Alembert’s principle. Galerkin method and Green
theory are applied to discrete the governing equation in the space domain. In
this paper, Runge-Kutta methodology was used to solve the governing equation
in the time domain. After analyzing with a series profile of road roughness in-
put, it is going to obtain the response of bridge output which is also the random
process. The FEM analysis totally agrees with the test results of KhueDong
Bridge. Furthermore, the road roughness also was investigated to clarify the ef-
fects on dynamic impact factor.
Finally, when road roughness condition changes from the type A to type E, dy-
namic impact factor has risen considerably.

Keywords: Dynamic impact factor (DIF), Road roughness, Truck, Super T

Girder Bridge, Monte-Carlo method, Runge-Kutta methodology.

1 Introduction

The early structural engineers found that under moving loads, structural dynamic
deformations and stresses can be significantly higher than those caused by corre-
sponding static loads. Additional dynamic effects are accounted for by dynamic im-
pact factors (DIF or IM) introduced in bridge design codes [1]. There are a large
number of studies on this topic including experimental DIF, analytical methods and
code specifications. The effect of road roughness on bridge dynamic behavior was
investigated by Coussy et al. [2]. In this paper, the suspended moving loads were
used. The results of this study showed that the DIF could be three times compared to
international design standards recommendation. The frequency domain method was
proposed by Lombaert and Conte [3] to analyze the interaction of vehicle and road
roughness. The effect of vehicle braking force to bridge dynamic was studied by
Xuan-Toan Nguyen et al. [4]. In this paper, three-axle vehicle-bridge interaction with

© Springer Nature Singapore Pte Ltd. 2020 99

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_12
100 X.-T. Nguyen and D.-T. Nguyen

different velocity was analyzed. In general, previous studies on dynamic interaction of

bridge-vehicle were for simple beam bridge. Nevertheless, only some of studies were
mentioned about effect of road roughness on dynamic behavior of multi-span Super T
Girder Bridge with link slab. Previous researches were lack of field measurement
result comparison. Therefore, to obtain fully understanding about relation of vehicle –
bridge interaction, further field measurements should be carried out. Current study
uses self-developed FEM program to investigate the effect of road roughness on dy-
namic interaction of truck and three-span Super T concrete girder with link slab.
Moreover, the FEM analysis and field measurement results at KhueDong Bridge
showed very good agreement.

2 Analysis of the dynamic interaction of the coupled vehicle–

bridge system

2.1 Model of the vehicle-bridge system

w
(z)
z 31 m 31 m 21 z 21 z 11
m11
37.6m 37.6m 37.6m
d31 k31 d21 k21 d11 k11 2.4m 2.4m
z 32 m 32 m 22 z22 m 21 z21
k33 d22 k22 d 21 k 21
Fig. 2. Schematic of multi-span Super T
d32
r3 r2 r1
w3 w2 w1 x
O
x3
x2
x1
L

Fig. 1. Truck-bridge interaction model Fig. 3. Bridge cross section

The model of vehicle - bridge dynamic interaction considering road roughness can be
seen in Figure 1. In detail, an Euler-Bernoulli beam with three-axle truck was used.
The mass of truck and good can be allocated to each axle of vehicle m11, m12 and m13
respectively. The weight of each axle denote m21, m22 and m23 respectively. wi(xi,t)is
the girder displacement in vertical direction at ith axle; z1i and z2i denote the displace-
ment of vehicle body and its axle in vertical direction at ith axle; k1i and k2i are the
spring stiffness of suspension and tire at ith axle; d1i and d2i denote the dashpot of sus-
pension and tire at ith axle respectively; the location of the ith axle was coordinated by
xi ; ri denote road roughness at ith axle and can be determined by Sun [5]:
M M
r ( t ) = ∑ 2Sr (ωk ) Δω .cos (ωk t + Φk ) =∑ 2Sr ( Ωk ) ΔΩ.cos (ωk t + Φk ) (1)
k =1 k =1

In which Sr(ωk) and Sr(Ωk) are the power spectral density of road roughness in terms
of frequency and wave number which are classified by ISO 8608:1995 [6]; ω and Ω
represents frequency and spatial frequency; M and Φk denote a positive integer and
independent random variable with uniform distribution from 0 to 2π.
Effects of Random Road Roughness on Dynamic Impact … 101

2.2 Equation of motion for the vehicle-bridge interaction

From truck-bridge interaction model in Fig.1 and using dynamic balance principle,
the equation of motion of the vehicle on the vertical direction can be determined by:
m1i .!!
z1i + d1i .z!1i − d1i .z!2i + k1i .z1i − k1i .z2i = −m1i .g (2)

m2i .!!z2i − d1i .z!1i + ( d1i + d2i ) .z!2i − k1i .z1i + ( k1i + k2i ).z2i = k2i ( wi + ri ) + d 2i .( w! i + r!i ) − m2i .g (3)

Governing equation of bridge girder can be written by Ray’s formula [7]:

⎛ ∂ 4ω ∂ 5ω ⎞ ∂ 2ω ∂ω n
EJ .⎜ 4 + θ 5 ⎟ + ρ m . 2 + β . = ∑ ⎡− ( m1i + m2i ).g − m1i .!!z1i − m2i .!!z2i ⎤⎦ .δ ( x − xi ) (4)
⎝ ∂x ∂x .∂t ⎠ ∂t ∂t i =1 ⎣

Where ρm, EJ, θ and β denote the initial value of bridge girder as: mass per unit
length of girder, bending stiffness, coefficient of internal friction and external friction
of bridge girder respectively; n=3 which is the axle number. δ(x-xi) denote the Dirac
delta function.
By using Galerkin method and Green theory, equation 2 to equation 4 was transferred
into matrix as following:
[ M e ].{q!!} + [Ce ].{q!} + [ K e ].{q} = { f e } (5)

Where {q̈ }, {q̇ }, {q}, {fe} denote acceleration vector, velocity vector, displacement
vector and forces vector of both truck and bridge girder.

⎧ wy 1 ⎫
⎧w !! ⎫ ⎧ w! ⎫ ⎧ w⎫ ⎧ Fw ⎫ ⎪φ ⎪
{q!!} = ⎪⎨ !!z1 ⎪⎬ ;{q!} = ⎪⎨ z!1 ⎪⎬ ;{q} = ⎪⎨ z1 ⎪⎬ ;{ f e } = ⎪⎨ Fz1 ⎪⎬ ;{w} = ⎪⎨ 1 ⎪⎬ (6)
⎪!!z ⎪ ⎪ z! ⎪ ⎪z ⎪ ⎪F ⎪ ⎪ wy 2 ⎪
⎩ 2⎭ ⎩ 2⎭ ⎩ 2⎭ ⎩ z2 ⎭ ⎪⎩ φ2 ⎪⎭

in which wy1, wy2 denote the displacement in vertical direction in the both end points
of bridge girder; ϕ1 and ϕ2 denote rotation angle in the both end points of bridge gird-
er; [Me], [Ce] and [Ke] denote the mass matrix, damping matrix and stiffness matrix of
both truck and bridge girder, respectively;

⎡ M ww M wz1 M wz 2 ⎤ ⎡Cww 0 0 ⎤ ⎡ K ww 0 0 ⎤
⎢ ⎥ ⎢ ⎥ ⎢ ⎥
[ e] ⎢
M = 0 M z1 z1 0 ⎥ [ e] ⎢
; C = 0 C z1 z1 C z1 z 2 ⎥ [ e ]
; K = ⎢ 0 K z1 z1 K z1 z 2 ⎥
(7)
⎢⎣0 0 M z 2 z 2 ⎥⎦ ⎢⎣Cz 2 w Cz 2 z1 Cz 2 z 2 ⎥⎦ ⎢⎣ K z 2 w K z 2 z1 K z 2 z 2 ⎥⎦

where Mww, Cww and Kww denote mass matrix, damping matrix and stiffness matrix of
the bridge girder, respectively, which can be determined in Zienkiewicz [8]. Matrices
Mzizi, Mwzi, Czizi, Cziw, Kzizi, Kziw are calculated by Xuan-Toan Nguyen et al [4] when
neglect rotational inertial moment of vehicle.
102 X.-T. Nguyen and D.-T. Nguyen

⎧− m11.g ⎫ ⎧ − m21.g + k21.r1 + d 21.r!1 ⎫

⎪ ⎪ ⎪ − m .g + k .r + d .r! ⎪
⎪⎪ ! ⎪⎪ ⎪⎪ 22 22 2 22 2 ⎪
n
⎪
Fw = ∑ ⎡− ⎣ ( m1i + m2i ) g ⎤⎦.Pi ; Fz1 = ⎨− m1i .g ⎬ ; Fz 2 = ⎨ ... ⎬ (8)
i =1 ⎪ !⎪ ⎪ − m .g + k .r + d .r! ⎪
⎪ ⎪ ⎪ 2i 2i i 2i i
⎪
⎩⎪− m1n .g ⎭⎪ ⎩ m2 n .g + k2 n .rn + d 2 n .r!n ⎪⎭n×1
⎪−

Using Runge-Kutta method to solve the equation 5 and obtain responses of bridge
girder in the time domain.

3 Numerical simulation and field measurement on KhueDong

Bridge

3.1 Parameters of KhueDong bridge and truck

KhueDong Bridge is multi –span super T girder with link slab, which located in
Danang city, Vietnam. The decks of KhueDong Bridge are connected at flexible
joints with 2.4m of length (Fig.2). The cross section bridge is illustrated in Fig. 3. The
parameters of Super T girder are referenced from design: L=36(m), E=30(Gpa),
θ=0.027, β=0.01, A=1.21(m2), J=0.537(m4); the properties of link slab (deck):
h=0.2(m), A=0.488(m2), J=0.00163(m4). The parameters of truck are given by the
manufacture catalogue and rechecked by site measurement: m11=5000(kg),
m21=260(kg), m12= m13=9000(kg), m22= m23=870(kg), k11=k12=k13= 2x106(N/m),
k21=k22=k23= 3.8x106(N/m), d11=d12=d13= 4000(N.s/m), d21=d22=d23= 8000(N.s/m).

3.2 Numerical analysis

By the site investigation, the road surface condition of KhueDong Bridge is assumed
at Type C-road (ISO 8608:1995): roughness coefficient Sr(Ω0)= 256×10−6 m3/cycle;
exponential γ=2; M=1000; the spatial frequency (Ωk) change from 0.011to 2.83 cy-
cle/m. Road roughness profiles can be simulated by Monte-Carlo method. A random
road roughness profile can be seen in Fig.5. In addition, the vehicle velocity on the
bridge is 36 Km/h (10 m/s). For every road roughness input, equation 5 is solved by
the Runge-Kutta methodology to achieve the static and dynamic displacement of su-
per T Bridge. The detail of analysis result will be discussed in section 3.3.

Fig.5. The profile of road roughness

3.3 Field measurement results

To compare with numerical results in section 3.2, field measurement of dynamic
response for super T concrete girder was conducted at KhueDong Bridge in Danang
city, Vietnam. Displacement sensors were located on the middle of 1st span to check
 Effects of Random Road Roughness on Dynamic Impact … 103

the dynamic displacement of super T girder. Using the excitation from the truck, the
girder vibration will be measured with various velocities. The truck velocity at 10, 20,
30 and 40 km/h were used due to the traffic speed regulation of bridge operator. The
dynamic displacement of super T girder was recorded and compared with the numeri-
cal results (Fig. 6).
The numerical results were totally agreed with the field measurement results. The
maximum differences of dynamic displacement between theory and experiment on
the middle of 1st span are 1.43%; 6.29%, 8.19% and 8.78% with truck velocity 10, 20,
30 and 40 km/h respectively. Thus, the numerical model mentioned in section 3.2 is
reliable.
0.5 0.5 1

FEM results FEM results 0.5 FEM results

0
0

a).
Experiment results

-0.5
b).
Experiment results
0
c). Experiment results

-0.5

Displacement (mm)
-0.5
Displacement (mm)

Displacement (mm)

-1
-1 -1
-1.5
-1.5 -1.5
-2
-2
-2
-2.5
-2.5
-2.5
-3
-3

-3 -3.5 -3.5

-3.5 -4 -4
0 0.5 1 1.5 2 2.5 3 3.5 4 0 0.5 1 1.5 2 2.5 3 3.5 4 0 0.5 1 1.5 2 2.5 3 3.5 4
Vehicle position (mm) 4 Vehicle position (mm) 4 Vehicle position (mm) 4
x 10 x 10 x 10

Fig.6. Dynamic displacement with different velocity: a) v=10 km/h; b) v=20 km/h;
c) v=40 km/h

3.4 Effect of road roughness on DIF of KhueDong Bridge

30
Table1. Statistical characteristics of DIF at 1st mid-span (1+IM) Histogram
25 (1+IM) Lognormal fit
Probability density function

Dynamic Impact Factor (1+IM)

20
Items Sr(Ωo) x10-6 m3/cycle
0 32 128 512 2048 8192 15

Mean
1.156 1.16 1.17 1.187 1.243 1.622 10
value
Maxi- 5
- 1.199 1.248 1.31 1.501 2.455
mum
0
Minimum - 1.122 1.093 1.063 0.986 0.79 0.9 1 1.1 1.2
1+IM
1.3 1.4 1.5 1.6

Standard
deviation
- 0.023 0.045 0.072 0.151 0.488 Fig.7. Dynamic impact factor at 1st
mid-span (grade B)
From the reliability of numerical model of above discussion, dynamic vehicle-bridge
interaction with various road roughness conditions will be investigated. Roughness
coefficient Sr(Ω0) were varied by range of [0, 32, 128, 512, 2048, 8192]×10−6
m3/cycle corresponding to the road roughness: ideal surface, Type A, B, C, D and E
(according to ISO 8608:1995). In these analyses, the truck velocity was fixed at
36km/h. Then, Monte-Carlo simulation method is used to created 100 road roughness
profiles for each the road roughness condition. With each road profile, static and dy-
namic displacements output of super T Bridge can be obtained by using Runge-Kutta
method to solve the equation 5. Consequently, dynamic impact factor was summa-
rized in Figure 7. In addition, the details about statistical characteristics of DIF calcu-
lation are described in Table 1. Obviously, when road roughness changed from type
A, type B, type C, the mean value of IM went up 0.63%, 1.09% and 3.0%, respective-
104 X.-T. Nguyen and D.-T. Nguyen

ly. Particularly, DIF value can rise to 11.26% and 54.47% as road roughness is type D
and E. Therefore, DIF value can exceed the recommended value in design standard
[1].

4 Conclusion

The effect of road roughness on DIF of super T girder was investigated by numerical
and experimental methods. The Monte-Carlo and Runge-Kutta method were applied
in this study. In addition, field measurement also carried out at KhueDong Bridge.
The theoretical analysis showed good agreement with experimental measurement.
Currently, the obtained results are as following:
• Various road roughness conditions were simulated by Monte-Carlo methodolo-
gy. From each roughness, DIF was calculated.
• DIF value can be exceeded the recommendation value in design code when
rough roughness are type D and E. In this study, dynamic impact factor can rise
54.47% when the road roughness changes to be type E.
• Road surface condition should be carefully considered when analyze DIF of
bridge structure. Especially, when the pavement have been deteriorated signifi-
cantly (type D and type E).

References
1. AASHTO, LRFD Bridge Design Specifications, AASHTO, Washington, DC, USA (2012).
2. Coussy, O., Said, M., and Van Hoore, J.P. “The influence of random surface irregularities
on the dynamic response of bridges under suspended moving loads”. Journal of Sound and
Vibration, 130(2): 3 13-320 (1989).
3. Geert Lombaert and Joel P. Conte. “Random vibration analysis of Dynamic Vehicle-
Bridge interaction due to Road Roughness”. Journal of Engineering Mechanics. 816-825
(2012).
4. Xuan-Toan Nguyen, Van-Duc Tran, Nhat-Duc Hoang. “A Study on the Dynamic Interac-
tion between Three-Axle Vehicle and Continuous Girder Bridge with Consideration of
Braking Effects”. Journal of Construction Engineering, Volume 2017 (2017).
5. L. Sun, “Simulation of pavement roughness and IRI based on power spectral density”,
Mathematics and Computers in Simulation 61(2003) 77–88.
6. ISO 8608:1995 ‘‘Mechanical vibration – Road surface profiles – Reporting of measured
data’’ issued by the International Organization for Standardiazation (ISO).
7. Ray, W. C., and Joseph, P. Dynamics of structures. 3rd Ed. Computers & Structures,
Berkeley, California (1995).
8. O.C. Zienkiewicz, R.L. Taylor, J.Z. Zhu. The Finite Element Method: Its Basis and Fun-
damentals, Seventh Edition. Butterworth-Heinemann, (2013).
 $GYDQFHG'HVLJQ6RIWZDUHIRU6WHHO&DEOH6WD\HG
%ULGJHV8VLQJ1RQOLQHDU,QHODVWLF$QDO\VLV

9LHW+XQJ7UXRQJDQG6HXQJ(RFN.LP
7KX\ORL8QLYHUVLW\7D\6RQ'RQJGD+DQRL9LHWQDP

WUXRQJYLHWKXQJ#WOXHGXYQ
6HMRQJ8QLYHUVLW\*XQMD'RQJ*ZDQJMLQ*X6HRXO6RXWK.RUHD

VHNLP#VHMRQJDFNU

$EVWUDFW&DEOHVWD\HGEULGJHVKDYHJDLQHGPXFKSRSXODULW\RZLQJWRWKHLUHV
WKHWLFDSSHDUDQFHHIILFLHQWXWLOL]DWLRQRIVWUXFWXUDOPDWHULDOVDQGWHFKQLFDODG
YDQWDJHV:KHQ WKH PDLQ VSDQ OHQJWK EHFRPHV ORQJHU PRUH DFFXUDWHDQDO\VLV
PHWKRGVDUHUHTXLUHGWRSUHGLFWWKHUHDOLVWLFEHKDYLRUVRIVWUXFWXUHVHVSHFLDOO\LQ
GHVLJQRIVWHHOFDEOHVWD\HGEULGJHVZKHUHQRQOLQHDUEHKDYLRUVVXFKDVVWUXFWXUDO
PDWHULDODQG JHRPHWULF QRQOLQHDULWLHVDUHFRQVLGHUDEOH 7KLV SDSHU SUHVHQWVDQ
LQQRYDWLYHVRIWZDUHQDPHGDV3$$3'IRUQRQOLQHDULQHODVWLFDQDO\VLVRIVWHHO
FDEOHVWD\HGEULGJHVVXEMHFWHGWRVWDWLFDQGVHLVPLFORDGLQJV7KHVROYHURIWKH
3$$3'LVGHYHORSHGXVLQJWKH)2575$1SURJUDPPLQJODQJXDJHZKLOHWKH
SUHDQGSRVWSURFHVVRUVRI3$$3'DUHZULWWHQXVLQJ&SURJUDPPLQJODQ
JXDJH ZLWK D XVHUIULHQGO\ DQG HDV\WRXVH LQWHUIDFH 7KH S\ORQV JLUGHUVDQG
FURVVEHDPVDUHPRGHOHGDVSODVWLFKLQJHEHDPFROXPQHOHPHQWVZKHUH ܲ െ ο
DQGܲ െ ߜHIIHFWVWKHLQLWLDOJHRPHWULFLPSHUIHFWLRQDQGWKHJUDGXDOVWLIIQHVV
GHJUDGDWLRQDUHFRQVLGHUHG7KHVWD\FDEOHVDUHPRGHOHGDVFDWHQDU\HOHPHQWVWR
FDSWXUHWKHFDEOHVDJHIIHFW7KHJHQHUDOL]HGGLVSODFHPHQWFRQWUROPHWKRG*'&
LVXVHGWRVROYHWKHVWDWLFDQDO\VLVDQGDWLPHKLVWRU\G\QDPLFDQDO\VLVLVHP
SOR\HG IRUVHLVPLFDQDO\VLV 6RPHH[DPSOHVRIVWHHOFDEOHVWD\HG EULGJHVVXE
MHFWHGWRERWKVWDWLFDQGVHLVPLFORDGVDUHVWXGLHG

.H\ZRUGV&DEOHVWD\HGEULGJHDGYDQFHGDQDO\VLVEHDPFROXPQFDWHQDU\

 ,QWURGXFWLRQ

,Q UHFHQW GHFDGHV FDEOHVWD\HG EULGJHV KDYH JDLQHG PXFK SRSXODULW\ RZLQJ WR WKHLU
HVWKHWLF DSSHDUDQFH HIILFLHQW XWLOL]DWLRQ RI VWUXFWXUDO PDWHULDOV DQG WHFKQLFDO DG
YDQWDJHV:KHQWKHPDLQVSDQOHQJWKEHFRPHVORQJHUPRUHDFFXUDWHDQDO\VLVPHWKRGV
DUHUHTXLUHGWRSUHGLFWWKHUHDOLVWLFEHKDYLRUVRIVWUXFWXUHVHVSHFLDOO\LQGHVLJQRIVWHHO
FDEOHVWD\HGEULGJHVZKHUHQRQOLQHDUEHKDYLRUVVXFKDVVWUXFWXUDOPDWHULDODQGJHRPHW
ULFQRQOLQHDULWLHVDUHFRQVLGHUDEOHUHIHU>@>@,QDVWHHOFDEOHVWD\HGEULGJHWKHPD
WHULDOQRQOLQHDULW\UHVXOWVIURPVWUHVV±VWUDLQEHKDYLRURIWKHVWHHOPDWHULDOZKHUHDVWKH
JHRPHWULFQRQOLQHDULWLHVFRPHIURPWKHVDJHIIHFWRIVWD\FDEOHVDQGD[LDOIRUFH±EHQG
LQJ PRPHQW LQWHUDFWLRQ DQG ODUJH GLVSODFHPHQW RI VWHHO S\ORQV JLUGHUV DQG FURVV
EHDPV
© Springer Nature Singapore Pte Ltd. 2020 105
C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_13
106 V.-H. Truong and S.-E. Kim

2YHUWKHODVWWKUHHGHFDGHVSUDFWLFDODGYDQFHGDQDO\VLV3$$PHWKRGVKDYHEHHQ

GHYRWHGVLQFHWKH\HIILFLHQWO\FDSWXUHWKHQRQOLQHDUEHKDYLRUVRIVWUXFWXUHVVHH>@>@
DPRQJRWKHUV:KHQXVLQJ3$$PHWKRGVWKHEHKDYLRUVRIVWHHOS\ORQVJLUGHUVDQG
FURVVEHDPVRIDVWHHOFDEOHVWD\HGEULGJHFDQEHPRGHOHGE\XVLQJWKHEHDPFROXPQ
DSSURDFKZLWKRQO\RQHRUWZRHOHPHQWVQHHGHGIRURQHVWUXFWXUDOPHPEHUVRWKHFRP
SXWDWLRQDOWLPHLVFRQVLGHUDEO\GHFUHDVHG%DVHGRQWKHEHDPFROXPQDSSURDFKVHYHUDO
FRPSXWHUSURJUDPVIRUQRQOLQHDULQHODVWLFDQDO\VLVRIVWHHOVWUXFWXUHVKDYHEHHQGHYHO
RSHGVXFKDV2SHQ6HHV>@'5$,1';>@DQG)5$0('>@+RZHYHUWKH3į
HIIHFWLVQRWFRQVLGHUHGLQWKRVHSURJUDPVKHQFHWKHUHVXOWVDUHXQUHOLDEOH
,QWKLVZRUNZHSUHVHQWDQLQQRYDWLYHVRIWZDUHQDPHGDV3$$3'EDVHGRQWKH
EHDPFROXPQ DSSURDFK IRUQRQOLQHDULQHODVWLFDQDO\VLV RI VWHHOFDEOHVWD\HG EULGJHV
7KHS\ORQVJLUGHUVDQGFURVVEHDPVDUHPRGHOHGDVSODVWLFKLQJHEHDPFROXPQHOH
PHQWVZKHUH3¨DQG3įHIIHFWVWKHLQLWLDOJHRPHWULFLPSHUIHFWLRQDQGWKHJUDGXDO
VWLIIQHVVGHJUDGDWLRQDUHFRQVLGHUHG7KHVWD\FDEOHVDUHPRGHOHGDVFDWHQDU\HOHPHQWV
WRFDSWXUHWKHFDEOHVDJHIIHFW7KHJHQHUDOL]HGGLVSODFHPHQWFRQWUROPHWKRG*'&LV
XVHGWRVROYHWKHVWDWLFDQDO\VLVDQGDWLPHKLVWRU\G\QDPLFDQDO\VLVLVHPSOR\HGIRU
VHLVPLFDQDO\VLV$XVHUIULHQGO\DQGHDV\WRXVHLQWHUIDFHLVGHYHORSHGWRIDFLOLWDWHWKH
PRGHOLQJSURFHVVDQGWKHSUHVHQWDWLRQRIDQDO\VLVUHVXOWV7KHVHPLKDUSW\SHRIDFD
EOHVWD\HGEULGJHVXEMHFWHGWRERWKVWDWLFDQGVHLVPLFORDGVLVVWXGLHGWRGHPRQVWUDWH
WKHHIILFLHQF\RIWKHGHYHORSHGVRIWZDUH

 1RQOLQHDU,QHODVWLF$QDO\VLVRI&DEOH6WD\HG%ULGJHV

)RUVWHHOFDEOHVWD\HGEULGJHDQDO\VLVWZRQRQOLQHDUHOHPHQWVVXFKDVWKHFDWHQDU\
DQGSODVWLFKLQJHEHDPFROXPQHOHPHQWVDUHLPSOHPHQWHGLQ3$$3',QWKHFDWHQDU\
HOHPHQWPRGHODVWD\FDEOHLVPRGHOHGDVDQHODVWLFFDWHQDU\HOHPHQWWKDWLVDVVXPHG
WREHSHUIHFWO\IOH[LEOHZLWKWKHVHOIZHLJKWGLVWULEXWHGDORQJLWVOHQJWK%DVHGRQWKH
H[DFWDQDO\WLFDOH[SUHVVLRQVRIWKHHODVWLFFDWHQDU\WKHUHDOLVWLFEHKDYLRUVRIVWD\FDEOH
PHPEHUVDUHFDSWXUHG,QWKHSODVWLFKLQJHEHDPFROXPQHOHPHQWWKHHOHPHQWUHPDLQV
HODVWLFH[FHSWDWLWVHQGVZKHUH]HUROHQJWKSODVWLFKLQJHVIRUP7KHVWDELOLW\IXQFWLRQV
SURSRVHGE\&KHQDQG/XL>@DUHXVHGWRFDSWXUHWKH 3  G DQG 3  ' HIIHFWV7KH
FROXPQUHVHDUFKFRXQFLO&5&WDQJHQWPRGXOXVFRQFHSW>@LVHPSOR\HGWRFRQVLGHU
LQLWLDOJHRPHWULFLPSHUIHFWLRQDQGUHVLGXDOVWUHVVHV7KH\LHOGVXUIDFHSURSRVHGE\2U
ELVRQHWDO>@LVDOVRXVHGWRFDSWXUHWKHSDUWLDOSODVWLILFDWLRQHIIHFWVRISODVWLFKLQJHV
LQWKHFDVHRIVPDOOD[LDOIRUFHDQGODUJHEHQGLQJPRPHQWV*HQHUDOO\RQHRUWZRHOH
PHQWVDUHHQRXJKIRUPRGHOOLQJDVWUXFWXUDOPHPEHUVRFRPSXWDWLRQDOWLPHLVVLJQLIL
FDQWO\ UHGXFHG'HWDLO RI WKH VWUXFWXUDO DIRUHPHQWLRQHG QRQOLQHDU LQHODVWLF EHKDYLRUV
DQGWKHFDWHQDU\DQGSODVWLFKLQJHEHDPFROXPQHOHPHQWVFDQEHIRXQGLQ>@
7KHVROXWLRQSURFHGXUHLPSOHPHQWHGLQ3$$3'IRUVWHHOFDEOHVWD\HGEULGJHVXQ
GHUVHLVPLFORDGLQJLVDWZRVWDJHDOJRULWKPZKHUHDVWDWLFDQDO\VLVLVFDUULHGRXWILUVW
IRUSUHGLFWLQJQRQOLQHDULQHODVWLFEHKDYLRUVRIWKHVWUXFWXUHXQGHUVWDWLFORDGVDQGWKHQ
IURPWKHVWUXFWXUDOGHIRUPDWLRQLQWKHVWDWLFDQDO\VLVDWLPHKLVWRU\G\QDPLFDQDO\VLVLV
SHUIRUPHG
)RUVWDWLFDQDO\VLV*'&PHWKRGSURSRVHGE\<DQJDQG6KLHK>@LVXVHGLQWKLV
ZRUNVLQFHLWVDGYDQWDJHVDUHWKHVWHSVL]HLVDXWRPDWLFDOO\DGMXVWHGLWFDQVHOI
Advanced Design Software for Steel Cable-Stayed Bridges … 107

DGDSWWRWKHFKDQJHRIWKHORDGLQJGLUHFWLRQDQGWKHDQDO\VLVLVUHODWLYHO\VWDEOHDW
WKHFULWLFDOSRLQWV'HWDLORIWKHQRQOLQHDUVROXWLRQSURFHGXUHIRUVWDWLFDQDO\VLVXVLQJ
*'&FDQEHIRXQGLQ>@
)RUWKHWLPHKLVWRU\G\QDPLFDQDO\VLVWKHGLUHFWLQWHJUDWLRQPHWKRGXVLQJ1HZPDUN
IDPLO\ PHWKRG >@ LV XVHG IRU VROYLQJ WKH HTXDWLRQ RI PRWLRQ 1HZWRQ5DSKVRQ
PHWKRGLVDOVRHPSOR\HGWRHOLPLQDWHWKHUHVLGXDOIRUFHVLQWLPHVWHSV'HWDLOVRQWKH
DIRUHPHQWLRQHGWLPHKLVWRU\G\QDPLFDQDO\VLVFDQEHIRXQGLQ>@

 ,QWHUDFWLYH*UDSKLFRI3$$3'

7KHLQWHUIDFH RIWKH 3$$3' VRIWZDUH LV GHVLJQHG YHU\ IULHQGO\ ZLWK WKUHH PDLQ
SDUWV SUHSURFHVVRU VROYHU DQG SRVWSURFHVVRU 7KH VROYHU LV FRGHG E\ XVLQJ WKH
)2575$1SURJUDPPLQJODQJXDJHZKLOHWKHSUHDQGSRVWSURFHVVRUVDUHGHYHORSHG
XVLQJ9LVXDO&SURJUDPPLQJODQJXDJH
7KHPDLQJUDSKLFXVHULQWHUIDFHRIWKH3$$3'VRIWZDUHLVSUHVHQWHGLQ)LJ)RXU
EDVLF PHQXV DUH SURYLGHG VXFK DV PDLQ WUHH LFRQ DQG UHVXOW PHQXV 7KH PRGHO RI
VWUXFWXUHVFDQEHVKRZQLQWKHSODQHRU'SHUVSHFWLYHYLHZVZKLOHWZRYLHZZLQGRZV
FDQEHXVHGDWVDPHWLPH7KHVWUXFWXUDOPRGHOFDQEHEXLOWE\UHDGLQJIURPDIRUPDWWHG
LQSXWILOHRUGLUHFWO\PRGHOLQJLQWKHSUHSURFHVVRU


)LJ0DLQJUDSKLFXVHULQWHUIDFHRI3$$3'

7KHVWUXFWXUHLVVROYHGE\XVLQJWKHDQDO\VLVPHQX7KH3$$3'VRIWZDUHSURYLGHV
ERWKVWDWLFDQGVHLVPLFDQDO\VHVRIVWUXFWXUHV7KHUHVXOWVDUHWKHQSUHVHQWHGLQWKHYLHZ
ZLQGRZXVLQJWKHUHVXOWPHQXDVVKRZQLQ )LJ$VFDQ EHVHHQLQWKLVILJXUHWKH
DQDO\VLVUHVXOWVRIWKHQRGDOGLVSODFHPHQWDQGUHDFWLRQIRUFHHOHPHQWIRUFHKLQJHDQG
EXFNOLQJ FDQ EH GLVSOD\HG LQ WKH SRVWSURFHVVRU 7KH RSWLRQ IRU DQLPDWLQJ VWUXFWXUH
GHIRUPDWLRQLVDOVRSURYLGHG
108 V.-H. Truong and S.-E. Kim


)LJ2XWSXWRIDQDO\VLVUHVXOWVLQ3$$3'

 &DVH6WXG\

7KHKDUSW\SHRIDVWHHOFDEOHVWD\HGEULGJHLQ)LJLVVWXGLHG7KHS\ORQVJLUGHUV
DQGFURVVEHDPVDUHVWHHOZLWKHODVWLFPRGXOXVRI*3D\LHOGVWUHVVRI03D
DQGZHLJKWSHUXQLWYROXPHRIN1P7KHZHLJKWSHUXQLWYROXPHRIFDEOHVLV
HTXDOWRN1P7KHHODVWLFPRGXOXV\LHOGVWUHVVDQGXOWLPDWHWHQVLOHVWUHQJWKRI
FDEOHVDUH*3D03DDQG03DUHVSHFWLYHO\7KHGHDGORDGLQFOXGLQJ
WKHGHFNZHDULQJVXUIDFHDQGWUDIILFEDUULHUVLVDSSOLHGDVN1PZKLOHWKHOLYH
ORDGLVDVVXPHGWREHWKHGLVWULEXWHGORDGDQGHTXDOWRN1PODQH>@
6(0,+$537<3(

                   

*LUGHU #  #  #  # 

<                                       


;
                                      
/HIW3\ORQV 5LJKW3\ORQV
    G 
  #

  #

   
   
   
   
=
<
'LDPHWHURIVWD\FDEOHV
   

   
 
W 
W 
6HFWLRQV




  :[

7RZHU *LUGHU &URVVEUHDP
/,9(/2$'&$6(// .1P


)LJ6HPLKDUSW\SHRIVWHHOFDEOHVWD\HGEULGJH
,QWKHILUVWFDVHVWXG\WKHVWDWLFDQDO\VLVRIWKHEULGJHXQGHUWKHORDGFRPELQDWLRQ
³6WUHQJWK,´'///VSHFLILHGLQ$$6+72/5)'>@LVFRQVLGHUHG7KH
Advanced Design Software for Steel Cable-Stayed Bridges … 109

XOWLPDWHORDG IDFWRU RI WKH EULGJH LV  )LJ VKRZV WKHWRWDOWUDQVODWLRQ RI WKH
EULGJHXVLQJ3$$3'


)LJ7RWDOWUDQVODWLRQRIWKHEULGJHDWXOWLPDWHORDGIDFWRU


)LJ7LPHKLVWRU\YHUWLFDOGLVSODFHPHQWDWWKHPLGVSDQ
,QWKHVHFRQGFDVHVWXG\WKHWLPHKLVWRU\G\QDPLFDQDO\VLVRIWKHEULGJHXQGHU(O
&HQWUR HDUWKTXDNH LV SUHVHQWHG 7KH VWDWLF ORDGV LQ WKLV FDVH VWXG\ LV '/ 7KLV
PHDQVWKDWRQO\GHDGORDGLVFRQVLGHUHG7KH(O&HQWURHDUWKTXDNHUHFRUGLVWDNHQIURP
WKHHDUWKTXDNHUHFRUGOLEUDU\RIWKH3$$3'VRIWZDUH1RWHVWKDWWKHUHDUHPDQ\HDUWK
TXDNHUHFRUGVSURYLGHGLQWKH3$$3'OLEUDU\VXFKDV,PSHULD9DOOH\.REH.RFDHOL
/RPD3ULHWDHWF7KHXVHUVDOVRFDQLQSXWWKHHDUWKTXDNHUHFRUGE\WKHPVHOYHV)LJ
SUHVHQWVWKHWLPHKLVWRU\YHUWLFDOGLVSODFHPHQWDWWKHPLGVSDQRIWKHEULGJH
110 V.-H. Truong and S.-E. Kim

 &RQFOXVLRQ
7KHIROORZLQJSRLQWVFDQEHVXPPDUL]HGIURPWKLVZRUN
¾$UREXVWVRIWZDUHLVSURSRVHGIRUQRQOLQHDULQHODVWLFDQDO\VLVRIVWHHOFDEOHVWD\HG
EULGJHVZKHUHWKHVROYHULVZULWWHQLQ)2575$1DQGWKHSUHDQGSRVWSURFHVVRUV
DUHGHYHORSHGXVLQJ9LVXDO&
¾%RWKVWDWLFDQGVHLVPLFDQDO\VHVDUHSURYLGHGLQWKHSURSRVHGVRIWZDUH
¾7KHUHVXOWVRIWKHVHPLKDUSW\SHRIWKHVWHHOFDEOHVWD\HGEULGJHVKRZWKDWWKHSUR
SRVHGVRIWZDUHFDQEHDSSOLHGLQSUDFWLFDOGHVLJQRIVWHHOFDEOHVWD\HGEULGJHV

$FNQRZOHGJHPHQW
7KLVUHVHDUFKLVIXQGHGE\9LHWQDP1DWLRQDO)RXQGDWLRQIRU6FLHQFHDQG7HFKQRORJ\
'HYHORSPHQW1$)267('XQGHUJUDQWQXPEHU

5HIHUHQFHV
 7KDL +7 .LP 6( 6HFRQGRUGHU LQHODVWLF RI FDEOHVWD\HG EULGJHV - )LQLWH (OHPHQWV LQ
DQDO\VLVDQGGHVLJQ
 7UXRQJ 9+ .LP 6( $Q HIILFLHQW PHWKRG RI V\VWHP UHOLDELOLW\ DQDO\VLV RI VWHHO FDEOH
VWD\HGEULGJHV$GYDQFHVLQ(QJLQHHULQJ6RIWZDUH
 7UXRQJ9+.LP6($UREXVWPHWKRGIRURSWLPL]DWLRQRIVHPLULJLGVWHHOIUDPHVVXEMHFW
WRVHLVPLFORDGLQJ-RXUQDORI&RQVWUXFWLRQDO6WHHO5HVHDUFK
 7UXRQJ9+.LP6($QHIILFLHQWPHWKRGIRUUHOLDELOLW\EDVHGGHVLJQRSWLPL]DWLRQRIQRQ
OLQHDULQHODVWLFVWHHOVSDFHIUDPHV6WUXFWXUDODQG0XOWLGLVFLSOLQDU\2SWLPL]DWLRQ

 +D 0+ 9X 4$ 7UXRQJ 9+ 2SWLPXP 'HVLJQ RI 6WD\ &DEOHV RI 6WHHO &DEOHVWD\HG
%ULGJHV8VLQJ1RQOLQHDU,QHODVWLF$QDO\VLVDQG*HQHWLF$OJRULWKP6WUXFWXUHV

 0F.HQQD))HQYHV*/HWDO2SHQV\VWHPIRUHDUWKTXDNHHQJLQHHULQJVLPXODWLRQ2SHQ
6HHV$YDLODEOHIURPKWWSRSHQVHHVEHUNHOH\HGX!3DFLILF(DUWKTXDNH(QJLQHHULQJ5H
VHDUFK&HQWHU8QLYHUVLW\RI&DOLIRUQLDDW%HUNHOH\ODVWDFFHVVHG
 3UDNDVK93RZHOO*+'5$,1';%DVHSURJUDPXVHUJXLGHYHUVLRQ$FRPSXWHU
SURJUDP GLVWULEXWHG E\ 1,6((&RPSXWHU DSSOLFDWLRQV 'HSDUWPHQW RI &LYLO (QJLQHHULQJ
8QLYHUVLW\RI&DOLIRUQLDDW%HUNHOH\
 .ULVKQDQ6+DOO-)0RGHOLQJVWHHOIUDPHEXLOGLQJVLQWKUHHGLPHQVLRQV,, (ODVWRILEHU
EHDP HOHPHQW DQG H[DPSOHV -RXUQDO RI (QJLQHHULQJ 0HFKDQLFV $6&(  

 &KHQ:)/XL(06WUXFWXUDO6WDELOLW\7KHRU\DQG,PSOHPHQWDWLRQ(OVHYLHU$PVWHUGDP

 &KHQ:)/XL(06WDELOLW\'HVLJQRI6WHHO)UDPHV&5&3UHVV%RFD5DWRQ)/
 2UELVRQ-*0F*XLUH:$EHO-)<LHOGVXUIDFHDSSOLFDWLRQVLQQRQOLQHDUVWHHOIUDPHDQDO
\VLV&RPSXWHU0HWKRGVLQ$SSOLHG0HFKDQLFVDQG(QJLQHHULQJ±
 <DQJ<%6KLHK066ROXWLRQPHWKRGIRUQRQOLQHDUSUREOHPVZLWKPXOWLSOHFULWLFDOSRLQWV
$,$$-±
 1HZPDUN10$PHWKRGRIFRPSXWDWLRQIRUVWUXFWXUDOG\QDPLFV-RXUQDORIWKH(QJLQHHU
LQJ0HFKDQLFV'LYLVLRQ±
 $$6+72/5)'%ULGJH'HVLJQ6SHFLILFDWLRQVWK(GLWLRQ
 Blind source separation technique for operational modal
analysis in presence of harmonic excitation

Van-Dong DO1, Thien-Phu LE2, Alexis BEAKOU1

1
Université Clermont Auvergne, CNRS, SIGMA Clermont, Institut Pascal, F-63000 Clermont-
Ferrand, France
{van_dong.do,alexis.beakou}@sigma-clermont.fr
2
LMEE, Univ Evry, Université Paris-Saclay, 91020 Evry cedex, France
thienphu.le@univ-evry.fr

Abstract. Operational modal analysis (OMA) known as response-only modal

identification, is extremely useful for large structures in civil engineering where
the excitation is difficult or even impossible to measure. The unknown excitation
is always assumed as white noise in OMA. In presence of harmonics, the white
noise assumption is not verified that makes the modal identification process dif-
ficult, and eventually leading to biased results. In recent years, blind source sep-
aration (BSS) techniques have been shown to be robust and efficient for OMA.
In this paper, a new BBS technique termed Sparse Component Analysis (SCA)
without white noise assumption, is applied for operational modal identification
in presence of harmonics. The performance of the proposed method is demon-
strated using numerical examples of a two-degrees-of-freedom system and a can-
tilever beam.
Keywords: Operational modal analysis, Harmonic component, SCA

1 INTRODUCTION

The dynamic behavior of a linear structure can be characterized by modal parameters

involving natural frequencies, damping ratios and mode shapes. Classically, modal pa-
rameters are extracted from an Experimental Modal Analysis (EMA) where both exci-
tation artificially generated by a hammer or a shaker, and its structural responses are
measured. For real structures, Operational Modal Analysis (OMA) which requires
structural responses only is more adequate because of several advantages: simple equip-
ment thus low cost, continuous use and real boundary conditions. However, the uncon-
trolled and unmeasured excitation of natural form such as wind, traffic, operational
loadings« is always assumed as white noise in OMA [1].
In presence of harmonics on excitation, for instance, structures having rotating com-
ponents such as buildings with fans/air-conditioners, gravity dams comprising turbines
and generators, the white noise assumption is not verified. Consequently, the harmonic
components cohabit with structural modes and that makes the modal identification pro-
cess difficult, even leading to biased results.

© Springer Nature Singapore Pte Ltd. 2020 111

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_14
112 V.-D. Do et al.

Blind Source Separation (BSS) technique is known as a statistical signal processing

tool that has been widely used to extract sources signals from the output signals only
without requiring explicit information about the source signals or mixing process. In-
dependent Component Analysis (ICA) and Second-Order Blind Identification (SOBI)
are two popular methods of BBS family for OMA. In addition of an assumption about
independent/uncorrelated sources, these methods still rely on white noise assumption
[2-3] and are only applicable for overdetermined cases [4]. In presence of harmonics,
their feasibility is thus limited. In this paper, we propose a new BSS method called
Sparse Component Analysis (SCA) applicable for underdetermined problems. Its effi-
ciency is verified for OMA in presence of harmonic excitation.

2 MODAL IDENTIFICATION BASED ON BLIND SOURCE

SEPARATION

The BSS problem coincides with the operational modal identification issue. On the
one hand, a linear instantaneous-mixing model is formulated as:
n
x(t ) As (t ) ¦ a s (t )
i 1
i i
(1)

The aim of BBS is to extract both the n original sources s (t ) and the mixing matrix
A  ƒmun from the m measured mixtures x(t ) . Depending on the number of measured
mixtures compared to the number of original sources, BBS is classified into three sorts,
overdetermined (m ! n) , determined (m n) and underdertermined ( m  n) cases.
On the other hand, for an n-degree-of-freedom (DoF) linear system, the system re-
sponses x(t ) can be expressed through modal expansion [3]:
n
x(t ) )q(t ) ¦ M q (t )
i i
(2)
i 1

where )  ƒnun is the square real-valued modal matrix and each column M i repre-
sents a mode shape; qi (t )  ƒn is the modal responses. It notes that there is a similarity
between the BBS model given in Eq. (1) and the modal expansion in Eq. (2). Therefore,
) and q(t ) can be recovered using BBS technique to extract modal parameters.
In presence of harmonics on excitation, there are additional components in measured
responses because of non-damped harmonic parts. These extra components can be con-
sidered as non-damped virtual modes of the system [5]. Therefore, ) and q(t ) in Eq.(2)
should be recovered by underdertermined BBS technique when the number of sources
q(t ) is more than number of measured responses x(t ) .
The theory of SCA and its application for OMA can be found in Ref [6] and [7].
Unlike traditional BSS techniques based on source independence, SCA requires sources
Blind source separation technique for operational modal … 113

to have sparsity in transformed domain. SCA comprises two main stages: mixing matrix
estimation and source recovery.
(a) (b)

Fig. 1. Scatter plots of 2 mixtures (involving 3 sources) in (a) time domain; (b) TF domain

Mixing matrix estimation. In order to sparsify the measured signals, the Short Time
Fourier Transform (STFT) was applied to Eq. (1) or Eq. (2) to obtain Eq. (3)

X (t , f ) AS (t , f ) (3)

Fig.1 shows the scatter plots of 2 mixtures including 3 latent sources. The measured
mixtures do not satisfy the assumption about sparsity in time domain (Fig.1.a), but they
have a sparse representation in time-frequency (TF) domain such as given in Fig.1.b.
Each straight line presents a latent source. The equation Eq. (3) becomes

­ S1 (t , f ) ½
­ X 1 (t , f ) ½ ª a11 a12 a13 º ° ° (4)
® ¾ « ®S 2 (t , f )¾
¯ X 2 (t , f )¿ ¬a21 a22 a23 »¼ ° °
¯ S 3 (t , f ) ¿

The key idea is that directions of the three straight lines correspond to the columns
of the mixing matrix A. Therefore, the mixing matrix A can be estimated by means of
a clustering algorithm from the scatter plot of the STFT coefficients X (t , f ) . In the first
main stage, K-means clustering algorithm is used to obtain the matrix A.
Source recovery. Once the mixing matrix A is estimated, in the second main stage,
the sources can be recovered by the l1-minimization sparse recovery technique [8] in
TF domain such as given in Eq. (5). The source recovery s (t ) can be obtained by trans-
forming S (t , f ) back to the time domain using an inverse STFT algorithm.

(P1): min S (t , f ) subject to AS (t , f ) X (t , f ) (5)

"1

Distinction between harmonic and structural components. s (t ) contains mono-

component sources that can be structural modes or harmonic components. Kurtosis
value and histogram [9] are used to distinguish and remove the sources s i (t ) corre-
sponding to harmonic excitation. For any mono-component, if its Kurtosis value is
equal to 3 and its histogram is in WKHIRUPRID³EHOO´this one corresponds to a structural
mode and if the Kurtosis value is equal to 1.5 and its histogram has two peaks, it is a
114 V.-D. Do et al.

harmonic component. The measured responses without harmonic components x R (t )

can be reconstructed as follows:
x Remove ith column ai and S i (t , f ) related to harmonic component in Eq. (3)
to obtain X R (t , f ) AR S R (t , f )
R
x Transform X R (t , f ) back to time domain to acquire x (t ) .
After removing the harmonic components, it is possible to use the Frequency Domain
Decomposition (FDD) method [10] to identify modal parameters and hence the name
SCA-FDD is chosen for the proposed method.

3 APPLICATIONS

A two DoFs system and a cantilever beam are used for numerical validation. The
two DoFs system is simulated with the mechanical properties given in Ref [11]. The
excitation force is generated by combining a harmonic excitation and a white-noise
excitation, located at second DoFs. The harmonic excitation has the form of
5 sin(2S .20.t ) and the Gaussian white-noise excitation has zero mean and unit standard
deviation. The details of a cantilever beam example (Figure 3.a) can be found in Ref
[9]. The developed SCA-FDD method is performed to identify modal parameters. The
signals are transformed into TF domain with Hanning window of 2048 and 75% over-
lapping. The identified modal parameters with their exact values in parentheses are re-
ported in Table 1.

Mode Frequency(Hz) Damping ratio (%) Mode shapes Kurtosis Conclusion

2 DoFs system

1.00(1.00)
1 10.25 (10.30) 0.32 (0.32) 3.09 (3.00) Structural
1.38 (1.39)
2 20.01 (20.00) - - 1.55 (1.50) Harmonic
1.00 (1.00)
3 30.11 (30.12) 0.95 (0.95) 3.11 (3.00) Structural
-0.72 (-0.72)
1.00 (1.00)
1 5.09 (5.15) 1.51 (1.50) 8.87 (8.87) 3.14 (3.00) Structural
19.23 (19.23)
Cantilever beam

2 7.08 (7.00) - - 1.61 (1.50) Harmonic

1.00 (1.00)
3 32.25 (32.28) 0.99 (1.00) 2.50 (2.48) 3.21 (3.00) Structural
-3.54 (-3.59)
1.00 (1.00)
4 91.04 (90.60) 0.48 (0.50) -0.80 (-0.81) 3.29 (3.00) Structural
1.56 (1.55)
Table 2. Identified modal parameters with SCA-FDD method.
For the two DoFs system, the scatter plot of two measured responses in TF domain
is given in Fig.2.a and three mono-component sources are given in Fig.2.c. The ob-
tained results show that the second source is harmonic. After removing the harmonic
Blind source separation technique for operational modal … 115

component, the reconstructed responses are shown in Fig.2.d. It can be seen that the
harmonic component of 20 Hz is quite eliminated from the original of measured re-
sponses such as given in Fig.2.b. It is noted that the identified frequencies, damping
ratios and mode shapes are very close to the exact values.
(a) (b)

(c) (d)

Fig. 2. 2 DoFs system: (a) scatter plot in TF domain; (b) Power Spectral Density (PSD) of
measured responses x(t ) ; (c) recovery source; (d) reconstructed responses x R (t )
(a) (b)

Fig. 3. (a) Description of the cantilever beam and (b) scatter plot in TF domain

For a cantilever beam example, the scatter plot of three simulated responses in TF
domain is given in Fig.3.b. It can be seen clearly four straight lines representing four
sources including three structural modes and a harmonic component. Modal parameters
are well identified by the SCA-FDD method despite the harmonic frequency is close to
the first natural frequency as given in Table 1.
116 V.-D. Do et al.

4 CONCLUSIONS

In this paper, the new SCA-FDD method is proposed for operational modal identi-
fication. In the proposed method, the white noise assumption is not required and it can
thus be applied to perform OMA in presence of harmonic excitation. Furthermore, this
method is applicable to underdetermined cases when the number of sensors measure-
ment is limited. Potentially, the SCA-FFD procedure is suitable to requirements of au-
tomated operational modal identification.
The performance of the proposed method is verified by two numerical examples.
The obtained results show that this technique is robust and efficient even though the
harmonic excitation is dominant or is very close to the natural frequency.

Acknowledgments. This work is funded by the European Union and by the Au-
vergne-Rhône-Alpes region through the CPER 2015-2020 program. Europe is commit-
ted to Auvergne with the European Regional Development Fund (FEDER).

References

1. T.-P. Le, P. Paultre, B. Weber, J. Proulx, P. Argoul, Modal identification based on ambient
excitation tests, in : Proceedings of 24th IMAC conference, 2006
2. J. Antoni, S. Chauhan, A study and extension of second-order blind sources separation to
operational modal analysis, Journal of sound and Vibration 332, pp 1079-1106, 2013
3. G. Kerschen, F. Poncelet, J.-C Golinval, Physical interpretation of independent component
analysis in structural dynamics, Mechanical Systems and Signal Processing 21 pp 1561-
1575, 2007
4. A. Sadhu, S. Narasimhan, J.Antoni, A review of output-only structural mode identification
literature employing blind source separation methods, Mechanical Systems and Signal
Processing 94 pp 415-431, 2017
5. P. Mohanty, D.-J. Rixen, Operational modal analysis in the presence of harmonic excitation,
Journal of sound and Vibration 270, pp 93-109, 2004
6. P. Bofill, M. Zibulevsky, Undertemined blind source separation using sparse representions,
Signal Processing 81, pp 2353-2362, 2001
7. Y. Yang, S. Nagarajaiah, Output-only modal identification with limited sensors using sparse
component analysis, Journal of sound and Vibration 332, pp 4741-4765, 2013.
8. E. Cand, J. Romberg, L1-MAGIC: Recovery of Sparse Signals via Convex Programming,
pp 1-19, 2005
9. T.-P Le, P. Agroul, Distinction between harmonic and structural components in ambient
excitation tests using the time-frequency domain decomposition technique, Mechanical
Systems and Signal Processing 52-53 pp 29-45, 2015
10. R. Brincker, C.-E. Ventura, P. Andersen, Damping estimation by frequency domain
decomposition, in: Proceedings of 19th IMAC conference, 2001
11. V.-D. Do, T.-P. Le, A. Beakou, Transmissibility based operational modal analysis in
presence of harmonics, in: Congrès international de Géotechnique-Ouvrages-Structures,
Springer, pp 972-980, 2017.
 Investigation into the response variability of a higher-
order beam resting on a foundation using a stochastic
finite element method

Ta Duy Hien1*, Bui Tien Thanh1, Nguyen Ngoc Long1, Nguyen Van Thuan2, Do Thi
Hang1
1
Department of Civil Engineering, University of Transport and Communications, HaNoi, Vi-
etnam
2
Department of Transportation Engineering, Nha Trang University, Nha Trang, Vietnam

*
Corresponding author: tdhien@utc.udu.vn

Abstract. This study deals with the response variability of a beam resting on an
elastic foundation using a stochastic finite element method (SFEM). Use of
SFEM for a higher-order beam on an elastic foundation with a random field of
elastic modulus is proposed. This random field of elastic modulus is assumed to
be a one-dimensional Gaussian random field. A random field is discretized by a
weighted integration method to build the stochastic finite element formulation. A
numerical example is presented to validate the proposed formulation, and the re-
sults are compared with those obtained using Monte Carlo simulation. The re-
sponse variability of the beam and effect of the parameter of the random field are
investigated in detail.

Keywords: Higher-order beam, weighed integration method, SFEM, random field.

1 Introduction

In civil engineering, the use of beams has become popular. To obtain stress resultants
and displacement of beams, various beam theories have been used. Gorbachev el al. [1]
studied an anisotropic inhomogeneous beam using Bernoulli—Euler Theory. Tekili el
al. [2] found an analytical solution for dynamic responses of Bernoulli—Euler compo-
site beams subjected to moving loads. Calim Faruk Firat [3] used the complementary
functions method and Durbin’s algorithm to investigate the dynamic response of curved
Timoshenko beams resting on a viscoelastic foundation. Xiang and Kang [4] employ a
meshless method and Reddy’s third-order shear deformation theory for the problem
regarding the static behaviour of laminated composite beams.

In recent years, SFEM has been a topic of active research [5-8]. For finite element im-
plementation, it is necessary to discretize such fields into random vector representa-
tions. Two methods of discretization are investigated — the field value over an element
at the midpoint of the element and the spatial average of the process over the element

© Springer Nature Singapore Pte Ltd. 2020 117

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_15
118 T. D. Hien et al.

[9, 10]. Hyuk Chun Noh [11] developed an SFEM using the weighted integral approach
to determine the response variability of in-plane and plate structures with multiple un-
certain elastic modulus and Poisson’s ratio. T.D. Hien et al. [12] computed the varia-
bility of displacements of a beam subjected to a moving load with various random pa-
rameters. N.X Hoang et al. [13] and T.D. Hien et al. [14] used geometric analysis to
compute the eigenvalue problem of composite structures with uncertain material prop-
erties.
In this work, the stochastic finite element for bending of beams with uncertain elastic
modulus is developed using higher-order shear-deformable beam theory in conjugation
with a weighed integration technique. The random field of elastic modulus is modelled
as a Gaussian random field. Numerical examples are employed in detail.

2 Finite element formulation

2.1 A finite element model for a higher-order shear-deformable beam theory

We consider a beam resting on an elastic foundation as in Fig.1:

kF
L
Fig. 1. Beam resting on elastic foundation.

It is assumed that interaction between beam and foundation comply with Winkler's hy-
pothesis. The displacement fields using Reddy's beam theory [15, 16] are calculated as:
⎛ 4 ⎞⎛ ∂w ⎞
u ( x, z ) = zψ ( x ) + z 3 ⎜ − 2 ⎟ ⎜ψ ( x ) + 0 ⎟
⎝ 3h ⎠ ⎝ ∂x ⎠ (1)
w ( x, z ) = w0 ( x )
The strain expression is:

∂ψ ⎛ 4 z 3 ⎞ 4 z 3 ∂ 2 w0
εx = ⎜z− 2 ⎟− 2
∂x ⎝ 3h ⎠ 3h ∂x 2
(2)
⎛ 4 z 2 ⎞ ∂w ⎛ 4 z 2 ⎞
γ xz = ψ ⎜1 − 2 ⎟ + 0 ⎜1 − 2 ⎟
⎝ h ⎠ ∂x ⎝ h ⎠
We use Hermite and Lagrange interpolation functions for approximation displace-
ment fields as follows:
⎧⎪ w0 ( x ) = N1 w1 + N 2θ1 + N3 w2 + N 4θ 2
⎨ (3)
⎪⎩ψ ( x ) = N1 ψ 1 + N 2 ψ 2
L L
Investigation into the response variability of a higherorder … 119

where:
⎧ x2 x3 x2 x3
N
⎪ 1 = 1 − 3 + 2 N 3 = 3 − 2
⎪ L2 L3 L2 L3
⎪ ⎛ x x2 ⎞ ⎛ x x2 ⎞
⎨ 2
N = x ⎜ 1 − 2 + 2 ⎟
N 4 = x ⎜− + 2 ⎟ (4)
⎪ ⎝ L L ⎠ ⎝ L L ⎠
⎪ L x x
⎪ N1 = 1 − N 2L = 1 −
⎩ L L
{q}e = {ψ 1 w2 θ 2 }
T
w1 θ1 ψ 2 (5)
The relationship strain-stress matrix is provided by:
⎧⎪ ε x ⎫⎪
⎨ ⎬ = [ B ]{q}e (6)
⎪⎩γ xy ⎪⎭e
The strain energy expression for bending is given as follows:
Le
⎧
⎪ 1 ⎡ ∂φx 4 3 ⎛ ∂ψ ∂ 2 w0 ⎞ ⎤ 1 ⎡⎛
2
4 ⎞⎛ ∂w0 ⎞ ⎤ ⎫
2
⎪
Ue = ∫ ∫ ⎨ E ⎢ y − 2 y ⎜ + 2 ⎟⎥
+ G ⎢⎜1 − 2 ⎟ ⎜ψ + ⎟ ⎥ ⎬dx
0 A 2
⎪ ⎣⎢ ∂x 3 h ⎝ ∂ x ∂x ⎠ ⎦⎥ 2 ⎣⎢⎝ h ⎠⎝ ∂x ⎠ ⎦⎥ ⎪
⎩ ⎭
(7)
The potential energy of a foundation is provided as follows:
2
1 Le
U F = ∫ kF ⎡⎣ w0 ( x )⎤⎦ dx (8)
2 0

2.2 Stochastic finite element method

In general, we assume that spatially random fields include a fluctuation part and a mean:

E ( x ) = E0 ⎡⎣1 + f ( x ) ⎤⎦ (9)
where f ( x ) is a one-dimensional Gaussian random field with mean equal to zero.
The element stiffness matrix includes a random field f ( x ) :

[k ]( ) = ∫V [ B]i [ D][ B] dV
e T
e
(10)

To apply the weighted integral method, the element stiffness matrix can be written as
both deterministic and stochastic parts:
[ k ]( ) = [ k ](0 ) + Δ [ k ]( )
e e e

(11)
(e)
Δ [k ] = ∫ e f ( x ) [ B ]i [ D ][ B ] j dV
T
V

The deviatoric stiffness matrix Δ [ k ] in Eq. (11) can be written as follows:

120 T. D. Hien et al.

nr
(e) (e)
Δ [k ] = ∑ [C ]r Yr( e ) (12)
r =1

Random variables Yr is represented by integration:

Yr( ) = ∫ x r f ( x )dx
e
(13)
Ωe

The solution can be found as follows:

[ K ]0 {U }0 = {F }
∂ {U } −1 ∂ [ K ]
(14)
(k )
= − [ K ]0 ( k ) { }0
U
∂dYi dYi

3 Numerical example

In this paper, the responses obtained by the weighted integration method are checked
by Monte Carlo simulation. We generate many different sample functions to describe
the fluctuation of the random field f ( x ) in equation (9) using a spectral representation
method [17, 18]. For convenience, we use the normalized stiffness of an elastic foun-
dation as follows:
k F L3
kf = (15)
EI
The auto-correlation functions for the respective random fields f ( x ) are assumed as
follows:
⎛ ξ ⎞
R (ξ ) = σ 2 exp ⎜ − ⎟ (16)
⎝ d ⎠
where R(ξ), σ, d are the auto-correlation function, coefficient of variation (COV) and
correlation distance for the random field of an elastic modulus.
We consider a simply supported rectangular beam subjected to uniform load q
=1000N/m. The length of beam L=5m and cross section: b=0.25m, h=0.4m. It is as-
sumed that the material properties are E0 =30GPa, and ν=0.3 and coefficient of varia-
tion of a random field of elastic modulus σ=0.1. The normalized stiffness of the elastic
foundation assumes kf =2 in this example.
The variability of response COV of displacement at the middle beam is defined as fol-
lows:

COV =
( )
Var U L
2
(17)
E (U )
L
2

( ) ( )
where E U L ,Var U L are expected value and variance of displacement at the mid-
2 2

dle beam.
Investigation into the response variability of a higherorder … 121

There is good agreement between the COV predicted by Monte Carlo simulation and
weighed integral method in Fig. 2. When the ratio correlation distance to length of beam
d/L is 10, the COV predicted by the weighed integral method is the asymptotic maxi-
mum value and is smaller than the coefficient of variation of a random field in elastic
modulus σ. This maximum COV depends on the proportion stiffness of the foundation
to the gross stiffness of the beam; in detail, if the stiffness of the foundation increases,
the maximum COV will decrease.

Fig. 2. Variability response COV as ratio d/L

4 Conclusions

The stochastic finite element method is proposed for bending of a high-order beam
resting on an elastic foundation with randomness in elastic modulus. The proposed sto-
chastic method is validated by comparing the response variability COV of deflection at
the middle beam predicted by the proposal method using a Monte Carlo approach. The
numerical examples clearly demonstrate that the randomness of elastic modulus signif-
icantly affects the structural responses of the beam. The effect of correlation distance
on variability response COV of displacement is clear — COV increases when correla-
tion distances also increase.

Acknowledgements

This research is funded by University of Transport and Communications (UTC) un-

der grant number T2019-CT-042.
122 T. D. Hien et al.

References

1. V. I. Gorbachev, T. M. Mel’nik. Formulation of Problems in the Bernoulli—Euler Theory of

Anisotropic Inhomogeneous Beams. Moscow University Mechanics Bulletin.73, 18-26 (2018).
2. S. Tekili, Y. Khadri, B. Merzoug, E. M. Daya, A. Daouadji. Free and Forced Vibration of
Beams Strengthened by Composite Coats Subjected to Moving Loads. Mechanics of Composite
Materials.52, 789-98 (2017).
3. F. F. Calim. Dynamic response of curved Timoshenko beams resting on viscoelastic
foundation. Structural engineering and mechanics : An international journal.59, 761-74 (2016).
4. S. Xiang, G. W. Kang. Meshless Solution of the Problem on the Static Behavior of Thin and
Thick Laminated Composite Beams. Mechanics of Composite Materials.54, 89-98 (2018).
5. G. Stefanou. The stochastic finite element method: Past, present and future. Computer Methods
in Applied Mechanics and Engineering.198, 1031-51 (2009).
6. W. K. Liu, T. Belytschko, A. Mani. Random field finite elements. International Journal for
Numerical Methods in Engineering.23, 1831-45 (1986).
7. Michael Kleiber, T. D. Hien. The stochastic finite element method : basic perturbation
technique and computer implementation: Wiley,(1992).
8. M. Kaminski. The Stochastic Perturbation Method for Computational Mechanics:
WIKEY,(2013).
9. A. Der Kiureghian, J.-B. Ke. The stochastic finite element method in structural reliability.
Probabilistic Engineering Mechanics.3, 83-91 (1988).
10. E. Vanmarcke, M. Grigoriu. Stochastic Finite Element Analysis of Simple Beams. Journal of
Engineering Mechanics.109, 1203-14 (1983).
11. H. C. Noh. Effect of multiple uncertain material properties on the response variability of in-
plane and plate structures. Computer Methods in Applied Mechanics and Engineering.195,
2697-718 (2006).
12. T. D. Hien, N. D. Hung, N. T. Kien, H. C. Noh. The variability of dynamic responses of
beams resting on elastic foundation subjected to vehicle with random system parameters. Applied
Mathematical Modelling.67, 676-87 (2019).
13. H. X. Nguyen, T. Duy Hien, J. Lee, H. Nguyen-Xuan. Stochastic buckling behaviour of
laminated composite structures with uncertain material properties. Aerospace Science and
Technology.66, 274-83 (2017).
14. T. D. Hien, B. T. Thanh, N. T. Quynh Giang. Uncertainty qualification for the free vibration
of a functionally graded material plate with uncertain mass density. IOP Conference Series: Earth
and Environmental Science.143, 012021 (2018).
15. C. M. Wang, J. N. Reddy, K. H. Lee. Chapter 2 - Bending of Beams. In: Lee CMWNRH,
editor. Shear Deformable Beams and Plates. Oxford: Elsevier Science Ltd. p. 11-38.(2000).
16. P. R. Heyliger, J. N. Reddy. A higher order beam finite element for bending and vibration
problems. Journal of Sound and Vibration.126, 309-26 (1988).
17. M. Shinozuka, G. Deodatis. Simulation of Stochastic Processes by Spectral Representation.
Applied Mechanics Reviews.44, 191-204 (1991).
18. M. Shinozuka, G. Deodatis. Simulation of Multi-Dimensional Gaussian Stochastic Fields by
Spectral Representation. Applied Mechanics Reviews.49, 29-53 (1996).
 Fatigue Analysis of Jack-up Leg Structures in Transit
condition

Dinh Quang Cuong1; Vu Dan Chinh2

1
Assoc. Prof. PhD, Faculty of Coastal and Offshore Engineering, National University of Civil
Engineering, Ha Noi, Viet Nam
2
MSc, Faculty of Coastal and Offshore Engineering, National University of Civil Engineering,
Ha Noi, Viet Nam
cuongdq.vctb@gmail.com

Abstract. In transit condition, jack-up platform is a type of floating structures

with large dimension. In towing duration, legs of the self-elevating unit are ele-
vated, so they do not directly subject to wave and current loads but only resist
inertia loads due to motion induced wave and other loads. Actually, fatigue
damages of the jack-up legs structure in transit condition have not been studied
clearly yet. However, Registers and Consultancy have been used 20% of total
fatigue life for transit life time in the analyses and designs [1]. The percentage
is approximately value, which has not accurately reflected the fatigue life in
transit condition. To have the more accurately results, it is necessary to solve
two important problems which are motions analysis and determination of inertia
forces on the legs structure induced by the motions. These problems will be
made clearly in the article and applied for 400ft jack-up platform fabricated in
Viet Nam.

Keywords: Jack-up structures; Transit condition; Fatigue analysis.

1 Introduction

In transit condition, Jack-up platforms can be towed by ships (Wet tow) or be carried
by barge (Dry Tow). Although only moving in a short time and in a mild sea-state,
but they still have cumulative damages induced by fatigue which has effects on the
total design life. Nowadays, the problems are mentioned in some researches and
standards [2], [3]. The article aims to clarify basic aspects of fatigue analysis of jack-
up leg structures in transit condition and its applications for Vietnamese conditions.

2 Determination of Hotspot Stresses of Jack-up Leg Structures

2.1 Wave loads and inertial forces

© Springer Nature Singapore Pte Ltd. 2020 123

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_16
124 D. Q. Cuong and V. D. Chinh

Floating structural motion components (Fig. 1)

Translations Rotations
1. Surge (along X) 4. Roll (about X)
2. Sway (along Y) 5. Pitch (about Y)
3. Heave (along Z) 6. Yaw (about Z)

Fig. 1. Definition of Axis Systems and

Floating Rigid Motions

The equation of motion

ሺ‫ ܯ‬൅ ‫ܣ‬ሻܷሷ ൅ ‫ܷܥ‬ሶ ൅ ‫ܭ‬௛௬௦ ܷ ൌ ‫ܨ‬ሺ‫ݐ‬ሻ (1)

where: M is structural mass matrix in global axis; A is added mass matrix determined
from radiation wave potential on the mean wetted body surface of hull; Damping
matrix C is determined from radiation wave potential; Hydrostatic Stiffness Matrix
Khys is determined from hydrostatic pressure effect on wet surface of hull [4]; U&, U&&:
vectors of motion velocities and accelerations; F(t): the vector of wave loads based on
incident wave potentials and diffraction wave potentials.

Inertial force
Motions at center of gravity of a jack-up platform are determined from (1) induces
inertial forces at ith mass point of the structure. The forces [5] include components of
inertial in translations, centrifugal inertial and tangential inertial forces.

2.2 Hotspot stresses determination

Assuming that incident waves are stationary random processes specified by spectrum
density functions ܵఎఎ ሺ߱ሻ, the response spectrums of the structures in equation (1) can
be expressed in formula as below:
ܵ௨௨ ሺ߱ሻ ൌ ܴ‫ܱܣ‬ሺ߱ሻଶ Ǥ ܵఎఎ ሺ߱ሻ (2)

where ܴ‫ܱܣ‬ሺ߱ሻ is Response Amplitude Operator of the structural response u(t).

According to [5], based on ܵ௨௨ ሺ߱ሻ, the inertial force spectrums at mass mi can be
determined depending on response accelerations of the structural central of gravity,
roll and pitch angles, distances from mi to the central of gravity. The structures will be
analyzed by finite element method, the stress spectrums SVV(Z) are determined at
hotspot of the structures including stress concentration factors [6]. Finally, the stress
ranges and number of cycles are determined by formula in [7]. Actually, joints of
jack-up legs are complicated, so the hotspot stress concentration factors (SCF) are
commonly determined by analysis of local FE models [5], [6].
Fatigue Analysis of Jack-up Leg Structures in Transit condition 125

3 Fatigue Analysis of Jack-up Leg Structures in Transit

Conditions

Assuming a hotspot of a jack-up structure resist m group of stresses ranges Si with

number of cycles ni. According to Palmgren-Miner rule [6], [7], fatigue cumulative
damage ratio of the hotspot is expressed by formula as below:
௡
‫ ܦ‬ൌ σ௠
௜ୀଵ ே
೔
(3)
೔

Commonly, jack-up legs are made of high strength steels with yield limits large than
500MPa, so fatigue limit cycles Ni can be determined based on Si (MPa) by the S-N
curve in DnV RP-C203 [8] as below:

‫ܰ݃݋ܮ‬௜ ൌ ͳ͹ǤͶͶ͸ െ ͶǤ͹‫ܵ݃݋ܮ‬௜

(4)

Fatigue life of the hot spot can be determined by the formula (5), with a safety factor
ߛ௙ . The factor value depends on standards [5], [6].
ଵ
ܶൌఊ
೑஽
(5)

This article establishes the general algorithm for fatigue analysis of Jack-up legs
structures in transit condition in Fig. 2.
126 D. Q. Cuong and V. D. Chinh

Fig. 2. Fatigue analysis algorithm for jack-up leg structures in transit condition

4 A Vietnamese Case Study

In this article, Tam Dao 05 Jack-up platform [9] is considered to analyze fatigue of
the leg structures in transit condition, using SACS software version 5.6.

4.1 Input data

Parameters of structure
- Length of hull: 70,4 m
- Width of hull: 76 m
- Height of hull: 9,5 m
- Draft: 7320 mm
- Total tonnage, cargo
and ballast: 24642 tons
Table 1. Input data [9]
Material Properties Loading condition
- <RXQJ¶V0RGXOXV - Platform weights are given in Weight
E = 20000 kN/cm2 Control Report of Ju-2000E version;
- Yield strength: - Environment data for all directions in
Fy = 54,5 kN/cm2 transit condition:
- Density of steel + Wave height: H = 3m
J =7,85 T/m3 + Period: T = 6s

4.2 Models
SACS model of TD-05 jack-up platform structure for legs fatigue analysis includes:
Mass, Hull and Legs models. Fatigue loading will be calculated based on wave data in
transit conditions (Fig. 3).
Fatigue Analysis of Jack-up Leg Structures in Transit condition 127

a) Equivalent hull model b) Jack up structural model

Fig. 3. Jack-up structural model in transit condition

4.3 Results
0RWLRQ¶V5$20RWLRQ¶V5$2RIWKHVWUXFWXUHDUHGHWHUPLQHGLQGLUHFWLRQVFRUUe-
sponding to wave directions in transit condition. The typical RAOs corresponding to
45o wave direction are expressed in Fig. 4.

Fig. 4. 0RWLRQ¶V RAO of TD-05 Platform corresponding to 45o wave direction

Joint SCF Calculation
SCFs of joints of the leg structures can be determined based on Local finite element
models using Ansys version 17 (Fig. 5). The maximum SCF corresponding to axial
force, in-plane bending and out-plane bending are 2.55, 2.43 and 2.33, respectively.

Fig. 5. Joint modeling and the local stresses due to axial force, in-plane bending moment, out-
plane bending moment of a brace

Stresses Spectrum and Fatigue Cumulative Damage

In transit condition, wave data is given in table 1 with Pierson Moskowitz Spectrum.
Fatigue damage of the structural hotspots are analyzed by SACS software. The maxi-
mum damage of joint 1009 of member 1009-1031 is presented in Fig. 6 as below.
128 D. Q. Cuong and V. D. Chinh

Wave spectrum Stress spectrum of top

hotspot of joint 1009

1031
S-N Curve

1009
Cumulative Damage

Fig. 6. Stress Spectrum and Fatigue Cumulative Damage Results of joint 1009

Fatigue Life Results in transit condition

Three joints with minimum fatigue life are listed in Table 2. As the results, there are
some comments:
- The jack-up structure is safe in transit conditions. However, the damages have a
certain influence of the total life of the jack-up structure.
- The fatigue damages per year are significant, but they are limited by short moving
duration of jack-up platforms.
- The wave data in Table 1 are only used for fatigue analysis corresponding to field
transit condition. In case of ocean transit condition, the fatigue damages will be larger
because of the effects of larger wave heights.
- The maximum damages are usually occurred for joints at contiguous locations be-
tween jack-up legs and hull. So in the operation condition, fatigue may be happened
at high levels. Otherwise, in the transit condition, it may be happened at lower levels
because the legs are elevated. So the total fatigue life of the jack-up structure has to be
evaluated from the combination of operating condition and transit condition.

Table 2. Summary of fatigue life of the structure in transit condition

Location Damage TransitFatig Design Life for
Joint Member
in section per year ue Life transit condition
1009 Brace 1009-1031 Top point 0.028 35.48 5
3014 Brace 3014-3025 Top point 0.0266 37.57 5
2012 Brace 2012-2026 Top point 0.024 41.23 5

5 Conclusions

Based on the research results of the article, there are some discussions as below:
Fatigue Analysis of Jack-up Leg Structures in Transit condition 129

Using 20% of total life for fatigue life in transit condition is only a reference value
in design. It is necessary to analyze fatigue damage of existing jack-up structures in
transit conditions in case of structural re-use assessment or life extension assessment.
At the present, there are only recommendations in current standards about effects
of fatigue of the jack-up platforms in transit condition [2], [3] but there are no detailed
guidelines and procedures to evaluate the fatigue damages.
The article establishes the algorithm to analyze fatigue of Jack-up leg structures in
transit conditions based on the combination of calculation theories of floating body
PRWLRQV JOREDO DQG ORFDO VWUXFWXUDO DQDO\VLV UDQGRP IDWLJXH DQDO\VLV«ZKLFK are
published in standards and textbooks and professional software. The algorithm is the
main result of the article. It has been successfully applied for Tam Dao 05 Jack-up
platform. It is one of products of national research project No.SPQG02b.01-01 which
performed by research team of ICOFFSHORE and verified by Technip FMC Vi-
etnam.
In the next time, the authors will publish a method to assess total fatigue life of
jack-up leg structures in combination of operating and transit conditions.

References
1. Zuo Xin, Fatigue Assessment Analysis of a Jack-up Platform Pile Leg Structure, Harbin
Engineering University, China, (2015).
2. /OR\G¶V 5HJLVWHU RI 6KLSSLQJ *XLGDQFH 1RWHV )DWLJXH RI -DFN-Ups During Tow, G.
Tseng/LR, (1992).
3. HSE, Review of the Jack-Ups: Safety in Transit, (JSIT) technical working group, BMT
Fluid Mechanics Ltd, (2003).
4. Newman, J.N., 'The Theory of ship motions', advances in Applied Mechanics, Vol.18,
(1980).
5. Det Norske Veritas (DnV) ± RP-C104, Self-elevating Units, (2012).
6. American Bureau of Shipping (ABS), Guidance Notes on Spectra-Based Fatigue Analysis
for Floating Production, Storage and Offloading (FPSO) System, (2002).
7. Barltrop, N. D. P., Adams, A. J.. Dynamics of Fixed Marine Structures, Third Edition. The
Marine Technology Directorate Limited (1991).
8. DnV-RP-C203, Fatigue Strength Analysis of Offshore Steel Structures, (2005).
9. F&G JU-2000E Construction Specifications, Jack-up drilling rig, Friede and Goldman
LTD, Houston, Texas, USA, (2013).
Finite element modeling of the TECCO protection system
for rock-fall under impact loading

TRAN Van Dang1, TRAN Dong2 and Dennis Gross3

1
Thuyloi University, 175 Tay Son, Dong Da District, Hanoi, Vietnam
2
National Civil Engineering University, 55 Giai Phong, Hanoi, Vietnam
3 Geobrugg AG, Aachstrasse 11 , CH-8590 Romanshorn, Switzerland

tranvandang@tlu.edu.com

Abstract. The TECCO protection system consists of the high strength compo-
nents such as the high tensile steel (HTS) wire mesh and the high strength
thread bar. This system is used in the construction for slope stabilization,
ground support, landslide protection, especially rock-fall protection. The pur-
pose of this study was to predict the behavior of the wire mesh under impact
loading of rock-fall. A finite element has been constructed in order to model the
impact between a rock of 0.5 tons and a rectangular wire mesh with high veloci-
ty. The mechanical response of the HTS material is assumed as elastic-plastic-
isotropic material. Further, the contact behavior within the steel wire mesh and
rock-fall are also considered in the model using the Coulomb friction model.
Thanks to ABAQUS/Explicit based on finite element method, the impact phe-
nomenon will be clarified. The model may help to reduce the number of exper-
iments with high cost.

Keywords: Rockfall,TECCO protection system, finite element method

1 Introduction

Rockfall is a common phenomenon in mountainous areas. The unpredictability of

frequency of rockfall endangers human lives and infrastructure. Rockfall starts with
the detachment of rocks from bedrock slopes. All bedrock slopes are subject to vari-
ous degrees of weathering, which may lead to fracturing, opening of joint and there-
fore to promotion of rockfall. Protective measures against rockfall can be taken by,
for example, constructing barrier fences and restraining nets, but these measures are
expensive and they deteriorate with time. The TECCO protection system is an innova-
tive engineered solution used effectively for this problem [1]. The TECCO system
presents many outstanding properties such as a low CO2 emissions having approxi-
mately 1/9 as compared with those of the shotcrete frame, not to break through the

© Springer Nature Singapore Pte Ltd. 2020 131

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_17
132 T. V. Dang et al.

botanic cover at the construction site [9]. Many projects have been carried out using
the TECCO system, for example, the project built at Mülheim, Germany, protects
against rockfalls and local instabilities for an area of 420 meters long and 12 meters
high. Or In Vietnam, the project for maintaining the national highway N6, in HoaB-
inh Province from Km128+602 to Km130+037, the TECCO system has been used for
preventing of rock-falls.
The system includes thehigh tensile wire mesh, the main nail, the spike plate, the
connection clip (see Fig.1).

Thread bar
(main nail)

Thread bar
HTS wire (main nail)
mesh

Fig.1. Elements of the TECCO system.

The common wire mesh is TECCO G65/3. The wire mesh features a diamond-shaped
individual mesh measuring 83 mm×143 mm. The individual diamond mesh surrounds
a caliber of 65 mm. The wire used of the mesh has a diameter of 3 mm. The strength
of the individual wire of mesh can reach1770 N/mm2.The feature of the wire mesh in
longitudinal and transversal tension is summarized in the Table 1.

Table 1.Characteristic values in tension of the TECCO G65/3 wire mesh [9]
Direction of Peak load Rupture de-
tension (kN/m) formation
(mm/m)
Longitudinal 165.7÷168 58.9÷61.8
Transversal 56.5÷59.6 198.5÷211.4
The spike plate is one of the most important elements of the TECCO system. This
special plate has been carefully designed in order to meet the best possible load and
force transmission between soil, mesh, plate and nail.

Fig.2.Spike plate and orientations.

Finite element modeling of the TECCO protection system … 133

Spike plates need to be oriented horizontally; long side must be horizontal. (Horizon-
tal installation covers 16 steel wires; vertical installation covers only 8 steel wires). It
is important that the spike plate is installed horizontally (see Fig. 2).
The connection clip is used for connecting the single wire meshes all together. The
connection clip T3 is made from high-tensile steel wire with a minimum tensile
strength of 1770 N/mm2 just like the high tensile steel wire meshes itself. The connec-
tion clip T3 has a diameter of 4.0 mm. The clip is only 60 × 21 mm and has two re-
versed end hooks on the one side of the clamp (see Fig.3).These connection clips can
transfer 100% of the necessary load between the individual mesh panels.

21 4
60

Fig.3.Geometry of the connection clip T3.

Today the design of the TECCO system for the construction is realized by using the
engineered software, named RUVOLUM [9]. The software RUVOLUM is based on
the theory and the experimental results that have been published on several research
works [2-8].

2 Numerical model

2.1 Model of the steel material

The steel material is assumed to be an elastic-plastic material and homogeneous. The
behavior of steel is described as a red curve in Fig.4.
Stress (N/mm2)

σ0=1770 N/mm2

ε0=0,02 Strain (%)

Fig.4. Behavior of steel [11].
134 T. V. Dang et al.

The elastic term is inside of (0, σ0). The elastic threshold is taken as 1770 N/mm2 for
the high tensile steel.
2.2 Contact model
The contact between the steel wire and the rock-fall is assumed using the Coulomb
friction model [11] (Fig.5).

Equivalent shear Critical shear stress

stress in default model

stick region Fig. 5. Coulomb friction model.

m:Constant friction
coefficient
Contact
pressure

3 Finite element model and results

The finite element model of the impact between the wire mesh and rock has been
realized in Abaqus/Explicit (see Fig.6). Two element parts have been realized in
which the wire mesh is modeled by using Beam element B31 of 2-node linear beam in
space. For the rockfall, the Discrete Rigid element: R3D4 of 4-node, 3-D bilinear
rigid quadrilateral has been used [11].

rockfall

Boundary
rope
Wire mesh

Fig.6. Finite element model of the impact between the wire mesh and rock.

A finite element has been primarily carried out in order to determine the properties of
the wire mesh which are used for the final model.
Finite element modeling of the TECCO protection system … 135

Fig.7 presents a comparison between the experiment of the wire mesh in tension and
the finite element model with good results.

Fig.7. Finite element model of the wire mesh in tension and the comparison between
experiment [9] and FE model.

Fig.8 presents the results of the numerical model which predict the behavior of the
wire mesh under impact loading of the rockfall. The diagram of the kinetic energy of
the rockfall shows a relevant result with an impact loading of the 0.5 tons of rockfall
under a velocity of 80 km/h.

Fig.8. Results of the numerical model with the velocity of 80 km/h of rockfall.
136 T. V. Dang et al.

4 Conclusion

The TECCO System was widely applied for many buildings in the world and in Vi-
etnam. The technology has been became a good solution for stabilization of slopes
and prevention of rock-fall hazard because it can finish thoroughly the problems.
Moreover, it represents a greening solution in construction.
This paper presented the TECCO system with elements and their properties. A finite
element model has been realized in Abaqus/Explicit to predicting the behavior of the
wire mesh under impact loading of the rockfall.

References
1. Wartmann, S.; Roth, A. (2005). Experimental and Numerical Modeling of Highly Flexible
Rockfall Protection Systems. International Symposium Society for Rock Mechnics and
Engineering (CSRME), China.
2. Loganatham, N., Balasubramaniam, A.S., Bergado, D.T.(1993). Deformation analysis of
embankments. J. Geotech. Engrg. ASCE. 199(8):1185-1206.
3. Baraniak, P.; Mrozik, M. (2010): Konstrukcjechroniąceprzedspływamigruzowymiorazspa-
dającymiodłamkamiskalnymi – testowanie, wymiarowanie, instalacja, użytkowanie – Pro-
tection systems against debris flow and rockfalls – testing, dimensioning, installation, ex-
ploitation, Konstrukcjestalowe w geotechnice – Steel structures in geotechnics, pages 79–
88; Poland.
4. Gerber, W.; Baumann, R.; Wartmann, S.; Buri, H.; Honegger, R.; Kaufmann, R.; Testi, R.;
Haller, B.; Toniolo, M. (2000). Guideline for the approval of rockfall protection kits, Swiss
Agency for the Environment Forests and Landscape (SAEFEL), Switzerland.
5. Muraishi, H.; Sano, S. (1997). Full Scale Rockfall test of ringnet barrier and components,
PWRI, Japan.
6. Wart mann, S.; Gerber, W. (2001): Field Testing of Rockfall Protection Barriers; Compar-
ison between the “Two Most Important Test Methods”, Swiss Federal Research Institute
WSL, Birmensdorf, Switzerland.
7. EUROCODE 7 (2004). Geotechnical design – Part 1: General rules.
8. Griffiths, D. V.; Lane, P. A. (1999). Slope stability analysis by finite elements, “Geotech-
nique”, Vol. 49 (3), pp. 387–403.
9. CałaFlumRoduner,RüeggerWartmann (2012). TECCO Slope Stabilization System and
RUVOLUM Dimensioning Method, ISBN 978-3-033-03296-5, Romanshorn, Switzerland,
AGH University of Science and Technology, Faculty of Mining & Geoengineering.
10. Test results (2016). National Research Institute of Mechanical Engineering of Vietnam. No
16-KKC.
11. Abaqus theory manual (2008). DassaultSystèmesSimulia Corp. Providence: Rhode, Island,
U.S.A.
 The study on the behavior of the simply supported beam
steel bridge structure without the intermediate bracing
system in the construction stage

Van Nam Le1, Anh Rin Nguyen1, Lien Thuc Pham1 and Ngoc Thi Huynh1*
1
VNUHCM-University of Technology, Ho Chi Minh City 72506, Vietnam
* Corresponding author

huynhngocthi@hcmut.edu.vn

Abstract. Steel bridge structure without intermediate bracing systems (IBS) has been
proposed. This structure has only been applied to the continuous I beam bridges in Ja-
pan. The construction of continuous bridges has many difficulties because of the solu-
tion of the deck cracking at the negative moment positions. That solution is complicated
and time-consuming. Moreover, the overall instability of main girder during construc-
tion stage is not considered. For simplifying the construction process as well as avoid-
ing above mentioned problem, the steel bridge structure without IBS has been inno-
vated for the simple span structure. Then, the models of the proposed structure and the
traditional steel bridge structure are simulated in construction stages in which the over-
all instability usually occurs. In addition, the destruction of structure caused by con-
struction load exceeding the bearing capacity of this structure also frequently happens.
Therefore, from analyzed data of those simulation the result could be found that 1) the
influence of the bar density (number of bars over an area unit) on the stability of the
proposed structure is only really clear when it reaches a specific value; 2) meanwhile,
the thickness of steel plate strongly affects to overall stability of this structure; and 3)
GHVSLWHWKLVVWUXFWXUH¶VFROODSVHGXHWRWKHRYHUall instability its material is still not used
to its full ability.

Keywords: Bracing systems; composite beam bridge; overall instability

1 Introduction

Nowadays, steel bridges are widely applied because of many outstanding advantages:
light structure, high bearing capacity, convenient for transportation, easy construction
and assembly. However, the save of maintenance costs and fast construction are still
very essential. Since then, the idea of removing some parts such as reinforcing ribs and
the IBS of the traditional steel bridge structure has been proposed. These components
are going to be replaced by new structural systems so that the structure still has enough
force-resistance and does not lose the overall stability during construction stage. This
structure is called the non-IBS bridge one. Bridge structure with non-IBS has been used
in Japan in recent years such as Tomari bridge, Hikifune bridge [1]. The structure of
the system including bearing beams, lateral beams is arranged at the abutment and pier

© Springer Nature Singapore Pte Ltd. 2020 137

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_18
138 V. N. Le et al.

positions. The I-shaped steel anchor bar helps the concrete deck slab prevents defor-
mation in small space and avoids sliding along main girder. In Vietnam, this is a new
bridge structure and although never applied in practice, it has received the attention
from Vietnamese researchers. A remarkable study presented a summary of the analysis
of load-bearing capacity and displacement of the continuous I-shaped girder system
without IBS for evaluating the applicability of this structure to reality [2]. In addition,
the stability of the both two structures has not been mentioned. Therefore, in this study
the overall stability, stress and deflection of the simply supported beam steel bridge
structure with and without IBS in the construction stage will be investigated by check-
ing the influence level of bracing system types and factors such as bar density and steel
plate thickness on them respectively. Through this investigation result, the paper aims
to show the confirmation of the applied possibility of this non-IBS structure in reality.

2 Methodology

To analyze the stability of the bridge structure without IBS, the finite element method
is used as a theoretical basis [3]. Set Ocr as critical load factor. The static equilibrium
equation of a structure at a deformed state is expressed as:
[K]{U}+[KG]{U}={P} (1)
where [K]: elastic stiffness matrix, [KG]: geometric stiffness matrix, {U}: displace-
ment of structure and {P}: applied load.
From Eq. (1) the equilibrium equation is rewritten as:
[K+OKG] {U} = {P} (2)
Set the equation [K+OKG] = [Keq]: Equivalent stiffness matrix.
In order for a structure to become unstable, the Eq. (2) must have a singularity. That is,
buckling occurs when the equivalent stiffness matrix becomes zero.
|[Keq]| < 0 (O>Ocr) Unstable equilibrium state
|[Keq]| = 0 (O=Ocr) Unstable state
|[Keq]| > 0 (O<Ocr) Stable state
Therefore, the instability analysis problem in Eq. (2) can be contracted to an eigenvalue
analysis problem.
[K+OiKG] = 0 (3)
where Oi: eigenvalue (critical load factor). The critical load equals the applied load mul-
tiplied by the critical load factor [3] rewriting as the following formula: {Pcr} =Ocr {P}
(where {Pcr}: critical load vector, {P}: applied load vector). Hence, with the known
applied load {P}, in case Ocr > 1, {P} is less than the critical load {Pcr} or in other
words,{P} has not reached the critical value. This time the system is stable. Case Ocr<1
or Ocr = 1, {P} has reached or exceeded the {Pcr}.This means that the system is unstable.

3 Analysis case

3.1 Object of analysis

In this study, the Luong Thuc Bridge (Ca Mau Province) has been selected as the
The study on the behavior of the simply supported beam … 139

research object [4]. Luong Thuc Bridge has a structure of traditional girder system, with
horizontal bracing system of I-shaped steel beams. The bridge has a total length of
90m, including 3 spans, maximum span length is 35.4m, with a beam length of 35.9 m.
The bridge width is 6m and the distance between the closest two main beams is 1.5m.
Horizontal beams at the bearings are 0.6m high and the intermediate bracing system
uses horizontal I-shaped 400 steel beams.

3.2 Numerical model of the steel bridge with and without IBS

In terms of non-IBS structure, IBS are removed while the I-shaped 150 steel anchor
bar and the steel plate are used to be replaced. The influence of only two factors the
steel plate thickness (t) and the distance between the closest two anchor bars (s) on the
stiffness of the non-IBS structure are considered. IBS or traditional structure is kept
from the original structure of Luong Thuc Bridge. In order to compare the stability of
the bridge structure with and without IBS more objectively, the structures in which the
intermediate cross beams in the Luong Thuc Bridge are replaced by X and K-shaped
IBS are also simulated. The main girder, cross girder and steel plate are simulated by
plate element with 4 nodes for assessing the overall performance of components in a
structural system. Midas/Civil software has been used to simulate for this structural.

3.3 Design loading

The strength I limit state and the service I are used to assess the stability and the
displacement of the bridge structure, respectively [5]. Therefore, the load acting on the
bridge in the construction stage includes dead load - self weight of all components in
the structure and the construction live load 4.8x10-4 MPa (CLL).

4 Results and discussion

4.1 Stability analysis of the bridge structure without IBS

The stability (critical factor Ocr) of the bridge structures is surveyed when the thickness
of steel plates (t) and the distance between the two nearest anchor bars (s) change.
5 Allowable critical factor s=600
Critical factor

s=700 s=800
s=900 s=1000
(Ocr)

s=1100 s=1200

0
8 9 10 11 12 13 14 15 16
Thickness (t)

Fig. 1. The relationship between thickness (t), distance (s) and critical factor
140 V. N. Le et al.

Fig. 1 shows that the thicker the t is and the shorter the s is, the higher the critical load
is because the critical factor increases. The phenomenon of this result is reasonable
because the increase of t and the decrease of s make the bridge structure
more rigid. This means that the structure will be more stable. However, the influence
of t and s on the stiffness of the structure is not similar. This difference will be discussed
in next section. From Fig. 1, it also can be concluded that the structures corresponding
to the survey cases C(s, t) above the red line (Ocr = 1) are stable because at this area the
critical load is greater than the applied load.

4.2 The stability evaluation of the structure without IBS in the construction
stage

In order to assess accurately the stability of the bridge structure without IBS, the stable
non-IBS bridge structures with the smallest and largest Ocr are selected to proceed com-
paring. Fig. 2 shows that the critical force of the traditional bridge structures are about
1.4 to 8 times greater than that of the non-IBS ones. Therefore, a study on designing
IBS for the traditional bridge structure should be conducted intensively. The non-IBS
bridge structures of which Ocr is a little bit larger than 1 are optimally designed.
10
8.06 8.20
8 Calculated critical factor
Critical factor

Allowable critical factor

6 4.83
3.68
4

2 1.06 1.06 1.05 1.15

0
. C(600, C(700, C(800, C (900, C (600, I X K .
9) 11) 12) 13) 16)

Type of intermediate bracing system

Fig. 2. The critical factors of the non-IBS and traditional bridge structures

4.3 The stress and the displacement of both structures

400
Displacement (m)

0.10
Stress (MPa)

0.08
Calculated stress
0.06
200 Allowable stress 0.04
0.02
0 0.00

Type of IBS Type of IBS

Fig.3. The maximum stress of struc- Fig.4. The maximum displacement of
tures in construction stage structures in construction stage
The study on the behavior of the simply supported beam … 141

In order to ensure the structure is not destroyed by overload during construction stage,
the analysis of stress and displacement of the two types of structures above are carried
out. Fig. 3 points out that in all structures the stress has not exceeded the critical stress
value during construction stage, even if three cases of them are unstable. Fig.4 shows
that the large displacement of the beam in construction will cause difficulties in creating
camber. Because the camber design is calculated as the sum of the final deflection of
the static load with a part of the vehicle load. The camber beam's creation will be con-
sidered in the future research.

4.4 The factors affecting the stability of the bridge structure without IBS
Influence of the number of I-shaped anchor bars (s) on the stiffness of structure
Influence of the number of I-shaped anchor bars (s) on the stability is determined by
the following parameter: 'O s-s' = Os - Os' u 100% (4)
W Os
s-s'
where: 'OW : the increase of critical factor from s' to s, corresponding to the increase
Ns' - Ns
percent in the number of bars (from s' to s) W = u 100
, N:the number of bars.
Ns

100%
t=8 t=9 t=10
t=11 t=12 t=13
t=14 t=15 t=16
'O

50%

0%
0.31~0.50 0.33~0.45 0.35~0.50 0.35~0.55 0.38~0.55 0.38~0.61 0.41~0.61 0.45~0.61 0.61~0.7 0.61~0.81
1600-1000 1500-1100 1400-1000 1400-900 1300-900 1300-800 1200-800 1100-800 800-700 800-600
59.7% 34.7% 40.7% 55.2% 43.9% 59.7% 48.8% 35.1% 15.3% 33.6%
Density of bars
Fig. 5. Influence of the number of I-shaped anchor bars (s) on the structure

The Fig.5 shows that when the number of anchor bars increases steadily about 35%,
with a bar density of less than 0.61 bars/m2, the extremes of percent increase in critical
factor is only 19.62% (case t = 16mm) much smaller than the percent increase of the
number of bars. However, when the bar density is greater than 0.61 bars/m2, the small-
est percent increase of critical factor is 59.94% (case t=8mm), much larger than the
percent increase (33.6%) of the number of bars.

Influence of the thickness of the steel plate (t) on the stiffness of the structure
Influence of thick steel plate (t) on the stability of the system is determined by the fol-
lowing parameter: t t Ot - Ot (5)
'OEn
n n1 = n1 u 100%
Ot
n1
tn tn 1
where: 'O : The percent increase of the critical factor from tn1 to tn ,
E

corresponding to the one in the thickness of plate E : E tn  tn 1

u 100
tn 1
142 V. N. Le et al.

Table. 1. Influence of the steel plate thickness (t) to the stability of the structure

1600 1500 1400 1300 1200 1100 1000 900 800 700 600
N=67 N=72 N=76 N=82 N=88 N=97 N=107 N=118 N=131 N=151 N=175
'O
10 8 60.1% 60.4% 57.9% 57.6% 56.8% 56.3% 50.6% 53.6% 57.5% 58.9% 80.7%
25%

'O
119 5 1.9% 52.3% 51.4% 49.7% 48.9% 49.0% 50.0% 50.6% 51.2% 52.6% 44.4%
22.2%

'O
1210 45.5% 46.0 % 42.9% 43.2% 43.1% 43.7% 44.6% 45.4% 46.2% 47.5% 41.3%
20%

'O
1311 40.6% 40.8% 40.3% 40.7% 38.7% 39.5% 40.3% 41.1% 42.0% 43.4% 51.7%
18.2%

'O
1412 36.5% 36.6% 36.2% 36.2% 35.1% 36.0% 36.9% 37.8% 38.6% 39.9% 46.4%
16.6%

'O
1513 32.5% 31.6% 32.1% 31.4% 32.4% 33.1% 34.0% 35.0% 35.2% 37.0% 37.2%
15.4%

'O
1614 28.7% 28.5% 29.9% 29.8% 30.0% 30.8% 31.7% 32.5% 33.4% 34.5% 34.5%
14.3%

The Table.1 shows that the percent increase of the critical factor is almost 2 times larger
than the one of the thickness in all bar arranged on the surface of the steel plate. This
demonstrate that the thickness of the steel plate greatly affects the stability of the struc-
ture.

5 Conclusions

When the bar density (number of bars over area unit) is greater than 0.61 bar/m2, the
change of number of I-shaped steel anchor bars significantly affect the stability of the
structure and vice versa. Meanwhile, when the thickness of steel plate (t) increases, the
percent increase of the critical factor is about 2 times greater than the one of thickness.
On the other hand, large displacements cause some difficulties in creating camber, but
this one is fully possible handle.
Although this structure collapses by the overall instability, the material is still not used
to its full ability. Therefore, in the construction stage, the structure is not obviously
destroyed by overload in terms of the stability. Hence, through above analysis results,
it can be concluded that the non-IBS bridge structure can completely be designed to
achieve the stability during construction stage. This is also one of the factors that can
confirm the high possibility of applying this structure to reality.

References
1. Panel BridgeTM, http://www. pnsastech. com.ph/Bridge-Products.php
2. Nguyen Thi Tuyet Trinh, Dao Duy Lam: Behavior analysis of new type of girder bridges
without intermediate beams. ISSN 0866-7012.2. Transport magazine, Vietnam (2015).
3. Analysis Reference of Midas Civil software.
4. Profile of Luong Thuc Bridge, 3E Steel
5. Vietnamese Highway Bridge Design Specification 11823: 2017 (VHBDS)
 Seismic behaviour of rammed earth walls: a time history
analysis

Quoc-Bao Bui 1,* and Tan-Trung Bui 2

1
Sustainable Developments in Civil Engineering Research Group, Faculty of Civil Engineer-
ing, Ton Duc Thang University, Ho Chi Minh City, Vietnam
2
University of Lyon, INSA Lyon, GEOMAS, 69100 Lyon, France
*
E-mail: buiquocbao@tdtu.edu.vn

Abstract. Rammed earth (RE) is a soil-based construction material which attracts

scientific researches because of its sustainable properties. This paper presents a
numerical study to investigate the in-plane seismic behaviour of RE walls. First,
an in-situ RE wall was modelled by the discrete element method (DEM). The
robustness of the model was checked by comparing dynamic properties of the
model with the experimental values. Then, an earthquake excitation was intro-
duced to the model to assess the seismic behaviour of the RE wall studied. The
signal was scaled at different amplitudes to assess the damages following differ-
ent earthquake intensities. The drifts were recorded and the seismic behavior was
discussed.

Keywords: Rammed Earth, DEM, Seismic Behaviour, Time History Analysis.

1 Introduction

Rammed earth (RE) is a construction material which is made by compacting the soil by
layers in formworks. The thickness of each compacted layer is about 10-15 cm. RE
without added binders is called “unstabilized RE” and when a binder is added (cement,
lime or geopolymer), it is called “stabilized RE”.
RE is interesting because of its low embodied energy and its positive hygric-thermal
behavior [1, 2]. Different aspects of this material was investigated by different studies,
on: mechanical, hygro-thermal behaviour and durability [3]. The seismic performance
of RE buildings is a topic which needs further investigations. The dynamic character-
istics of RE buildings were investigated in an experiment study [4]. The seismic per-
formance of RE walls was investigated in other studies by experimental [5 -7] or nu-
merical [8] approaches. Several studies used the nonlinear static method (pushover)
which is robust, but the dynamic phenomena are not directly considered. This paper
uses the time history analysis in which the discrete element method (DEM) and a real
earthquake signal were applied.

© Springer Nature Singapore Pte Ltd. 2020 143

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_19
144 Q.-B. Bui and T.-T. Bui

2 Numerical investigations

A RE wall is a superposition of different layers. The DEM is a relevant method to model

RE walls because the interlayers can be considered. The 3DEC code [9] was used. The
RE wall was modelled as an assemblage of discrete intralayer blocks and the interlayers
(with interface laws [10]). The intralayers are assumed to be homogeneous and iso-
tropic which are divided into internal elements for stress, strain, and displacement com-
putations. The failure criterion used in this study was the Mohr-Coulomb surface with
a tension cut-off behavior [9]. The interlayers consider shear, tensile failures and the
dilation.

2.1 Characterization of parameters

A parametric study was carried out in a previous study [10] to identify the values used
in DEM. The best parameters were identified by varying in the numerical model and
seeking the values which provided the closest numerical results to the experimental
ones. The parameters obtained are presented in Table 1 which are also logical with the
experimental results obtained by other studies [11, 12].

Table 1. Parameters of intralayer blocks and interlayers used in DEM

Tensile strength Cohesion Friction angle Normal stiffness

(kPa) (kPa) (°) (GPa/m)
Interlayers 113 113 38 60
Intralayers 133 133 45 -

An in-situ unstabilized RE wall having 50-cm thick was chosen for the validation of
the numerical model. In-situ vibrational measurements were performed on this wall by
using the accelerometers, during the construction phase and on the complete house after
the construction [4]. The wall’s dynamic characteristics were measured. The static char-
acteristics were determined by uniaxial compression tests on cylindrical specimens: the
compressive strength, density, Young’s modulus and Poisson’s ratio obtained were of
1.9 MPa; 20kN/m3; 470 MPa and 0.22, respectively [13]. The eigenfrequencies ob-
tained by the model had less than 5% of difference when compared to that obtained by
the measurements, which shows the relevancy of the model and the parameters used
for the case of small deformation (in-situ vibrations).

2.2 Approach adopted

This paper investigates the current type of RE buildings in France: two-storeys with
load-bearing unstabilised RE walls which have 50-cm of thickness. Their seismic per-
formance is investigated for an almost dry state (2-3% of moisture content [4]). Influ-
ences of the moisture content and the out-of-plane behaviour on the earthquake perfor-
mance are not considered in the present study.
Seismic behaviour of rammed earth walls: a time history analysis 145

The seismic performance of a RE building depends on different parameters: earth-

quake action (seismicity zone, soil type, site factors), the structure’s dynamic charac-
teristics (natural frequencies, mode shapes, damping), the material’s characteristics
(compressive, tensile strengths, Young’s modulus, density) and the structural typology
(spans, liaisons between structural and non-structural elements). Therefore, the earth-
quake performance of each building should be studied case by case by structural engi-
neers. However, the methods used by the engineers are usually simplified. The ap-
proach adopted in this paper is to use the advanced methods which are DEM, the time
history analysis and the incremental dynamic analysis (IDA) [14] to provide the most
general answer for the topic about the earthquake performance of RE buildings.
In seismic design, the standards usually assumes that the seismic shear force is dis-
tributed at the floor and roof levels of the structure; then, the shear force continues to
be distributed to the vertical structural elements. The seismic performance of a RE wall
depends on the mass supported which comes from the vertical loads (dead and live
loads); when this mass increased, the unfavourable effect increases [7]. In general, an
internal wall supports more mass than the external wall; so the seismic performance of
internal walls is worse than the external walls. So the seismic performance of the inter-
nal walls is studied in the present paper.
The RE wall is modelled where a simplified configuration of the RE wall is chosen
in the safe side. The wall has 2.5-m-length (like the in-situ wall presented above). Ver-
tical light elements between the walls (wooden and glass infill) were not considered. In
general, load-bearing RE walls have more than 2.5-m of length, the length chosen in
this paper tends to underestimate the seismic performance, which is in the safe side.
The wall includes a 0.4-m concrete base and 2.3-m RE wall (Fig. 1).

Fig. 1. RE wall modelled by DEM.

The vertical loads were calculated as indicated in Eurocode 8 [15]. These loads were
converted into an equivalent mass which concentrated at the wall’s head by increasing
the density of the top beam of the wall. Then, the wall was loaded at the top by a in-
plane horizontal load, to simulate an earthquake. The damping was neglected for sim-
plification which is in the security side because the damping decreases the vibration.
146 Q.-B. Bui and T.-T. Bui

2.3 Time history analyses

The El Centro earthquake signal (May 1940, NS component) was chosen which is a
well-known strong earthquake and is often used in dynamic studies. The real signal
(100% earthquake loading) has a maximum acceleration of 0.319g at the time of 2.02s.
Each earthquake has specific properties (frequencies, amplitude, duration, directions,
distance) so an earthquake signal can not be generalized for all earthquakes; however,
a well-known signal is chosen in the present study to assess the dynamic behavour of
RE walls. The IDA (Incremental Dynamic Analysis) is applied by multiplying the
earthquake acceleration with different scale ratios: 100% (real signal), 70%, 50% and
30%. The dynamic loading was introduced to the model with the time rate Δt of 0.013s.
The number of loading points is high to reproduce the dynamic effects by the model
(enough when Δt < T/8 where T is the predominant period which is 0.05-0.1s for RE
buildings [4]).

3 Results

The responses under earthquake loadings of the studied wall were measured by the
displacements at the top and at the base of the wall. The relative displacements between
the top and the base were determined. The relative displacements were used for the
determination of the “drifts” (relative displacement ⁄ storey height) which are usually
assumed as an index for the seismic evaluation [16]. The synthesis of the top wall’s
displacements is illustrated in Fig. 2 for all excitations studied. Table 2 presents the
synthesis of the maximum drifts obtained.

Fig. 2. Absolute displacements obtained at the top of the wall.

From Fig. 2, it is observed is that residual displacements (after the excitations) exist
for all cases studied. It is worth noting that the case of 50% amplitude has the absolute
displacement higher than the case of 30% amplitude, however, in term of relative dis-
placements, the case of 30% amplitude has a slightly higher value but these values are
still close, which indicates no significant difference of drifts between the cases 50%
and 30% loading. This result suggests that with excitations lower than 50% loading, the
wall’s behaviour remains in the elastic domain, while for higher loading, more damages
Seismic behaviour of rammed earth walls: a time history analysis 147

appear in the wall which increase the drift. This hypothesis is supported by observing
the damage states at the intralayers and the interlayers at different loading amplitudes
(which cannot be presented here due to the limited space): few damage zones were
observed both at intralayers and interlayers for the cases of 30% and 50% loading.

Table 2. Synthesis of results obtained

amax (% loading) Max. relative displacement Max. drift
0.319 g (100%) 12.98 mm 0.564 %
0.223 g (70%) 5.89 mm 0.256 %
0.160 g (50%) 2.35 mm 0.102 %
0.096 g (30%) 2.68 mm 0.117 %

4 Conclusions and prospects

A criterion was recently proposed to evaluate the seismic performance of RE structures

[16]. This criterion proposed a drift of 0.3% as a limit where the RE wall damage
changes from “moderate” to “extensive”; this drift limit can be used for current RE
buildings. By applying this drift limit for the present study, the investigated wall did
not have satisfying seismic performance for the case of 0.319 g (drift > 0.3%), however,
for other cases of loading (0.223 g, 0.160 g and 0.096 g) the studied wall had satisfying
performance (drift < 0.3%).
Therefore, the present study shows that two-storey RE buildings can have satisfying
in-plane seismic performance until the ground acceleration peaks of 0.23g. A 0.32-g-
acceleration corresponding to strong earthquakes caused serious damages for the stud-
ied wall, so two-storey-RE-buildings without reinforcements are not recommended in
high seismicity zones. Further studies with other seismic loadings, other building ar-
chitectures or with shaking tables will be interesting to compare with the results ob-
tained in the present paper.

References
1. Morel JC, Mesbah A, Oggero M, Walker P. Building houses with local materials: means to
drastically reduce the environmental impact of construction. Build. Environ. 36, 10, 1119–
1126 (2001).
2. Soudani L, Fabbri A, Morel JC, Woloszyn M, Chabriac PA, Wong H, Grillet AC. Assess-
ment of the validity of some common assumptions in hygrothermal modeling of earth based
materials. Energy and Buildings 116, 498-511 (2016).
3. Bui QB, Morel JC, Reddy BVV, Ghayad W. Durability of rammed earth walls exposed for
20 years to natural weathering. Build. Environ. 44, 5, 912–919 (2009).
4. Bui QB, Hans S, Morel JC, Do AP. First exploratory study on dynamic characteristics of
rammed earth buildings. Eng. Struct. 33, 12, 3690–3695 (2011).
5. Hamilton III HR., McBride J, Grill J. Cyclic testing of rammed-earth walls containing post-
tensioned reinforcement. Earthquake Spectra 22, 937–959 (2006).
148 Q.-B. Bui and T.-T. Bui

6. Silva RA, Oliveira DV, Miranda T, Cristelo N, Escobar MC, Soares E. Rammed earth con-
struction with granitic residual soils: The case study of northern Portugal, Constr. Build.
Mater. 47, 181-191 (2013).
7. El-Nabouch R, Bui QB., Plé O, Perrotin P. Assessing the in-plane seismic performance of
rammed earth walls by using horizontal loading tests, Eng. Struct. 145, 153–161 (2017).
8. Bui QB, Bui TT, Limam A. Assessing the seismic performance of rammed earth walls by
using discrete elements, Cogent Engineering 3: 1200835 (2016).
9. Itasca, 3DEC - Three Dimensional Distinct Element Code, V4.1, Itasca, Minneapolis (2011).
10. Bui TT, Bui QB, Limam A, Morel JC. Modeling rammed earth wall using discrete element
method, Contin. Mech. Thermodyn. 28, 1–2, 523–538 (2015).
11. El-Nabouch R, Bui Q-B, Perrotin P, Plé O. Experimental and numerical studies on cohesion
and friction angle of rammed earth material, 6th BIOT, Paris, July 10-13 (2017) ASCE.
12. Araki H, Koseki J, Sato T. Tensile strength of compacted rammed earth materials, Soils and
Foundations 56, 189–204 (2016).
13. Bui QB, Morel JC, Hans S, Walker P. Effect of moisture content on the mechanical charac-
teristics of rammed earth, Constr. Build. Mater. 54, 163–169 (2014).
14. Vamvatsikos D., Cornell C. A. Incremental dynamic analysis, Earthquake Engng Struct.
Dyn.31:491–514 (2002).
15. EN 1998-1:2004, Eurocode 8: Design of structures for earthquake resistance, European
Committee for Standardization (2005).
16. El-Nabouch R, Bui Q B, Plé O, Perrotin P, Rammed earth under horizontal loadings: Prop-
osition of limit states, Constr. Build. Mater. 220, 238-244 (2019).
 Analysis of Two-Directional Seismic Deterioration of
Steel Box Columns

Tran Tuan Nam1

1
Ho Chi Minh City University of Technology (HUTECH), 475A Dien Bien Phu, HCMC,
Vietnam
tt.nam@hutech.edu.vn

ABSTRACT. This paper addresses the deterioration of the steel box column
subjected to simultaneous biaxial moment and axial force, which caused local
buckling, consequent decrease in base shear capacity and eventual collapse of a
full-scale 4-story steel building tested at the E-Defense shake table. Two-
directional deterioration was observed in the columns which were subjected to a
major few cycles in a particular direction and later to the largest cycle in a sig-
nificantly rotated direction. The multi-spring (MS) element that consists of
springs discretizing the column cross section was used to analyze the moment
deterioration by local buckling.

Keywords: Collapse test · Steel box column · Local buckling · Two-directional

deterioration · Multi-spring (MS) element

1 Introduction

Strength loss of steel columns is caused by not only axial compression increase but
also local buckling effect. As recorded in a shake-table experiment on a full-scale
four-story steel building [1], the frame eventually collapsed by the soft story mecha-
nism due to local buckling of columns. In order to analyze this phenomenon, two-
dimensional (2D) numerical methods were adopted by several researchers. Column
deterioration after buckling was considered by authors [2] and [3], but the changing of
axial force caused by the seismic building overturning moment was not included.
Detailed three-dimensional (3D) finite element models used by authors [4] and [5],
therefore, would be more appropriate to analyze. However, those studies required
large modeling and computational efforts. It is not really necessary for the most of
structural components responding elastically as long as simpler analytical techniques
are available.
This study utilizes a simplified analysis technique that can model the column
elastoplastic behavior under varying axial load and biaxial bending moment. The
multi-spring (MS) element that consists of springs discretizing the column cross sec-
tion is extremely efficient and accounts for accumulated column damage based on the
strain history of each spring. Such an element is adopted in this study and appears to

© Springer Nature Singapore Pte Ltd. 2020 149

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_20
150 T. T. Nam

well analyze the two-directional moment deterioration behavior as recorded in the

test.

2 Full-scale steel building collapse test

2.1 Building specimen and input ground motion

The building specimen is a full-scale four-story steel moment frame (Fig. 1). It was
shaken to collapse on the E-Defense shake-table facility located in Hyogo Prefecture,
Japan, in September 2007. The columns are constructed using 300×300×9 cold-
formed steel square tubes with relatively large width-to-thickness ratios (b/t = 33).
The ground acceleration histories recorded at Takatori station during the 1995 Japan
Kobe earthquake (herein referred to as Takatori motion) were used as the input for the
test. Experimental responses of the building specimen under the weak, strong and
collapse shaking levels were formerly reported in [6] and [7]. Discussing the collapse
response of the building, the study [8] mentioned that the deterioration of columns
due to local buckling governed the soft-story mechanism.

Fig. 1. Building specimen: (a) overview, (b) elevation and plan, (c) post-collapsed configura-
tion, (d) hinging of first-story column ends due to local buckling.

2.2 Deterioration of first-story columns caused by local buckling

The experimental results at 100% Takatori motion level showed that the deterioration
of first-story columns due to local buckling caused the soft-story collapse mechanism.
To clarify the deteriorating sequence of all six columns in the first story at 100%
Takatori level, time-histories of their axial forces (including gravity load), as well as
their bottom moments MY-bot and MX-bot are shown in Fig. 2. In order to clarify the
strength loss due to local buckling, this study utilizes the combined strength factor R
proposed by Inoue [9] for square tube columns, counting for the simultaneous interac-
tion of axial force and biaxial moments (as shown in Eq. 1, where Ny is yield axial
strength, and Mp is plastic moment in case of zero axial force) in relation with story
drift ratio r characterizing the magnitude of column deformation.
Analysis of Two-Directional Seismic Deterioration … 151

R ¨¨
2

¸¸   «

4 § N · max  M X , M Y  3 ª min M X , M Y  º» (1)
3 © Ny ¹ Mp 4« Mp »
¬ ¼

Fig. 2. (a) 1st-story drift ratio orbit, and time histories of (b) axial forces, (c) bottom moments,
(d) combined strength factor R of 1st-story columns, and (e) 1st-story drift ratio (100% Takatori)

If the column is stable, its magnitude of factor R should increase in accordance

with the increase of story drift r. However as shown in Fig. 2, during the first cycle,
moment deterioration was observed in some columns, initiated from the decrease of
MY-bot and followed by the decrease of MX-bot even when the story displacement load-
ing was still progressing. The story displacement increased along the rX direction
faster than along the rY direction, thus causing earlier local buckling and consequent
deterioration of MY-bot. At the same time, even though rY was not very large, the ca-
pacity reduction of MX-bot also occurred, because the cross section had changed due to
the local buckling effect. Such sequence of deterioration to MY-bot and successively to
152 T. T. Nam

MX-bot also repeated in the following cycles. During the pre-collapse time, the widely
difference of tensile and compressive loads developed in columns.

3 Modeling concept

The buckling zone at the column ends (Fig.3(a)) is modeled using the multi-spring
element. Discretization scheme for the square tube column is displayed in Fig. 3(b),
where the element comprises totally 40 springs distributed over the cross section such
that the geometric properties (i.e. area and moment of inertia) are closely generated.
The MS HOHPHQW¶VURWDWLRQDOFKDUDFWHULVWLFLVDFKLHYHGE\SUHVXPLQJWKHILQLte length
lh of buckling zone. For the square tube column, lh is set equal to the column depth
based on the observations from experiment.
The nonlinear stress-strain properties of a spring element which are based on this
finite length are illustrated in FLJFZKHUHıy is yield stress. The degradation rule
IRUWKHFRPSUHVVLYHEXFNOLQJ ]RQHLVVSHFLILHGE\FRQVLGHULQJ <DPDGDHWDO¶VD[LDO
stress-strain rule [10] constructed from pure compression tests of stub columns with
b/t ratios varying from 15 to 35.

Fig. 3. Structural model and material property of a multi-spring element.

4 Dynamic analysis of two-directional column deterioration

This section presents time-history analysis of two selected columns (B1 and B3) in
the first story (Fig. 4). Assuming the inflection point remains stable at the middle
height of the column, the analytical column models are constructed as cantilever col-
umns with the height of H/2 = 1750 mm, where H is the actual first-story column
height in the test frame. Displacement control is applied to the cantilever column
model, wherein the tip end is subjected to the bi-lateral displacement history so as to
reproduce the identical chord rotation as the first-story drift ratio recorded in the col-
lapse test at 100% Takatori motion level.
In each Fig. 4(a) or Fig. 4(b), the upper graph plots the interacting relation between
MX-bot (horizontal axis, related to rY) and MY-bot (vertical axis, related to rX) as well as
the yield surface represented by dotted line. The two lower graphs in each group show
WKH K\VWHUHWLF UHODWLRQVKLSV ³MY-bot vs. rX´ and ³MX-bot vs. rY´ IURP OHIW WR ULJKW Ue-
Analysis of Two-Directional Seismic Deterioration … 153

spectively. Note that the vertical axis is also reversed to define the moment coordi-
nates consistent with the displacement coordinates presented earlier in Fig. 2(a).
To clarify the moment interaction surface, the curves relating to the elastic loading
time before 3.2 sec, i.e. the start of the first cycle, are omitted. The stop point marked
by circle symbol is related to 6.57 sec, the time when the building completely col-
lapsed.
During the first two cycles of loading close to the rX direction, MY-bot developed
earlier than MX-bot and rapidly reached to yield surface. However, the difference of
axial force variation caused different responses of MY-bot of two columns during this
time. MY-bot of column B1 which had the smallest compression force hence was quite
stable on both ±rX and +rX loading directions [Fig. 4(d)]. In contrast, the MY-bot capaci-
ty of column B3 [Fig. 4(h)] was reduced significantly, especially on the ±rX side,
where they were carrying quite large compression forces.
Local buckling must have occurred considerably to those columns during this time,
thereby limiting their moment capacity. Consequently, the development of MX-bot was
limited, particularly in the last half cycle until collapse at nearly 0.2 rad of rY defor-
mation. The increasingly significant levels of MX-bot deterioration during that time
(characterized by the decreasing ratio of the average remaining capacity to the nomi-
nal plastic moment Mp) are easily recognizable from columns B1 [Fig. 4(e), ratio
0.67] to B3 [Fig. 4(i), ratio 0.31]. The moment interaction surface of column B1 [Fig.
4(a)] appeared to show fully, whereas the one of column B3 [Fig. 4(c)] was much
more shrinking, especially near the collapse time.
In general, the complex interaction between varying axial force and two-directional
moment deteriorations was closely simulated by the MS model.

Fig. 4. Analysis results of column bottom moments, applying varying axial force and
lateral displacement histories by 100% Takatori level.
154 T. T. Nam

5 Conclusions

The study clarified that any deterioration of column moment due to local buckling in
the X direction, for instance, may limit the capacity of moment in the Y direction, and
vice versa, which can be explained based on the shrinkage of moment yield surface.
The MS element was used to model the buckling zone at the column end proved its
advantage in the study. Dynamic analysis of isolated cantilever columns shows the
efficiency of MS element method in simulating column responses under sophisticated
axial load conditions.

Acknowledgements
The author acknowledges Prof. Kazuhiko Kasai (Tokyo Institute of Technology), Dr.
Bruce F. Maison (California), Prof. Satoshi Yamada (Tokyo Institute of Technology),
and Dr. Yuko Shimada (Chiba University) for their enthusiastic technical support.

References
1. Kasai, S., Ooki, Y., Motoyui, S., Takeuchi, T., Sato, E.: E-Defense tests on full-scale steel
buildings, part 1 - experiments using dampers and isolators, Proceedings of Structural
Congress, ASCE, Long Beach (2007).
2. Maison, B.F., Kasai, K., Deierlein, G.: ASCE-41 and FEMA-351 evaluation of E-Defense
collapse test. Earthquake Spectra, 25 (4), 927±953 (2009).
3. Lignos, G.D., Hikino, T., Matsuoka, Y., Nakashima, M.: Collapse assessment of steel
moment frames based on E-Defense full-scale shake table collapse test. Journal of Struc-
tural Engineering (139), 120±132 (2013).
4. Tada, M., Tamai, H., Ohgami, K., Kuwahara, S., Horimoto, A.: Analytical simulation uti-
lizing collaborative structural analysis system. Proceedings of the 14th World Conference
on Earthquake Engineering, Beijing (2008).
5. Yu, Y.J., Tsai, K.C., Weng, Y.T., Lin, B.Z., Lin, J.L.: Analytical studies of a full-scale
steel building shaken to collapse, Engineering Structures (32), 3418-3430 (2010).
6. Yamada, S., Suita, K., Tada, M., Kasai, K., Matsuoka, Y., Shimada, Y.: Collapse experi-
ment on 4-story steel moment frame, part 1 ± outline of test results. 14th World Confer-
ence on Earthquake Engineering, Beijing, China (2008).
7. Suita, K., Yamada, S., Tada, M., Kasai, K., Matsuoka, Y., Shimada, Y.: Collapse experi-
ment on 4-story steel moment frame, part 2 ± detail of collapse behavior. 14th World Con-
ference on Earthquake Engineering, Beijing, China (2008).
8. Kasai, K., Nam, T., Maison, B.: Structural collapse correlative analysis using phenomeno-
logical fiber hinge elements to simulate twoǦdirectional column deteriorations, Earthquake
Engineering & Structural Dynamics 45 (10), 1581-1601 (2016).
9. Inoue, I.: Theory and design of steel structures (in Japanese), Kyoto University Press, Kyo-
to (2003).
10. <DPDGD6$NL\DPD+DQG.XZDPXUD+³3RVW-buckling and deteriorating behavior of
box-VHFWLRQ VWHHO PHPEHUV´ LQ -DSDQHVH -RXUQDO RI 6WUXFWXUal and Construction Engi-
neering, AIJ, No.444, pp. 135-143, 1993.
 Numerical analysis of double-layered asphalt pavement
behaviour taking into account interface bonding condi-
tions
Minh Tu LE 1,2, Quang - Huy NGUYEN 1 and Mai Lan NGUYEN 3
1
INSA de Rennes, 20 avenue des Buttes de Coësmes, CS 70839,
F-35708 Rennes Cedex 7, France
minh-tu.le@insa-rennes.fr, Quang-Huy.Nguyen@insa-rennes.fr
2
University of Transport Technology, 54 Trieu Khuc Street, Thanh Xuan Distr,
Hanoi, Vietnam
leminhtu@utt.edu.vn
3
IFSTTAR/MAST, CS4 - F 44344 Bouguenais Cedex, France
mai-lan.nguyen@ifsttar.fr

Abstract. Bonding conditions between pavement layers have an important influ-

ence on the responses of pavement structure. This paper deals with numerical
analysis for the stresses, strains and deflections of double-layered asphalt pave-
ment structure. The constitutive relations are based on the layered homogeneous
half-space theory developed by Burmister (1943). In this work, in order to im-
prove the modelling, the actual interaction between the layers is taken into ac-
count by considering a horizontal shear reaction modulus which represents the
interface bonding condition. The numerical model is validated and applied
through an original case study where falling weight deflectometer (FWD) tests
are carried out on two full-scale pavement sections with two different bonding
conditions at the interface between asphalt layers. The results indicate that the
considered numerical model allows evaluating and quantifying the interface
bonding condition in determining by backcalculation the horizontal shear reac-
tion modulus at the pavement interface.

Keywords: Asphalt pavement, Interface bonding, Analytical, Shear reaction

modulus, Numerical analysis, FWD.

1 Introduction

The condition of this interface bonding has an important impact on the distribution of
pavement stress and strain fields, and affects therefore the corresponding performance
and service ability of pavement. According to Khweir and Fordyce (2003) [1] and Ngu-
yen el al. (2016) [2], failures of asphalt concrete road surface due to poor bond between
asphalt layers may reduce the road surface service life by 40% to over 80 %. Many
researches in the literature have demonstrated the influence of the bond condition bet
tween layers on the performance of asphalt pavement [3-4]. Recently in the new Tech-
nical Committee PIM of the RILEM starting in 2016 [5-6], interlayer characterization
in multilayer pavement will be investigated through dynamic shear tests.
© Springer Nature Singapore Pte Ltd. 2020 155
C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_21
156 M. T. Le et al.

There are three kinds of bonding conditions of overlay pavement, which are un-
bonded, partially bonded and fully bonded overlays. To ease the modelling of the struc-
ture and the computation process, most methods assume that the layers are fully bonded
or unbonded to each other [7-8].
In this paper, the bonding condition at the interface could be then described as a
relative displacement between the two asphalt layers using this parameter. Furthermore,
an experimental case study has been done which validates the numerical model and
allows determining and comparing actual bonding conditions of two different pavement
sections.

2 Theoretical background and improvement

2.1 Theory of elastic half space for calculation of stresses in the pavement
structure
Typical multilayered pavement structure in cylindrical coordinates, where r and z are
the cylindrical coordinates in the radial and vertical directions, respectively. Each layer
is homogenous, isotropic, and horizontally infinite. The bottom layer is an infinite half-
VSDFH7KHHODVWLFPRGXOXVDQG3RLVVRQ¶VUDWLRRIWKHLth layer are Ei and Q i . In an asym-
metric cylindrical coordinate system the stress by tensor.

W rT i W rz i º»
ª V 
« rr i
«
V kh (r,T , z) i   VTT i WT z i »»
« W rT i
« »
¬ WT z i V zz i »¼
« W rz 
i

Fig. 1. Components of stresses in cylindrical coordinates

The pavement is subjected to a uniform circular load with radius and pressure mag-
nitude q. The elastic solution to the problem can be analytically derived from the lay-
ered elastic theory [7].

Fig. 2. Multilayered pavement structure

/HWȡ U+DQGȜ ]+,QZKLFK+LVWKHGLVWDQFHIURPWKHVXUIDFHWRWKHXSSHUERXQGDU\

of the lowest layer; J0 and J1 are Bessel function of the first kind and order 0 and 1
respectively; m is a parameter; Ai,Bi,Ci,Di are constant of integration to be determined
from boundary and continuity conditions. The superscript i varies from 1 to n and refers
to the quantities corresponding to the ith layer. A star super is placed on these stresses
Numerical analysis of double-layered asphalt pavement … 157

and displacement due to a vertical load of ±mJ0PȡQRWWKHactual stresses and dis-
placements due to a uniform q distributed over a circular are of radius. It can be verified
by substitution that:

*
­°> Ai  Ci (1  2Q i  mO ) @ e  m ( Oi  O ) ½°
(V )zz i mJ 0 (m U ) ®  m ( O  Oi 1 ) ¾
(1)
°¯ > Bi  Di (1  2Q i  mO @ e °¿

J1 (m U ) º °­> Ai  Ci (1  mO ) @ e i  °½
 m(O O )
ª
(V rr* )i mJ
« 0 ( m U )  ® ¾
¬ U »¼ ¯°> Bi  Di (1  mO ) @ e  m ( O  Oi1 ) ¿° (2)
 m ( Oi  O )  m ( O  Oi 1 )
2Q i mJ 0 (mU ) ª¬Ci e  Di e º¼
J1 ( m U )
*
(V TT )i
U
^> A  C (1  mO )@ e
i i
 m ( Oi  O )
 > Bi  Di (1  mO @ e  m ( O  Oi1 ) `
(3)
2Q i mJ 0 (m U ) ª¬Ci e  m ( Oi  O )  Di e  m ( O  Oi1 ) º¼
(W rz* )i ^
mJ1 (m U ) > Ai  Ci (2Q i  mO ) @ e  m ( Oi  O )  > Bi  Di (2Q i  mO ) @ e  m ( O  Oi1 ) ` (4)
1 Q i
(u* )i
Ei
^
J1 (mU ) H > Ai  Ci (1  mO ) @ e m ( Oi  O )  > Bi  Di (1  mO @ e  m ( O  Oi1 ) ` (5)

* 1 Q i ­°> Ai  Ci (2  4Q i  mO ) @ e  m ( Oi  O )  ½°
(w )i  J 0 (m U ) H ®  m ( O  Oi 1 ) ¾
(6)
Ei °¯> Bi  Di (2  4Q i  mO @ e °¿
* *
Where (u )i and (w )i : horizontal and vertical displacements of layer i.
2.2 Improvement taking into account actual bonding condition
The stresses and displacements as a result of load q are obtained by using the Hankel
f
R*
transform: R qD ³ J1 (mD )dm (7)
0
m
:KHUHĮ D+5* is the stress or displacement as a result of load ±mJ0Pȡ); R is the
stress or displacement as a result of load q. Based on the homogeneous half-space of
layered theory developed by Burmister (1943) most exiting works [7] have studied for
only two cases: unbonded and fully bonded. In this study, general case is taking into
account the actual bonding condition with the layers interface behaviour according to
Goodman's constitutive law [9].

K s 'u W rz
(8)
ZKHUHǻXis the relative horizontal displacement of the two faces at the interface, Ks is
the horizontal shear reaction modulus at the interface. So thereIRUHȜ Ȝi the continuity
conditions:

(V zz* )i (V zz* )i 1 ; (W rz* )i (W rz* )i 1 ; (W rz* )i K s ª¬(u* )i 1  (u* )i º¼ ; ( w* )i ( w* )i 1 (9)

158 M. T. Le et al.

According by Al Hakim (1997) [10] and Nguyen et al. (2016) [2], the Ks value can
be selected as follows:
- Ks < 0.01 MPa/mm: Unbonded
- 03DPP”. s”03DPP3DUWLDOO\ERQGHG
- Ks > 100 MPa/mm: Fully bond.
2.3 Backcalculation method
Falling weight deflectometer (FWD) testing has been widely practiced to assess
structural condition of pavement in determining by backcalculation the moduli of pave-
ment layers. In this work, the vertical displacements on the pavement surface is deter-
mined by the following a function:
ªf º
w 0 F « ³ f (emh , e mh , h, K s ,Q , E) J 0 (mU ) J1 (mD )dm » (10)
¬0 ¼
Direct calculations are firstly made for different pavement surface positions using
equation (10) with pavement layers properties (thicknesses, moduli and PoissRQ¶VUDWLR),
and the horizontal shear reaction modulus at the interface Ks as the inputs. These cal-
culations are then repeated until that the calculated values of vertical displacements are
as close as the ones measured by FWD.

3 Application of theoretical improvement for an experimental

case study

To validate the pavement analysis model, surface deflections measured by FWD on two
different pavement sections with varying values of interface condition are compared
with analytical values. FWD tests are carried out using a circular plate of 300 mm in
diameter. The positions of the velocity sensors are as follow: 0, 300, 450, 600, 900,
1200, 1500, 1800, 2100 mm. FWD tests are performed at three different locations for
each pavement section with the same load of 65 kN. The temperature measured in the
asphalt layer is about 22°C.

Table 1. Structure pavement of section I (SI) and section II (SII)

Layer E Thickness SI Thickness SII PoissRQ¶VUDWLR Material type

(Mpa) (cm) (cm)
1 8630 6.6 6.3 0.35 Asphalt
Interface Predicted by backcalculation method
2 9570 4.6 3.9 0.35 Asphalt
3 184 290 290 0.35 Subgrade
Concrete
4 55000 - - 0.25
Substratum
Observed that have good results between predicted and measured deflections, nu-
merical results indicate that the horizontal shear reaction modulus at the interface (K s)
will significantly affect pavement structure behavior. These values are calculated in
Numerical analysis of double-layered asphalt pavement … 159

accordance with the initial assumption of the two sections: good bond condition in sec-
tion I with Ks > 100 MPa/mm (Ks: 531, 109 and 131MPa/mm) and partially bonded in
VHFWLRQ,,ZLWK03DPP”Ks ”03DPPKs: 74, 76 and 69 MPa/mm).
Section I Section II

0 1 2 0 1 2
Surface deflections (—m)

0 0

Surface deflections (—m)

100 100
200
200
300
300 Measured Measured
400
400 Predicted, Ks=531 500 Predicted, Ks=74
(MPa/mm) (MPa/mm)
500 600
Distance from load center (m) Distance from load center (m)

a, Point 1 d, Point 4

0 1 2 0 1 2
Surface deflections (—m)

Surface deflections (—m)

0 0
100 100
200 200
300
300 Measured Measured
400
400 500
Predicted, Ks=109 Predicted, Ks=76
500 (MPa/mm) 600 (MPa/mm)
Distance from load center (m) Distance from load center (m)

b, Point 2 e, Point 5

0 1 2
Surface deflections (—m)

0 1 2
Surface deflections (—m)

0 0
100 100
200 200
300 Measured 300 Measured
400 400
Predicted, Ks=131 Predicted, Ks=69
500
500 (MPa/mm) (MPa/mm)
600
Distance from load center (m) Distance from load center (m)
c, Point 3 f, Point 6
Fig. 4. Predicted and measured deflections at section I (points 1, 2, 3) and section II
(points 4, 5, 6)
160 M. T. Le et al.

The obtained shear modulus are close to the ones from previous laboratory experi-
mental studies which were done on double-layered specimens, closer to the ones from
cyclic load tests [4] than from static load tests [2].

4 Conclusion

Numerical analysis of multilayer pavement structure is improved by taking into ac-

count actual bonding condition at the interface between asphalt layers. A partial bond,
represented by the shear modulus (Ks), is assumed for the last one together with the
classical layered homogeneous half-space theory. This original approach allows more
accurate analysis of multilayer pavement structure. Experimental application on full-
scale pavement sections using FWD measurements and backcalculation, this improved
method allows to predict the shear reaction modulus (Ks) in addition to the layers mod-
ules. Further research will focus at studying the behavior of asphalt pavement structure
under the whole histories of the FWD dynamic loadings. Moreover, linear viscoelastic
behavior can be applied for asphalt material layers for pavement analysis at structure
level.

References
1. Khweir K. and Fordyce D. Influence of Layer Bonding on the Prediction of Pavement Life.
Proceedings of the Institution of Civil Engineering Transport 156, 213-218. 2003.
2. Nguyen N.L., Dao V.D., Nguyen M.L., Pham D.H. (2016) Investigation of Bond Between
Asphalt Layers in Flexible Pavement. In: Chabot A., Buttlar W., Dave E., Petit C., Tebaldi
G. (eds) 8th RILEM International Conference on Mechanisms of Cracking and Debonding
in Pavements. RILEM Bookseries, vol 13. Springer, Dordrecht
3. Brown SF, Brunton JM. The influence of bonding between bituminous layers, Highways
and Transportation, 31(5), 16±7. 1984.
4. Diakhate M., Millien A., Petit C., Phelipot-Mardelé A., Pouteau B. Experimental investi-
gation of tack coat fatigue performance: Towards an improved lifetime assessment of pave-
ment structure interfaces, Construction and Building Materials, 25 (2011), 1123±1133.
.2011.
5. RILEM International Conference on Mechanisms of Cracking and Debonding in Pavement,
Nantes, 2016.
6. RILEM Technical Committee PIM (Phase and Interphase behaviour of bituminous Mate-
rial) TG3: Pavement Multilayer system. Activity Starting 2016.
7. Huang, Y. (2003). Pavement analysis and design, Prentice Hall, Englewood Cliffs, NJ, 2003.
8. %XUPLVWHU'0³7KHWKHRU\RIVWUHVVDQGGLVSODFHPHQWVLQOD\HUHGV\VWHPVDQGDSSOLFD
WLRQV WR WKH GHVLJQ RI DLUSRUW UXQZD\V´ +LJKZ 5HV %RDUG3URF $QQX 0HHW 23,
í
9. Goodman RE, Taylor RL and Brekke TL³$PRGHOIRUWKH0HFKDQLFVRI-RLQWHG5RFNVRI
6RLO0HFKDQLFVDQJ)RXQGDWLRQ'LYLVLRQ´$6&(SS- 659. 1968.
10. Hakim AI. An improved backcalculation method to predict flexible pavement layers moduli
and bonding condition between wearing course and base course, Liverpool John Moores
University, Liverpool, UK. 1997.
 3DUDPHWULFQXPHULFDOVWXG\RQDQRYHOHQHUJ\KDUYHVWHU
XVLQJLURQJDOOLXPDOOR\DQGVWUDLQUHVSRQVH

7XDQ0LQK+D6DLML)XNDGDDQG7RVKL\XNL8HQR
+R&KL0LQK&LW\8QLYHUVLW\RI7HFKQRORJ\+87(&++R&KL0LQK&LW\9LHW1DP

.DQD]DZD8QLYHUVLW\.DQD]DZD-DSDQ
hm.tuan@hutech.edu.vn

$EVWUDFW:LWKWKHDLPRIFRQWULEXWLQJWRHIILFLHQWVWUXFWXUDOKHDOWKPRQLWRULQJ
DSSURDFKHVWKLVVWXG\SURSRVHVDPLFURHQHUJ\KDUYHVWLQJVWUDLQJDXJHXWLOL]LQJ
LURQJDOOLXPDOOR\FDSDEOHRIJHQHUDWLQJHOHFWULFDOHQHUJ\IURPG\QDPLFVWUDLQ
UHVSRQVHV7KHLURQJDOOLXPDOOR\LVDGXFWLOHPDJQHWRVWULFWLYHPDWHULDOZLWKD
KLJKSLH]RPDJQHWLFFRQVWDQWJRRGPDFKLQDELOLW\DQGDODUJHLQYHUVHPDJQHWR
VWULFWLYHHIIHFWE\ZKLFKPDJQHWL]DWLRQFDQEHYDULHGE\PHFKDQLFDOVWUHVV7KH
GHYLFHKDVDVLPSOHVWUXFWXUHDFRPELQDWLRQRIDIODWSODWHRILURQJDOOLXPDOOR\
DQGDVWDLQOHVVVWHHOIUDPHWKHIRUPHUDWWDFKHGWRWKHODWWHU$YDULDWLRQLQVWUHVV
DSSOLHGDWWKHDOOR\SODWH\LHOGVDWLPHYDULDWLRQRIWKHIOX[JHQHUDWLQJDYROWDJH
RQWKHZRXQGFRLOV7KHVWUDLQGDWDLVDOVRPHDVXUHGDWWKHVDPHWLPH7RDFKLHYH
KLJKHIILFLHQF\LQERWKSRZHUJHQHUDWLRQDQGGHIRUPDWLRQPHDVXUHPHQWHIIHF
WLYHGHVLJQIRUWKHQRYHOGHYLFHLVUHTXLUHG7KHUHIRUHSURWRW\SHILQLWHHOHPHQW
PRGHOVZHUHFRQVWUXFWHGEDVHGRQWKHQXPEHURIFRPPHUFLDOGHVLJQVDYDLODEOH
RQWKHPDUNHW7KHQSDUDPHWULFVWXGLHVZHUHFDUULHGRXWE\FKDQJLQJWKHVKDSHV
KHLJKWVDQGZLGWKVRIWKHGHYLFHVLQRUGHUWRH[DPLQHWKHLUSRVVLEOHHIIHFWVRQ
WKHGHIRUPDWLRQEHKDYLRURIWKHSURSRVHGVWUXFWXUH)LQDOO\GHVLJQSURFHGXUHV
IRUWKHSURSRVHGGHYLFHZHUHUHFRPPHQGHGEDVHGRQWKHVHQXPHULFDOVWXGLHV

.H\ZRUGV(QHUJ\+DUYHVWHU6WUDLQ5HVSRQVH0DJQHWRVWULFWLYH0DWHULDO

 ,QWURGXFWLRQ

7KH PDLQ SUREOHP ZLWK KHDOWK PRQLWRULQJ V\VWHPV IRU DFWXDO VWUXFWXUHV LV EDWWHU\
SRZHU2YHUWKH\HDUVYLEUDWLRQEDVHGSRZHUJHQHUDWLRQWHFKQRORJ\LVXWLOL]HGHIIHF
WLYHO\LQYDULRXVILHOGV7KHWHFKQRORJ\KDVRSHQHGXSPDQ\RSSRUWXQLWLHVIRUEDWWHU\
IUHHZLUHOHVVHOHFWURQLFVDQGWKHSRZHUIRUWKHPRGXOHLVUHWULHYHGIURPYLEUDWLRQ7KH
VHOISRZHUHGDXWRQRPRXVZLUHOHVVVHQVRUV\VWHPLVDOVREHQH¿FLDOLQVWUXFWXUDOKHDOWK
PRQLWRULQJWRLQIRUPV\VWHPDEQRUPDOLW\RIFLYLOHQJLQHHULQJZRUNVVXFKDVEULGJHV
6RIDUPDQ\VWXGLHVUHODWLQJWRPLFURHQHUJ\KDUYHVWHUXVLQJLURQ±JDOOLXPDOOR\KDYH
EHHQ SHUIRUPHG ,Q  8HQR DQG <DPDGD GHYHORSHG D ELPRUSK YLEUDWLRQ HQHUJ\
KDUYHVWHU>@LQZKLFKWZRURGVRIWKHLURQJDOOLXP)H*DDOOR\ZHUHHPSOR\HGDQG
FDSDEOHRISURGXFLQJP:FP)RXU\HDUVODWHUWKHHQHUJ\KDUYHVWHUZDVLPSURYHG
E\XVLQJOHVVYROXPHRI)H*DDQGSHUPDQHQWPDJQHWDVFRPSDUHGWRWKHFRQYHQWLRQDO
RQH>@5HFHQWO\DQHZYLEUDWLRQSRZHUJHQHUDWLQJGHYLFHXVLQJD)H*DSODWHOLNH

© Springer Nature Singapore Pte Ltd. 2020 161

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_22
162 T. M. Ha et al.

HOHPHQWDGKHUHGWRD8VKDSHGPDJQHWLFIUDPHLVXQGHULQYHVWLJDWLRQ>@7KHGHYLFH
LVUHTXLUHGWRJHQHUDWHSRZHUIURPPLOGWRKLJKLQWHQVLW\YLEUDWLRQV
,QWKLVVWXG\DPLFURSRZHUJHQHUDWLQJVWUDLQJDXJHXVLQJD)H*DSODWHLVSURSRVHG
7KHGHYLFHLVH[SHFWHGWRERWKPHDVXUHVWUDLQYDULDWLRQDQGWRJHQHUDWHHOHFWULFDOHQHUJ\
IURPDPELHQWYLEUDWLRQV7KHSULQFLSOHRIRSHUDWLRQLVEDVHGRQWKHLQYHUVHPDJQHWR
VWULFWLYHHIIHFWRIWKHSODWH$SDUDPHWULFVWXG\ZDVFRQGXFWHGRQVHYHUDOSUHOLPLQDU\
GHVLJQVRIWKHJDXJHXVLQJILQLWHHOHPHQWPRGHOVWRH[DPLQHWKHSHUIRUPDQFHRIHDFK
GHVLJQ7KHVKDSHWRWDOKHLJKWDQGZLGWKRIWKHJDXJHDQGWKHOHQJWKRIWKH)H*D
SODWHZHUHPRGLILHGZLWKGLIIHUHQWRSWLRQV)LQDOO\SUHGLFWHGUHVXOWVIURPWKHQXPHULFDO
PRGHOVZHUHDQDO\]HGWRSURSRVHUHFRPPHQGDWLRQVIRUIXUWKHURSWLPDOGHVLJQ

 3ULQFLSOHDQGFRQILJXUDWLRQRIWKHGHYLFH

0DJQHWRVWULFWLRQLVDSURSHUW\RI)H*DDOOR\WKDWFDXVHVWKHPWRFKDQJHWKHLUVKDSH
GXULQJWKHSURFHVVRIPDJQHWL]DWLRQ7KHRSHUDWLQJSULQFLSOHRIWKHGHYLFHLVEDVHGRQ
WKHLQYHUVHPDJQHWRVWULFWLYHHIIHFWWKDWWKHPDJQHWL]DWLRQFKDQJHVZLWKVWUHVVDVVKRZQ
LQ)LJ,IWKHUHDUHXQLIRUPWHQVLOHDQGFRPSUHVVLYHVWUHVVHVJHQHUDWHGLQWKHORQJL
WXGLQDOGLUHFWLRQRID)H*DSODWHWKHPDJQHWLFIOX[LVLQFUHDVHGRUGHFUHDVHGE\WKH
LQYHUVHHIIHFW7KHUHIRUHYROWDJHVDUHLQGXFHGLQWKHZRXQGFRLOVDURXQGWKH)H*D
SODWHGXHWRWLPHYDU\LQJPDJQHWLF¿HOGVDQGWKHYLEUDWLRQHQHUJ\LVKDUYHVWHG


)LJ,QYHUVHPDJQHWRVWULFWLYHHIIHFW

7KHLGHDRIWKHSURSRVHGVWUDLQJDXJHFDPHIURPDFRQYHQWLRQDO3LVKDSHGGLVSODFH
PHQWWUDQVGXFHUZKLFKLVXWLOL]HGWRPHDVXUHWKHFUDFNRSHQLQJGLVSODFHPHQWRFFXUULQJ
ZLWKLQHDFKJDXJHOHQJWKRQWKHVXUIDFHRIWKHFRQFUHWH7ZRSUHOLPLQDU\GHVLJQVRI
WKHVWUDLQJDXJHZHUHFRQVLGHUHGVHH)LJVD±E'HVLJQ)LJDLVDFRPEL
QDWLRQRIDIODWSODWHRILURQJDOOLXPDOOR\ DQGD3LVKDSHGVWDLQOHVVVWHHOIUDPHWKH
IRUPHUDWWDFKHGWRWKHODWWHU5HJDUGLQJ'HVLJQWKH)H*DDOOR\LVPRXQWHGGLUHFWO\
WRWZRYHUWLFDOSODWHVRQERWKVLGHVDQGDFWVDVDWRSSODWH)LJD7KHVL]HRIVWDLQ
OHVVVWHHOEDVHSODWHLVIL[HGZLWKDZLGWKRIPPDQGDOHQJWKRIPPZKLOHWKH
ZLGWKDQGWKLFNQHVVRIWKH)H*DSODWHDUHPPDQGPPUHVSHFWLYHO\)RU'HVLJQ
WKHWKLFNQHVVRIWKHVWDLQOHVVVWHHOEDVHSODWHVYHUWLFDOSODWHVDQGWKHWRSSODWHDUH
PPPPDQGPPUHVSHFWLYHO\)RU'HVLJQWKHWKLFNQHVVRIDOOVWDLQOHVVVWHHO
SODWHVLVDVVXPHGWREHPPWRLQFUHDVHWKHRYHUDOOULJLGLW\RIWKHIUDPH,QWKLVVWXG\
WKHHIIHFWLYHQHVVRIDGHVLJQLVSUHVXPHGWREHDVVRFLDWHGZLWKDQLQFUHDVHLQWKHVWUDLQ
PHDVXUHGRQWKH)H*DSODWH
Parametric numerical study on a novel energy harvester … 163

 
D'HVLJQ      E'HVLJQ
)LJD±E7ZRSUHOLPLQDU\GHVLJQVRIWKHSURSRVHGVWUDLQJDXJH


)LJ$QDO\VLVDSSURDFK

 3DUDPHWULFVWXG\

 $QDO\VLVPRGHODQGPHWKRG
7R DQDO\WLFDOO\ LGHQWLI\ WKHVWUDLQEHKDYLRURI WKH )H*DSODWH WKLV VWXG\ SHUIRUPHG
QXPHULFDODQDO\VHVXVLQJ',$1$7RUHSURGXFHWKHDFWXDOZRUNLQJVWDWHIRUWKH
WZR SUHOLPLQDU\ GHVLJQV RI WKH VWUDLQ JDXJH WKH DQDO\WLFDO PRGHO HQFRPSDVVHG WZR
DQDO\VLVDSSURDFKHV5HJDUGLQJWKHHOHPHQWW\SHVXVHGLQWKHDQDO\VLVDSSURDFK)LJ
WKHVWDLQOHVVVWHHOWRSSODWHWKHYHUWLFDOSODWHVDQGWKH)H*DSODWHZHUHPRGHOHGDV
IODW VKHOO HOHPHQWVZKHUHDV WKUHHGLPHQVLRQDO PHPEUDQH HOHPHQWV ZHUH XWLOL]HG IRU
WKHPRGHOVRIWKHVWDLQOHVVVWHHOEDVHSODWHVZKLFKZHUHH[SHFWHGWRVXSSRUWRQO\WUDQV
ODWLRQDO GHJUHHV RI IUHHGRP DQG LQSODQH ORDGV 7KH VWUDLQ JDXJHZDV FRQVWUDLQHG DW
IRXUFRUQHUVE\OLQHDUVWLIIQHVVVSULQJV7KHVSULQJSUHYHQWHGWUDQVODWLRQDORQJWKHYHU
WLFDOGLUHFWLRQ=D[LVDQGLQWKHWUDQVYHUVHGLUHFWLRQ<D[LVDQGWKHVHVSULQJVZHUH
RQO\XVHGLQRUGHUWRUXQWKHVLPXODWLRQV)RUWKHPRGHOVRIWKHDQDO\VLVDSSURDFK
IODWVKHOOHOHPHQWVZHUHHPSOR\HGIRUDOOSODWHVDQGWKHGHYLFHZDVFRQVWUDLQHGDWIRXU
FRUQHUVE\YHU\ORZVWLIIQHVVVSULQJV7RREWDLQWKHGHIRUPDWLRQRIWKHGHYLFHWKHEDVH
SODWHVDUHVXEMHFWHGWRDSUHGHWHUPLQHGGLVSODFHPHQWRIPPWRZDUGVWZRRSSRVLWH
GLUHFWLRQVRQWKH;D[LV
,QDGGLWLRQIRUWKHQXPHULFDOPRGHORI'HVLJQULJLGOLQNVZLWKFRQQHFWHGWUDQVOD
WLRQVDW;<DQG=GLUHFWLRQVZHUHHPSOR\HGWRUHSURGXFHGWKHDGKHVLRQEHWZHHQWKH
)H*DSODWHDQGWKHVWHHOIUDPH5HJDUGLQJ'HVLJQWKHERQGLQJIRUFHEHWZHHQWKH
)H*DSODWHDQGWKHYHUWLFDOVWHHOSODWHZDVDVVXPHGWREHFRPSOHWH7KHOLQHDUVWDWLF
DQDO\VLVZDVHPSOR\HGIRUWKHDQDO\VLVFRPPDQG7KHVWUDLQYDOXHVRIWKH)H*DSODWH
164 T. M. Ha et al.

ZHUHUHFRUGHGDQGWKHQFRPSDUHGWRHYDOXDWHWKHSHUIRUPDQFHRIWKHSURSRVHGVWUXF
WXUHV
7KHVWDLQOHVVVWHHOSODWHVZHUHDVVXPHGDVOLQHDUHODVWLFPDWHULDOVZLWK<RXQJ¶VPRG
XOXVRI1PPDQGD3RLVVRQ¶VUDWLRRI7KHPDVVGHQVLW\ZDVGHILQHGDV
 JFP 0RUHRYHU WKH )H*D SODWH ZDV FRQVLGHUHG PDGH RI LURQJDOOLXP DOOR\
)H*DZLWK<RXQJ¶VPRGXOXVRIDURXQG1PP7KHPDVVGHQVLW\DQG
3RLVVRQ¶VUDWLRZHUHVHOHFWHGDVJFPDQGUHVSHFWLYHO\

 $QDO\VLVFDVHV
VFHQDULRVZHUHXVHGIRUWKHSDUDPHWULFVWXG\VHH7DEOH,QSDUWLFXODU6FHQDULRV
DQGZHUHHPSOR\HGIRULQYHVWLJDWLQJWKHHIIHFWLYHQHVVRIWKHWZRSUHOLPLQDU\GH
VLJQV LH 'HVLJQ  DQG 'HVLJQ  UHVSHFWLYHO\ 5HJDUGLQJ 6FHQDULRV   WKH WRWDO
KHLJKWRIWKHGHYLFHZDVLQFUHDVHGIURPPPWRPPDQGPPUHVSHFWLYHO\
6FHQDULRVZHUHVXSSRVHGWRSURYLGHWKHHIIHFWRIWKHWRWDOZLGWKDVFRPSDUHGZLWK
6FHQDULRV0RUHRYHUWKHDQDO\VLVDSSURDFKZDVDSSOLHGIRU6FHQDULRV±6SH
FLILFDOO\6FHQDULRV±ZHUHDOVRXWLOL]HGWRLQYHVWLJDWHWKHYDULDWLRQWUHQGRIWKHVWUDLQ
RI)(*DSODWHRZLQJWRWKHFKDQJHLQWKHWRWDOKHLJKW6FHQDULRZDVFRQVLGHUHGIRU
HYDOXDWLQJWKHHIIHFWRIWKHOHQJWKRI)H*DSODWHE\UHGXFLQJWKLVSDUDPHWHUWRPP
ZKLFKLVPPOHVVWKDQWKHVFHQDULRVSUHYLRXVO\DQDO\]HG

7DEOH$QDO\VLVVFHQDULRVIRUSDUDPHWULFVWXG\

/HQJWKRI
+HLJKW :LGWK $QDO\VLV
)H*DSODWH 'HVLJQ
PP PP PHWKRG
 PP
6FHQDULR     $SSURDFK
6FHQDULR     $SSURDFK
6FHQDULR     $SSURDFK
6FHQDULR     $SSURDFK
6FHQDULR     $SSURDFK
6FHQDULR     $SSURDFK
6FHQDULR     $SSURDFK
6FHQDULR     $SSURDFK
6FHQDULR     $SSURDFK
6FHQDULR     $SSURDFK

 5HVXOWVDQGGLVFXVVLRQ
(IIHFWRIWKHSUHOLPLQDU\GHVLJQV7KHVWUDLQUHVXOWVRIWKH)H*DSODWHZHUHREWDLQHG
DV—‫ܭ‬DQG—‫ܭ‬IRU6FHQDULRVDQGUHVSHFWLYHO\'LYLGLQJWKHVWUDLQRIWKH
)H*DSODWHE\WKHUHODWLYHVWUDLQRQ;GLUHFWLRQRFFXUULQJZLWKLQWKHJDXJHZLGWK\LHOGV
Parametric numerical study on a novel energy harvester … 165

DQGZKLFKZHUHFRQVLGHUHGWREHWKHHIIHFWLYHUDWHRIHDFKGHVLJQPRGHO
HPSOR\HGLQWKLVVWXG\7KHUHIRUHWKHREWDLQHGGDWDRI6FHQDULRZDVDSSUR[LPDWHO\
WLPHVJUHDWHUWKDQWKHLGHQWLILHGYDOXHRI6FHQDULR,QRWKHUZRUGVZLWKWKHVDPH
SUHGHWHUPLQHGGLVSODFHPHQWRQWKH;D[LVWKHSHUIRUPDQFHRIWKHSUHOLPLQDU\GHVLJQ
1RZDVVLJQLILFDQWO\KLJKHUWKDQWKDWRIWKHGHVLJQ1R

(IIHFWRIWKHWRWDOKHLJKWRIWKHGHYLFH7KHWRWDOKHLJKWRIWKHGHYLFHZDVLQFUHDVHG
IURPPPWRPPDQGPPDQGWKHQWKHFKDQJHLQWKHVWUDLQUHVXOWVZDVPHDV
XUHGDQGSORWWHGLQ)LJDDORQJZLWKWKHFRUUHVSRQGLQJHIIHFWLYHUDWHV7KHUHVXOWV
RI6FHQDULRVDQGVKRZWKDWWKHLQFUHDVHLQWKHWRWDOKHLJKWRIWKHGHYLFHKDGD
QHJDWLYHUHODWLRQVKLSZLWKWKHLQFUHDVHVLQWKHVWUDLQRIWKH)H*DSODWHDQGWKHHIIHFWLYH
UDWH0RUHRYHUWKHUHVXOWVZHUHVLPLODUWRWKRVHREWDLQHGZLWKUHVSHFWWR6FHQDULRV
XVLQJWKHDQDO\VLVDSSURDFK

(IIHFWRIWKHWRWDOZLGWKRIWKHGHYLFH$VFDQEHVHHQLQ)LJEWKHLQFUHDVHLQWKH
WRWDOZLGWKKDGQRVLJQLILFDQWHIIHFWRQWKHVWUDLQYDOXHRIWKH)H*DSODWHZKHUHDVLW
KDGDSRVLWLYHUHODWLRQVKLSZLWKWKHHIIHFWLYHUDWH7KHH[SODQDWLRQRIWKLVRXWFRPHLV
WKDWZKHQWKHJDXJHZLGWKLQFUHDVHGZLWKWKHVDPHSUHGHWHUPLQHGGLVSODFHPHQWRI
PPWRZDUGVWZRRSSRVLWHGLUHFWLRQVRQWKH;D[LVWKHUHZDVDGHFUHDVHLQWKHUHODWLYH
VWUDLQRFFXUULQJZLWKLQWKHJDXJHZLGWKOHDGLQJWRWKHLQFUHDVHLQWKHHIIHFWLYHUDWH

   


  

(IIHFWLYHUDWH

(IIHFWLYHUDWH
  

6WUDLQȝ‫ܭ‬

6WUDLQȝ‫ܭ‬

   


  

6WUDLQRI)H*DSODWH 6WUDLQRI)H*DSODWH
  
 (IIHFWLYHUDWH
(IIHFWLYHUDWH
   
         
+HLJKWPP
  :LGWKPP

D$QDO\VLVUHVXOWVRI6FHQDULRVDQGE$QDO\VLVUHVXOWVRI6FHQDULRVDQG
)LJD±E(IIHFWRIKHLJKWDQGZLGWKRIVWUDLQJDXJHRQWKHVWUDLQYDULDWLRQRI)H*DSODWH

 
 6WUDLQRI)H*DSODWH
(IIHFWLYHUDWH 

(IIHFWLYHUDWH



6WUDLQȝ‫ܭ‬


 






 
 
6FHQDULR

)LJ(IIHFWRIWKHOHQJWKRI)H*DSODWHRQLWVVWUDLQYDULDWLRQ
166 T. M. Ha et al.

(IIHFWRIWKHOHQJWKRI)H*DSODWH)URP)LJWKHVWUDLQRIWKH)H*DSODWHZDV
REWDLQHGDV—‫ܭ‬DQG—‫ܭ‬DQGWKHFRUUHVSRQGLQJHIIHFWLYHUDWHVZHUHHVWLPDWHG
WREHDERXWDQGIRU6FHQDULRVDQGUHVSHFWLYHO\+HQFHDVHDFK
PRGHOZDVXQGHUWKHVDPHGHIRUPDWLRQRFFXUULQJZLWKLQWKHJDXJHZLGWKWKHUHZDVDQ
LQFUHDVHLQWKHHIIHFWLYHUDWHZLWKUHVSHFWWRWKHGHFUHDVHLQOHQJWKRIWKH)H*DSODWH

 &RQFOXVLRQV

7KLVVWXG\SURSRVHVDPLFURHQHUJ\KDUYHVWLQJVWUDLQJDXJHXWLOL]LQJLURQJDOOLXPDO
OR\FDSDEOHRIJHQHUDWLQJHOHFWULFDOHQHUJ\IURPG\QDPLFVWUDLQUHVSRQVHV3DUDPHWULF
VWXGLHVZHUHFDUULHGRXWE\FKDQJLQJWKHVKDSHKHLJKWDQGZLGWKRIWKHGHYLFHWRH[
DPLQHLWVSRVVLEOHHIIHFWVRQWKHGHIRUPDWLRQEHKDYLRURIWKHSURSRVHGVWUXFWXUH7KH
PDLQFRQFOXVLRQVGUDZQIURPWKLVVWXG\DUHDVIROORZV

x 7KHSHUIRUPDQFHRIWKHSUHOLPLQDU\GHVLJQZDVVLJQLILFDQWO\KLJKHUWKDQWKDWRI
WKHGHVLJQ
x 7KHLQFUHDVHLQWKHWRWDOKHLJKWRIWKHGHYLFHKDGDQHJDWLYHUHODWLRQVKLSZLWKWKH
LQFUHDVHVLQWKHVWUDLQRIWKH)H*DSODWHZKLOHWKHLQFUHDVHLQWKHWRWDOZLGWKKDGQR
VLJQLILFDQWHIIHFWRQWKHVWUDLQYDULDWLRQRIWKHSODWH
x 7KHUHZDVDQLQFUHDVHLQWKHHIIHFWLYHUDWHZLWKUHVSHFWWRWKHGHFUHDVHLQOHQJWKRI
WKH)H*DSODWH
7KHUHVXOWVLQWKHSUHVHQWVWXG\DUHREWDLQHGSXUHO\IURPWKHQXPHULFDODQDO\VHVWKLV
ZRUNLVDVWDUWSRLQWIRUDVLPSOHQRYHOYLEUDWLRQDOSRZHUJHQHUDWLQJVWUDLQJDXJHXVLQJ
PDJQHWRVWULFWLYHPDWHULDO7KHIXUWKHUQXPHULFDOVWXG\VKRXOGEHFDUULHGRXWWRSUHGLFW
DFFXUDWHO\WKHEHKDYLRURIWKHSURSRVHGGHYLFHLQDFWXDOVLWXDWLRQVDQGFDQWDUJHWLP
SURYHPHQWVLQUHODWLRQWRWKHGHVLJQVDQGVLPXODWLRQPHWKRGV

$FNQRZOHGJPHQW

7KLVUHVHDUFKZDVSURPRWHGE\WKH5HJLRQDO,QQRYDWLRQ(FRV\VWHPV3URJUDPIURPWKH
0LQLVWU\RIHGXFDWLRQFXOWXUHVSRUWVVFLHQFHDQGWHFKQRORJ\-DSDQ7KHDXWKRUVZLVK
WR WKDQN WKH FRQFHUQHG SDUWLHV IRU WKHLU YDOXDEOH FROODERUDWLRQ VXEFRQVXOWDQWV DQG
VXSSRUW

5HIHUHQFHV
 8HQR7<DPDGD63HUIRUPDQFHRIHQHUJ\KDUYHVWHUXVLQJLURQJDOOLXPDOOR\LQIUHHYL
EUDWLRQ,(((7UDQVDFWLRQVRQ0DJQHWLFV±
 8HQR73HUIRUPDQFHRILPSURYHGPDJQHWRVWULFWLYHYLEUDWLRQDOSRZHUJHQHUDWRUVLPSOH
DQGKLJKSRZHURXWSXWIRUSUDFWLFDODSSOLFDWLRQV-RXUQDORI$SSOLHG3K\VLFV$

 8HQR70DJQHWRVWULFWLYHORZFRVWKLJKSHUIRUPDQFHYLEUDWLRQSRZHUJHQHUDWRU-RXUQDO
RI3K\VLFV&RQIHUHQFH6HULHV
 Numerical investigation of derailment loading on
composite fibre transoms for implementation in the
Sydney Harbour Bridge

Olivia Mirza[1], Maryam Hosseini[1] and Micah Fountain[1]

1
Western Sydney University, Locked Bag 1797, Penrith, 2751 New South Wales, Australia
o.mirza@westernsydney.edu.au, m.hosseini@westernsydney.edu.au
and 18376601@student.westernsydney.edu.au

Abstract. This paper investigates the behaviour of composite fibre transoms

subjected to derailment loading through a numerical analysis. The timber tran-
soms currently installed on the Sydney Harbour Bridge railway track are soon
to be replaced due to their low service lifespans and regular maintenance re-
quirements. Whilst steel and concrete alternatives achieve elongated service
lives, each possess unique limitations associated with installation and servicea-
bility. In light of this, recent studies have focused on the development of com-
posite fibre technology (CFT) transoms. Whilst previous studies have analysed
CFT transom response to static and quasi-static loading conditions, limited re-
search is available regarding the performance of composite fibre transoms,
when subjected to derailment induced impact loading. In order to bridge this
knowledge gap, numerical analysis have been utilised to investigate the internal
VWUHVV GLVWULEXWLRQ DQG GHWDLOHG IDLOXUH EHKDYLRXU RI :DJQHU¶V &)7 WUDQVRPV
under impact loading.

Keywords: Bridge rehabilitation, numerical analysis and strengthening.

1 Introduction

Transoms are a fundamental component of railway track systems, used to transfer

imposed axle loading to the track substructure. Since the initiation of railway systems
in 1925, transoms have been predominantly manufactured from hardwood timber [1].
Over 2.5 billion timber transoms are currently installed in worldwide railway systems,
including the railway track of the Sydney Harbour Bridge [2 and 3]. Whilst the work-
ability and mechanical properties of hardwood timber transoms appeal to railway
application, timber is highly susceptible to biological decay and chemical degradation
[4]. Consequentially, timber transoms require regular maintenance and often suffer
truncated service lives of less than 20 years [5]. In recent years, the development of a
more durable transom, which achieves similar performance and workability character-
istics to a timber transom, has presented a significant challenge to the railway industry
[6]. Extensive research has verified that neither steel nor concrete alternatives are
ideal for the replacement of timber bridge transoms, such as those installed on the

© Springer Nature Singapore Pte Ltd. 2020 167

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_23
168 O. Mirza et al.

Sydney Harbour Bridge. Whilst studies have analysed CFT transom response to static
and quasi-static loading conditions, limited research is available, regarding the per-
formance of composite fibre transoms, when subjected to derailment induced impact
loading [7]. In order to bridge the existing knowledge gap, this paper presents a nu-
merical analysis of the behaviour of WagneU¶V &)7 WUDQVRPV XQGHU LPSDFW ORDGLQJ
conditions. ABAQUS finite element software was utilised to investigate the internal
VWUHVV GLVWULEXWLRQ DQG GHWDLOHG IDLOXUH EHKDYLRXU RI :DJQHU¶V &)7 WUDQVRPV ZKHQ
subjected to impact loading.

2 Finite Element Modelling

ABAQUS finite element analysis software was utilied to investigate the effects of
GHUDLOPHQW LPSDFW ORDGLQJ RQ :DJQHU¶V &)7 WUDQVRPV 7ZR WUDQVRP PRGHOV ZHUH
constructed to compare the performance of different connectors. They are Ajax One-
side and Lindapter Hollobolt shear connectors within the composite fibre transoms.

2.1 Material Properties

:DJQHU¶V&)73URSHUWLHV
$V :DJQHU¶V &)7 H[KLELWV DQLVRWURSLF EHKDYLRXU WKH SURSHUWLHV ZHUH GHILQHG XVLQJ
WKH µHQJLQHHULQJ FRQVWDQWV¶ PHWKRG 7KH YDOXHV XVHG ZLWKLQ the ABAQUS model
were based on the manufacturers design information and are presented in Table 1.
Additionally, failure stress was defined in each principle direction. The values used
IRU:DJQHU¶V&)7PDWHULDOEHKDYLRXU ZLWKLQWKLVVWXG\DUHLQDJUHHPHQW Zith those
defined by [8].

Table 1:DJQHU¶V&)7properties.
Engineering Constant Value
Young's modulus
‫ܧ‬1 (MPa) 10800
‫ܧ‬2 (MPa) 10800
‫ܧ‬3 (MPa) 36300
3RLVVRQ¶VUDWLR
‫ݒ‬12 0.09
‫ݒ‬13 0.28
‫ݒ‬23 0.28
Shear modulus
‫ܩ‬12 (MPa) 4280
‫ܩ‬13 (MPa) 4280
Fig. 1. Stress strain relationships for steels
‫ܩ‬23 (MPa) 4280
properties.
Steels Properties
The stress±strain characteristics of steels properties include shear connectors and steel
stringers. A piecewise linear approach was found to be reasonably accurate to repre-
sent the stress±strain relationship. Their behaviour is initially elastic after which
yielding and strain hardening developed as shown in Fig. 1.
Geometry, Element Type and Mesh
Numerical investigation of derailment loading on composite fibre … 169

C3D8R elements were selected for all deformable instances within the numerical
model, as their quadrilateral geometry is well suited to the cellular nature of the
ABAQUS models mesh. All parts within this numerical analysis have been adequate-
ly partitioned to produce relatively uniform structural meshes for analysis. Additional-
ly, as impact loading results in high strain rates, mesh surrounding the zone of impact
is prone to excessive distortions. To prevent this phenomenon element deletion has
been enabled, which allows the model to ignore element distortion once the ultimate
element capacity has been achieved.
Contact Properties, Boundary Conditions and Load Application
Table 2. Contact Definitions. Frame Restraint
Surface Pair Interaction Impactor
Steel Plate
Impactor ± Steel Hard Frictionless
Plate
Steel Plate ± Tran- Tie Constraint
som Anti-crush
Anti-crush Inserts ± Tie Constraint Inserts
Transom Bolts
Transom ± bolts Tie Constraint Transom
Bolts ± I beam Tie Constraint Steel Beam
Transom ± Frame Hard Frictionless Fig. 2. Boundary conditions.
Transom ± I-Beam Hard Frictionless
Two contact definitions were used within the model. Tie constraints were defined
between parts which did not separate from each other over the duration of the test. For
instance, the anti-FUXVKLQVHUWVZHUHWLHFRQVWUDLQHGWRWKHLQVLGHRIWKH:DJQHU¶V&)7
hollow sections. Furthermore, surfaces which experienced relative movement were
modelled with a general explicit hard frictionless contact. This contact prevents sur-
face penetration and minimizes computation time by neglecting friction. The contact
properties of each surface to surface interaction are listed in Table 2.
The base of the I beam was defined as pin supported, to simulate interaction be-
tween the ground surface and restraints used to prevent uplift within the experiment.
All degrees of freedom except displacement in the vertical Y-direction were re-
strained for the rigid impactor to simulate the experimental restraints. The three re-
straints are presented in Fig. 2.
Load application was simulated by defining a rotary inertia for the rigid impactor
in conjunction with gravity loading. In order to reduce convergence time, the rigid
impactor was positioned 1mm above the transom with a predefined velocity corre-
sponding to a 2m drop height. To account for energy loss within the experimental
apparatus, the impactor initial velocity was manually lowered until the second impact
occurred at 0.82s, in agreement with the experimental data.

3 Results and Discussions

3.1 The Effect on Ajax Transoms

The deflection of the Ajax transom at the point of impact is presented in Fig. 3. The
peak downwards deflection after the initial impact was recorded as 44.38mm and
170 O. Mirza et al.

42.43mm during the two experimental trials. The peak deflection calculated by
ABAQUS was 44.38mm, corresponding to discrepancies of 0% and 4.4% respective-
ly. These insignificant discrepancies were attributed to experimental variation. The
maximum upwards deflection as calculated by ABAQUS was 37.26mm.
Stress contours are presented in Fig. 4. The elements shaded grey within Fig. 4, in-
dicate areas of local failure at a time interval of 0.015s. Whist maximum stress values
are exaggerated at tie constraint junctions, the model clearly demonstrates lateral
failure behaviour around the bolt holes and at the region in contact with the inner edge
of the top flange of the steel stringer. These calculations accurately predict the lateral
fractures observed within the experimental specimen displayed in Fig. 4.

Fig. 3. Comparison of Ajax transom FEA and Fig. 4. Comparison of calculated failure re-
experimental deflection. gions with experimental failure (Ajax).
3.2 The Effect on Lindapter Transoms

The deflection of the Lindapter transom at the point of impact is presented in Fig. 5.
The peak downwards deflection after the initial impact was recorded as 43.99mm and
45.66mm during the two experimental trials. The peak deflection calculated by
ABAQUS was 44.73mm, corresponding to a discrepancy of 1.65% and 2.05% respec-
tively. These minor variations were both less than 5% and therefore deemed accepta-
ble. The maximum upwards deflection according to the ABAQUS model was
40.77mm.
:DJQHU¶V &)7 SDQHOV DFKLHYH ODWHUDO IDLOXUH VWUHQJWK RI MPa. The elements
shaded grey within Fig. 6, indicate areas of local yielding at a time interval of 0.015s.
Whist maximum stress values are exaggerated at tie constraint junctions, the model
clearly demonstrates lateral yielding behaviour around the bolt holes and at the region
in contact with the inner edge of the top flange of the steel stringer. These calculations
accurately explain the corresponding lateral fractures observed within the experi-
mental specimens, displayed in Fig. 6.
Numerical investigation of derailment loading on composite fibre … 171

Fig. 5. Comparison of Lindapter transom FEA Fig. 6. Lindapter transom lateral fractures.
and experimental deflection.
3.3 Comparison of Transom Behaviour

Fig. 7. Ajax and Lindapter transom deflection comparison.

As illustrated in Fig. 7, both the Ajax and Lindapter transoms responded to the initial
impact in a similar manner. The Ajax transom experienced 0.8% lower downwards
peak deflection. Conversely, due to the smaller area of confinement, the Lindapter
transom model experienced an 8.6% higher peak in upward deflection The key differ-
ence in shear connector performance was evident during the free vibration of the tran-
som before the second impact. As displayed in Fig. 7, the Ajax transom experienced a
faster decrease in amplitude than the Lindapter transom. This is due to the greater
VXUIDFHDUHDRIWKH$MD[EROWLQFRQWDFWZLWKWKHLQVLGHRIWKH:DJQHU¶V&)7WUDQVRP
Whilst the deflection curves are similar, it is evident that the Ajax shear connector
produced more confinement than the Lindapter shear connector due to its larger sur-
face area. Due to the improved restraint of the Ajax shear connectors, larger stresses
developed around the Ajax connector bolt holes during upward deflection of the tran-
som. Ultimately this resulted in more severe cracking failure around the Ajax bolts
than the Lindapter bolts. This phenomenon is evident within the experimental tran-
soms, as displayed in Fig. 4 and Fig. 6.

4 Conclusions

ABAQUS finite element analysis software was utilised to conduct a numerical analy-
sis on the strXFWXUDOEHKDYLRXURI:DJQHU¶V&)7WUDQVRPVZKHQVXEMHFWHGWRGHUDLl-
172 O. Mirza et al.

ment impact loading. The model was validated through a comparison of deflection
data with that recorded during the experiment. Based on the finite element analysis
the following conclusions can be made:
‡ The model predicted peak deflection with discrepancies of less than 7%.
‡ The model predicted lateral failure behaviour of both the Ajax and Lindapter tran-
VRPVDWWKHMXQFWLRQEHWZHHQWKH:DJQHU¶V&)7SDQHODQGIODQJHRIWKHVWHHOVWULQJHU
This failure was observed within the experimental specimens.
‡ Whilst the deflective behaviour of each transom was similar, the Ajax transom
experienced a faster decrease of amplitude and in turn higher stress concentrations
around its shear connectors, due to the larger area of surface contact.
‡ Whilst localised yielding was observed around the bolt holes, neither transom ex-
perienced global yielding.
7KH QXPHULFDO LQYHVWLJDWLRQ YHULILHG WKDW :DJQHU¶V &)7 WUDQVRPV DUH DEOH WR DGe-
quately withstand derailment impact loading. Due to their excellent derailment load-
ing capacity, long service lifespan, lightweight nature and ease of installation, Wag-
QHU¶V&)7WUDQVRPVDUHUHFRPPHQGHGIRULPSOHPHQWDWLRQZLWKLQWKHUDLOZD\V\VWHP
of the Sydney Harbour Bridge.

References
1. Manalo, A, Aravinthan, T, Karunasena, W & Ticoalu, A.: A review of alternative materi-
als for replacing existing timber sleepers. Composite Structures 92(3), 603-11 (2010).
2. Ticoalu, A, Aravinthan, T & Karunasena, W.: An investigation on the stiffness of timber
sleepers for the design of fibre composite sleepers. Proceedings of the 20th Australasian
Conference on the Mechanics of Structures and Materials (ACMSM 20), pp. 865-70
(2008).
3. Griffin, D, Mirza, O, Kwok, K & Kaewunruen, S.: Composite slabs for railway construc-
tion and maintenance: A mechanistic review. The IES Journal Part A: Civil & Structural
Engineering 7(4), 243-62 (2014).
4. Ferdous, W & Manalo, A.: Failures of mainline railway sleepers and suggested remedies ±
Review of current practice. Engineering Failure Analysis 44, 17-35 (2014).
5. Mateus, EP, Gomes da Silva, MDR, Ribeiro, AB & Marriott, PJ.: Qualitative mass spec-
trometric analysis of the volatile fraction of creosote-treated railway wood sleepers by us-
ing comprehensive two-dimensional gas chromatography. Journal of Chromatography A
1178(1), 215-22 (2008).
6. Ferdous, W & Manalo, A.: Failures of mainline railway sleepers and suggested remedies ±
Review of current practice. Engineering Failure Analysis 44, 17-35 (2014).
7. Talos, A, Mirza, O, Kirkland, B and Assey A.: Experimental investigation on composite
steel and precast reinforced concrete transom for Sydney Harbour Bridge under impact
loading. 13th International Conference on Steel, Space and Composite Structures, 31 Janu-
ary - 2 February 2018, Perth, Australia (CD ROM) (2018).
8. Mirza, O, Shill, SK & Johnston, J,: Performance of Precast Prestressed Steel-Concrete
Composite Panels Under Static Loadings to Replace the Timber Transoms for Railway
Bridge. Engineering Failure 19, 30-40 (2019)
 Finite Element Analysis of Reinforced Concrete beams
subjected to combined actions

Tuan-Anh NGUYEN1,2, Quang-Huy NGUYEN1 and Hugues SOMJA1

1 Structural Engineering Research Group - INSA de Rennes, 20 avenue des Buttes de Coesmes,
CS 70839, F-35708 Rennes Cedex 7, France
2 Section of Structural Engineering - University of Transport and Communication, Vietnam

quang-huy.nguyen@insa-rennes.fr

Abstract. An enhanced multi-fiber beam element suitable for the analysis of re-
inforced concrete members subjected to combined loadings is presented. The
model is developed using displacement-based formulation with small displace-
ment assumption. The section kinematics is based on the kinematic assumptions
of a two-node Timoshenko beam and enhanced by introducing additional degrees
of freedom at each section in order to take into account the warping phenomenon.
A system of fixed points is created and interpolated by Lagrange functions and
polynomials. In order to take into account the contribution of stirrups, a discreti-
zation of control sections into different regions following its material response is
applied. As a result, the basic assumptions of the Modified Compression Field
Theory with a secant-stiffness formulation is used to represent the constitutive
material model for reinforced concrete. The model is validated by comparing to
the theoretical formulations and several experimental tests. The simulations in-
clude a variety of monotonic load conditions under bending, shear and torsion for
specimens with rectangular section.

Keywords: 3D beam multi-fiber, multi-axial efforts, torsion, warping, rein-

forced concrete

1 Introduction

Most civil engineering structures are subjected to multi-axial actions. Thus, in order to
completely simulate a structure in real conditions, its model must be able to take into
account the multi-axial efforts, including the normal efforts (axial force and bending
moments) and the tangential efforts (shear forces and torsional moments). For rein-
forced concrete (RC) structural components, multi-axial stress conditions can be inves-
tigated using others special analysis methods, such as strut-and-tie models based on the
space truss analogy or sectional approach models. Since more than thirty year, the sec-
tional analysis approach has been developed with the use of fiber (or multi-fiber) dis-
cretization models associated with frame elements. The introduction of normal effects
in the non-linear analysis of RC fiber-frame elements has been widely developed and
successfully applied in the models of Scordelis [1] or Taucer & Filippou [2]. On the
other hands, the tangential effects, especially shear forces, have also been investigated

© Springer Nature Singapore Pte Ltd. 2020 173

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_24
174 T.-A. Nguyen et al.

in fiber models by many researchers following various modeling strategies, such as

Vecchio & Collins with a dual-section analysis method [3], Bentz with the longitudinal
stiffness method [4] or Saritas & Filippou with a fiber-frame model with fixed patterns
for the shear strain distribution [5].
However, taking into account the coupling between normal and tangential effects is
not straightforward and attracts many attentions of researchers with different ap-
proaches. Moreover, the inclusion of the torsional effect in the modeling of RC mem-
bers is also a complex task, as it requires special handling of the warping phenomenon.
Knowing that the plane-section theory becomes insufficient to represent the coupling
effects at sectional levels, Bairan & Mari proposed a superposition of the enhanced
warping-distortion field with the classical plane-section field [6]. A discretization of
cross-section into different regions according to its stress state was also regarded as an
effective approach by some researchers, such as Rahal [7] or Navarro-Gregori [8].
In this present work, the authors want to develop an enhanced 3D multi-fiber model
for RC members in the framework of displacement-based formulation, which is capable
of predicting the entire sectional response under bending-shear-torsion coupling actions
and representing the warping phenomenon under shear-torsion as well as its influence.
This proposed model draws inspiration from several fiber-frame models, takes into ac-
count their advantages such as the superposition of displacement field in the model of
Bairan & Mari, the sectional discretization presented by Navarro-Gregori and the inter-
polation of Lagrange polynomials for warping deformation, proposed in the model of
Le Corvec [9]. The effectiveness of the proposed model are validated by comparing its
results with the analytical, numerical and experimental ones.

2 Enhanced 3D beam FE formulation

The proposed model is developed base on a 2-node Timoshenko beam, with 12 DoFs
in each element. Otherwise, several additional DoFs will be added in order to take into
account the warping phenomenon of torsional effects.

2.1 Section kinematic

Using multi-fiber finite element approach, the proposed model is described as a frame
element with a set of cross-sections along its longitudinal axis and a system of integra-
tion points (of fiber) at each cross-section. As mentioned above, the plane-section the-
ory is insufficient to reproduce the complete sectional deformation under shear and tor-
sional effects (Fig. 1a). According to the proposition of Bairan & Mari, the material
displacement field (with 3 components (U,V,W)) is decomposed into two domains: a
classical field describing the rigid body motion, consistent with the plan-section beam
theory; and an enhanced field referring to warping-distortion phenomena (Fig. 1b):

ܷ௠ ‫ݑ‬ሺ‫ݔ‬ሻ െ ‫ߠݕ‬௭ ሺ‫ݔ‬ሻ ൅ ‫ߠݖ‬௬ ሺ‫ݔ‬ሻ ൅ ܷ ௘ ሺ‫ݔ‬ǡ ‫ݕ‬ǡ ‫ݖ‬ሻ

௣
ࢊ௙௠ ൌ ࢊ௙ ൅ ࢊ௙௘ ௠
ൌ ൭ܸ ൱ ൌ ቌ ‫ݒ‬ሺ‫ݔ‬ሻ െ ‫ߠݖ‬௫ ሺ‫ݔ‬ሻ ൅ ܸ ௘ ሺ‫ݔ‬ǡ ‫ݕ‬ǡ ‫ݖ‬ሻ ቍ (1)
௠
ܹ ‫ݓ‬ሺ‫ݔ‬ሻ ൅ ‫ߠݕ‬௫ ሺ‫ݔ‬ሻ ൅ ܹ ௘ ሺ‫ݔ‬ǡ ‫ݕ‬ǡ ‫ݖ‬ሻ
Finite Element Analysis of Reinforced Concrete beams … 175

Fig. 1. a) Section deformation under normal efforts (axial, bending) and tangential efforts
(shear, torsion). b) Decomposition of a material axial displacement.

In this present work, only warping is taken into account, so the enhanced transversal
displacements Ve and We become zero. For the warping displacement Ue, according to
Saint-Venant theory it is a multiplication of the warping profile (representing by Saint-
Venant warping function) and the warping distribution (representing by the torsional
twist). However, in this model, these two parameters will be interpolated by a grid of
fixed points, which is defined in the axis direction x, with index i, and in the cross-
section followed by the direction y and z with index j and k respectively (Fig. 2).

Fig. 2. System of fixed interpolation points in the axis element and cross-section. The number
of interpolation points may be varied according to the need for precision.

Then, at any of these fixed points, with coordinate (xi,yj,zk), the enhanced displace-
ment values are defined as independent additional DoFs. These values are interpolated
to the total displacement field using Lagrange polynomials. The enhanced axial dis-
placement of a material point in the element is defined as follows:
௘
ܷ ௘ ሺ‫ݔ‬ǡ ‫ݕ‬ǡ ‫ݖ‬ሻ ൌ σ௜ ‫ܮ‬௜ ሺ‫ݔ‬ሻ σ௝ǡ௞ ܵ௝௞ ሺ‫ݕ‬ǡ ‫ݖ‬ሻ ‫ݑ‬௜௝௞ (2)

where ueijk is the enhanced axial displacement value; Sjk is the 2D Lagrange polynomials
and Li is the 1D Lagrange polynomials. Under the small displacement hypothesis, the
following kinematic relation is established to obtain the total strains of a material point:
௠ డ௎ ೘ ೔ డ௅ ሺ௫ሻ ௘
‫ߝۓ‬௫௫ ൌ డ௫ ൌ ߝ௫ െ ‫ߢݕ‬௭ ൅ ‫ߢݖ‬௬ ൅ σ௜ డ௫ σ௝ǡ௞ ܵ௝௞ ሺ‫ݕ‬ǡ ‫ݖ‬ሻ ‫ݑ‬௜௝௞
ۖ డ௎ ೘ డ௏ ೘ డௌೕೖ ሺ௫ሻ ௘
௠
ߛ௫௬ ൌ ൅ ൌ ߛ௬ െ ‫ߢݖ‬௫ ൅ σ௜ ‫ܮ‬௜ ሺ‫ݔ‬ሻ σ௝ǡ௞ ‫ݑ‬௜௝௞ (3)
డ௬ డ௫ డ௬
‫۔‬
ۖ ߛ ௠ ൌ డ௎ ೘ ൅ డௐ ೘ ൌ ߛ ൅ ‫ ߢݕ‬൅ σ ‫ ܮ‬ሺ‫ݔ‬ሻ σ డௌೕೖሺ௫ሻ ‫ݑ‬௘
‫ ە‬௫௭ డ௭ డ௫ ௭ ௫ ௜ ௜ ௝ǡ௞ ௜௝௞ డ௭

2.2 Material response and variational formulation

Once the strain vector ef is obtained at each material point, an appropriate material con-
stitutive relation is applied to determine the material stresses sf: sf=kf ef, with kf the
material stiffness matrix. The material constitutive model in this work is borrowed from
176 T.-A. Nguyen et al.

the idea of Navarro-Gregori, which is accorded to the sectional analysis and is formu-
lated general enough for being used for any shape of cross-section in further studies.
In this model, the cross-section is discretized in 3 zones according to the stress state
that they are subjected to (Fig. 3-a). In each zone, an appropriate constitutive model is
established to determine the regional stiffness matrix Ks,1D, Ks,2D or Ks,3D : an uniaxial
elastoplastic law for steel in the 1D-zone, a secant-stiffness-based formulation for con-
crete in the 2D-zone [10] and an extension into three dimensional of the 2D-formulation
for the 3D-zone [11]. Then, for the entire section, the sectional stiffness matrix is ob-
tained by summarizing all these regional stiffness matrices: Ks =Ks,1D +Ks,2D +Ks,3D.
Next, let the element be subjected by a virtual displacement įG, and the principle of
virtual work gives an equation of the element equilibrium:

ߜࢃ௜ ൌ ߜࢃ௘ ֜ ‫׮‬௏ ߜࢋ௙் ࢙௙ ܸ݀ ൌ ߜ்ࢗ௘ ࡽ௘ ൅ ‫׬‬௅ ߜࢊ௦ ࡼ௨ (4)

where the virtual internal work is represented by the left-hand side, while the virtual
external work is expressed by the right-hand; qe is the nodal displacement; Qe is the
external nodal forces and Pu is the external uniform loading. Then, the element equilib-
rium are established with the definition of the element stiffness matrix Ke:

ࡷ௘ ࢗ௘ ൌ ࡽ௘ ൅ ‫׬‬௅ ࡺ௦ ࡼ௨ ݀‫ݔ‬ (5)

3 Numerical examples

3.1 Beam subjected to torsion in linear elastic domain.

In order to represent the influence of the warping phenomenon on the torsional re-
sponse, a simple beam subjected to pure torsion at mid span is investigated (Fig. 3-b).
The twist angle calculated by the model gives a very good correlation comparing to the
theoretical formulation and the result of software RDM7 (Table 1). In this table, we can
see that neglecting the effect of warping will have a strong influence on the twist angle.
The warping profile of cross-section is represented in Fig. 4-a and gives a good
demonstration of the phenomenon. In Fig. 4-b,c, the shear stress distribution of the
classical plan-section field and the enhanced warping field are also plotted and we un-
derstand that due to the warping phenomenon, the value of shear stress is reduced, so
do the section rigidity, then the twist angle increases as a consequence. The total shear
stress distribution at mid-span section are represented in Fig. 4-d,e and it is completely
different to the distribution following the classical plan-section theory.

3.2 Beam subjected to combined shear-bending-torsion

In this case, the model capacity of taking into account the coupling of actions is ana-
lyzed. The TMV2 specimen, tested by Staley in 1978 is used as reference [12]. The
section dimensions, reinforcing details and material properties are shown in Fig. 3-a. In
Finite Element Analysis of Reinforced Concrete beams … 177

both curves representing the bending moment-defection and the torque-twist relation-
ship, a good correlation between numerical and experiment results is achieved (Fig. 5-
a,b). Moreover, in this example the model also represents its capacity when simulating
an element with various section disposition. The analysis time is taken about 10-15
minutes for a model built up from 2-4 elements.

Table 1. Mid span twist angle for cantilever beam under mid span torque.

Theory RDM7 No warping Proposed model

Twist angle (10-3 rad) 10.49 10.49 5.78 10.52
Relative error (%) 0 44.9 0.3

Fig. 3. a) Discretization of cross-section following material stress state. b) Simple beam sub-
jected to pure torsion and system of interpolation points for enhanced warping field. c) Outline
of test setup and section details in Staley tests [12].

Fig. 4. At mid-span section: a) Warping displacement profile under torsion. b, c) Warping and
classical plan-section shear stress distribution comparison. d, e) Total shear stress distribution.

Fig. 5. Specimen TMV2 in Staley test [12]: a) Bending moment versus deflection of mid-span.
c) Torque-twist curve.
178 T.-A. Nguyen et al.

4 Conclusion

In this paper, an enhanced multi-fiber FE beam model, taking into account the coupling
of actions, the warping phenomenon and the contribution of stirrups in a reinforced
concrete element, has been developed. The coupling between normal forces and shear
and, in particularly, torsion is done with a simple formulation and a reasonable compu-
tational efficiency.
The model is formulated in displacement-based framework, however, the sectional
model is general enough to apply in any finite element model or formulation, such as
force-based or flexibility-based. The nonlinear geometry has also been included in the
model and will be introduced in another specific paper. The numerical-experimental
comparison shown validates the model reliability and the capacity to determine the ex-
perimental behavior of RC beams under combined shear-torsion-flexion actions.

References
1. Scordelis, A. C.: Computer Models for Nonlinear Analysis of Reinforced and Prestressed
Concrete Structures. In: PCI Journal, vol. 29, pp. 116-135 (1984).
2. Taucer, F., Filippou, F. C. : A fiber beam-column element for seismic response analysis of
reinforced concrete structures. In: Technical Report, Earthquake Engineering Research Cen-
ter, University of California, Berkeley (1991).
3. Vecchio, F. J., Collins, M. P.: Predicting the response of reinforced concrete beams sub-
jected to shear using modified compression field theory. In: ACI Structural Journal, vol. 85,
pp. 258-268 (1988).
4. Bentz, E. C.: Sectional Analysis of Reinforced Concrete members. Ph.D thesis, University
of Toronto, Toronto, Canada (2000).
5. Saritas, A., Filippou, F. C.: A beam finite element for shear critical reinforced concrete
beams (2006).
6. Bairan J. & Mari, A.: Multiaxial-coupled analysis of RC cross-sections subjected to com-
bined forces. In: Engineering Structures, vol. 29, pp. 1722-1738 (2007).
7. Rahal, K.: Combined Torsion and Bending in Reinforced and Prestressed concrete beams
using simplified method for combined stress-resultants. In: ACI Structural Journal, vol. 104,
pp. 402-411 (2007).
8. Navarro-Gregori, J., Miguel, P., Fernandez, M. & Filippou, F.: A 3d numerical model for
reinforced and prestressed concrete elements subjected to combined axial, bending, shear
and torsion loading. In: Engineering Structures, vol. 29, pp. 3404-3419 (2007).
9. Le Corvec, V.: Nonlinear 3D frame element with multi-axial coupling under consideration
of local effects. Ph.D thesis, University of California, Berkeley (2012).
10. Vecchio, F. J., Collins, M. P.: The modified compression-field theory for reinforced concrete
elements subjected to shear. In: Journal of the American Concrete Institute, vol. 83, pp. 219-
231 (1986).
11. Vecchio, F. J., Selby, R. G.: Toward compression-field analysis of reinforced concrete sol-
ids. In: Journal of Structural Engineering, vol. 117, p. 1740-1758 (1991).
12. Staley, R. F.: Torsion, bending and shear in rectangular reinforced concrete beams ± an ex-
perimental study. In: Report no. Uniciv R-178, The University of New South Wales, Sydney
(1978).
 Resistance of Cross Laminated Timber members under
axial loading - a review of current design rules

Xin Li1, Mahmud Ashraf*1, Mahbube Subhani1, Bidur Kafle1 and Paul Kremer1, 2
1
School of Engineering, Deakin University, Geelong Waurn Ponds, VIC 3216, Australia
2
XLam Australia and New Zealand
xinli@deakin.edu.au

Abstract. Cross laminated timber (CLT) has become an important type of

engineered timber product due to its superior material properties, making it
suitable for mass timber construction. CLT is manufactured from naturally
grown sources, and hence, it offers significant environmental benefit within the
construction sector in comparison to its traditional counterparts such as concrete
and steel. As a naturally grown material, timber products possess some inherent
variabilities in material properties. These uncertainties make it difficult to
develop universally accepted design rules for timber products, as the basic
material properties may vary significantly based on its origin. This paper
presents a review of current design rules for CLT under axial loads. Design
procedures outlined in Eurocode, North American and Chinese standards as
well as other relevant CLT handbooks have been thoroughly investigated.
Experimental evidences available from literature have been used to compare the
performance of those design guidance in predicting axial resistance of CLT
members, where appropriate. Overall, it was observed that despite having
considerable variations in mechanical properties, similar reliability concepts are
adopted for specifying characteristic values, design values as well as for
devising design rules for CLT structures. Once the design rules are critically
evaluated based on test evidences, some design rules for CLT members under
ultimate limit states have been summarised herein.

Keywords: Cross laminated timber, axial loading, design rules.

1 Introduction

Cross laminated timber (CLT), a prefabricated engineered timber product, is made of

at least three orthogonal layers of sawn boards or wood-based boards that are
arranged and laminated crosswise to each other using structural adhesives [1]. CLT
was first developed in Europe [2] and has been recognized for its obvious beneficial
properties around the world including US, Canada, China, Australia and New
Zealand, to name a few. Due to its unique orthogonal layup, CLT is reported to
withstanding considerable in-plane and out-of-plane stresses making it suitable for
mid-rise construction. From a design point of view, carrying capacities of axial loaded
CLT members are strongly dependent on their compressive and tensile performances.

© Springer Nature Singapore Pte Ltd. 2020 179

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_25
180 X. Li et al.

However, apart from some research on out-of-plane compression [3-7], the

performance of CLT when subjected to in-plane compression and tension have not
been thoroughly investigated. In Europe, although properties and design for CLT have
been regulated via international European Technical Approvals (ETA) since 2006,
and the first European standard for CLT (EN 16351) was released in 2015 but CLT
has not been adopted in the current timber design code (Eurocode 5). Newly released
North American [1, 8] and Chinese standards (GB) [9] include specific section for CLT,
but only regulate limitations related to production and construction. For now, there is
no universally accepted test procedures or design rules for CLT products.
In this paper, available test and design procedures outlined in European guidance
will be thoroughly investigated and compared with those proposed in National Design
Specification for wood constructions (NDS 2018) [8], GB 50005 [9] and other relevant
CLT handbooks [2,10]. The general design concept will be explained followed by a
review of test configurations and characteristic properties of CLT under axial loading.
Available design guidance for relevant CLT members shall be summarized from the
open literature, whilst conclusions at the end.

2 General design concept

Although design rules for CLT has not been covered by most national standards, a
general design concept has been developed and is well accepted by the relevant
designers. Based on engineering safety, the reliability of structure [8-11] can be ensured
by considering possible influencing factors. Therefore, characteristic values, design
values and design rules for CLT structures in different regulations are somewhat
comparable. Basic design starts from obtaining the characteristic values of strength,
Rk (fk), followed by the determination of design value of resistance, Rd, by taking other
influence factors into account. The very basic design principle in the ultimate limit
state is to ensure that the respective design value of stress, Nd is equal to or smaller
than the respective design value of resistance, Rd as shown in Eq. 1:
Nd d Rd (1)

3 Characteristic properties of CLT

Characteristic properties of axially loaded CLT members are generally determined by

two methods ± experimental method using standard testing procedure and empirical
method using calibrated relationships. Most of the experimental evidences are based
on European CLTs, and those outcomes are discussed in details in the following
sections.

3.1 Test configurations

Available testing methods for CLT are summarized in Table 1. Testing set-up
procedures for CLT panels are outlined in EN 408 unless different requirements are
given in EN 16351. CLT panels subjected to compression perpendicular to plane are
Resistance of Cross Laminated Timber members … 181

suggested to be tested following EN 16351. The height of the sample shall be the
overall thickness of the tested panel. In the up-to-date draft prEN 16351:2018, further
provisions on using 150 mm square cross-section samples have been made. It is,
however, worth noting that in aforementioned standards, there is no specified
guideline for testing CLT under in-plane compression. More importantly, there is no
test standard for CLT panels under tension primarily due to lack of research [12].
Table 1. Available testing methods for CLT
Out-of-plane In-plane
Categories Standards
Compression tension Compression tension
EN 16351/ prEN 16351 ×
CLT PRG 320
ISO 16696-1

3.2 Characteristic values of strength, Rk (fk)

Out-of-plane compression. The most comprehensive experimental investigation on

uniformly stressed full-section European specimens under out-of-plane compression
was conducted by Halili [3]. A total of 217 prisms were tested with primary focus on
the influence of annual ring orientation, layup and ratio of the thicknesses of adjacent
layers. Obtained test results indicated relatively higher elastic and strength properties
in CLT than homologous GLT duHWR³ORFNLQJHIIHFWV´
In Eurocode 5, a load introduction dependent coefficient kc,90 is recommended to
consider partial loaded contributions to the higher bearing resistances and stiffness.
Accepting this concept, Salzmann [4] focused on the positions of point loads applied
on continuously supported CLT panels. Serano and Enquist [5] investigated the effects
of the position of line loads on linearly supported CLT panels for simulating CLT
slabs clamped by walls. They observed a close relationship between bearing capacity
and stress dispersions as well as the position of loading. Brandner and Schickhofer et
al. [6] conducted further research and analysis on system properties, and proposed a
stress dispersion model in which stress is spread at an angle of 45° (ĮL) in longitudinal
direction and at 15° (ĮT) in transverse direction. Thus, the correlation between CLT
compression prisms and CLT structural members in regards to load introduction and
load transmission (Fig. 1) are expressed as shown in Eq. 2. Thiel and Brandner [7]
provided recommended values for the coefficient kc,90,CLT.
fc,90,CLT = kc,90,CLT Â fc,90,CLT,prim (2)
The characteristic compressive strength perpendicular to plane of completely
stressed prism is stated in EN 16351 with fc,90,CLT,k = 3 N/mm2 for all types of wood, if
no test results are present.

Fig. 1. Load introduction and load transmission in compression perpendicular to grain stresses [6]
182 X. Li et al.

In-plane compression. CLT properties in regard to compression in plane have not

been sufficiently investigated. On a conservative basis, the effects of transverse area
are ignored. Fig. 2 illustrates the net area of a cross section within the parallel loaded
layers. CLT in-plane compressive strength is assumed as being equal to the bending
strength of the corresponding CLT strength class [12] or as shown in Eq. 3:
fc,0,CLT,net,k = ksys,c,0× fc,0,l,k (3)
where fc,0,l,k is the tension strength of timber lamella parallel to the grain direction;
ksys,c,0 is the coefficient to account for the systemic effect in tension due to the load
distribution of the lamellae [11,12].

(a) Net area in major direction (b) Net area in minor direction
Fig. 2. Net area in longitudinal layers and transverse layers

Tension. EN 14080 recommends that the tension strength of GLT perpendicular to

the plane should be considered as being equal to that of a single lamella. Due to very
limited research on CLT subjected to out-of-plane tension, it is suggested to be in
accordance with GLT, and may be expressed as shown in Eq. 4:
ft,90,CLT,k = ft,90,k (4)
Similar to in-plane compression, only the net area of cross section Anet within the
parallel loaded layers are considered in in-plane tension properties. And the same
concept of system factor [11,12] is used as shown in Eq. 5:
ft,0,CLT,net,k = ksys,t,0× ft,0,l,k (5)

4 Codified design parameters of CLT

4.1 Design value of resistance, Rd

The design value of resistance Rd in Eurocode is defined by taking the partial safety
factor Ȗm and the modification factor kmod into account, and is expressed as shown in
Eq. 6:
Rk
Rd kmod ˜ (6)
Jm

where, Ȗm is taken as 1.25 for CLT; kmod is used to consider the effect of load duration
and moisture content, and is taken as 0.8 for occupancy classes 1 and 2 [15]. In Chinese
standard [9], the design values of CLT are taken as equal to those obtained for singular
lamina used as the CLT outermost layers. In North American standard [1,8], reference
GHVLJQ YDOXHV VKDOO EH REWDLQHG IURP WKH FURVV ODPLQDWHG WLPEHU PDQXIDFWXUHU¶V
literature or code evaluation report. Detailed adjustment factors for each loading
condition are elaborated in NDS [8].
Resistance of Cross Laminated Timber members … 183

4.2 Comparison of characteristic values

Choosing a commonly used European homogeneous CLT with strength grade of
CL24 as a reference, comparisons among standards are summarized in Table 2.
Table 2. Characteristic values in standards (MPa)
Characteristic values of strength Rk Design values of resistance Rd
Categories Categories GB
Literature ETA GB NDS Literature ETA (TC13- NDS
B/C24)
fc,90,CLT,k 3.0 2.5 2.5 fc,90,CLT,d 1.92 1.6 1.8/2.4
fc,0,CLT,net,k 24 21 17.2 No fc,0,CLT,net,d 15.35 13.4 10/12.5 No
ft,90,CLT,k 0.5 - - details ft,90,CLT,d 0.32 - 0.47/0.63* details
ft,0,CLT,net,k 16 14 14 ft,0,CLT,net,d 10.24 9.0 7.0/7.5
Note: TC-13 refers to European spruce; C24 refers to imported European timber with strength grade of
C24;
*
For an engineered simplification, ft,90,CLT,d = 1/3 fv,0,CLT,d is assumed in GB.

Characteristic values of strength between ETA and GB are nearly the same. The
mismatching in fc,0,CLT,net,k is due to the different definitions between standards i.e.
fc,0,CLT,net,k = fm,CLT,k in ETA, while fc,0,CLT,net,k = fc,0,l,k in GB. When it comes to fm,CLT,k ,
GB gives a similar value of 23.3 N/mm2.
Design values of resistance in ETA and GB, for safety considerations, are typically
lower than those obtained from research. GB is slightly conservative than ETA, but
still comparable.

4.3 Design of CLT

The verification of out-of-plane compression must fulfil the following conditions as
shown in Eq. 7 and 8:
V c , 90 , d d kc , 90 ,CLT ˜ fc , 90 ,CLT , d (7)
N90 , d d kc , 90 ,CLT ˜ fc , 90 ,CLT , d ˜ Aef (8)
where, ıc,90,d is the design value of compression stress perpendicular to grain; N90,d is
the design value of compression force; Aef is the effective contact area [15]; kc,90,CLT is
the stress dispersion factor [15] as discussed in section 3.2. In case of CLT subjected to
in-plane compression, the actual compressive stress must not exceed the design value
of resistance. Only the net area is considered, as shown in Fig. 2. This has to satisfy
the conditions shown in Eq. 9 - 10:
V c ,0 , net , d d fc ,0 ,CLT , net , d (9)
N d d fc ,0 ,CLT , net , d ˜ Anet (10)
where, ıc,0,net,d is the design value of compression stress parallel to grain; Nd is the
design value of compression force.
There is no regulation in Europe for the case of CLT subjected to out-of-plane
tension. Similarly, NDS [8] suggests that design induces tension stress perpendicular to
plane should be avoided whenever possible. In case of CLT subjected to tension
184 X. Li et al.

parallel to plane, the actual tensile stress shall also be obtained from the net section
area, Anet and Eq. 11 and 12 must be satisfied:
V t ,0 , net , d d ft ,0 ,CLT , net , d
(11)
N d d ft ,0 ,CLT , net , d ˜ Anet
(12)
where, ıt,0,net,d is the design value of tensile stress parallel to grain; Nd is the design
value of tensile force.

5 Conclusions

As an innovative engineered bio-composite material, CLT has experienced rapid

development over the last decade, and is expected to be used widely in the
construction sector as an environmentally friendly alternative to the traditional
building material. However, as a naturally grown material, some inherent variabilities
in material properties make it difficult to develop widely accepted design rules, which
is a pre-requisite for wider use of CLT. This paper presented a review of current test
configurations and design rules for CLT under axial loads to identify the differences
and gaps in this field. As CLT was first produced in Europe, current Eurocode and
ETA provide most comprehensive technical details for CLT; more formal regulations
are expected to be included in the revision of Eurocode 5 (EN 1995-x-x). As to
material properties, characteristic values are comparable between ETA and GB.
Despite some guidance on testing procedures for out-of-plane stresses, extensive
research is required to characterise in-plane compression and tension behaviour of
CLT panels. Identified gaps will be address in an ongoing research project and will be
reported in due course.

References
1. PRG-320, Standard for performance-rated cross laminated timber. (2018).
2. Karacabeyli, E., Douglas, B.: CLT Handbook: cross laminated timber - U.S. edition (2013).
3. Halili, Y., Versuchstechnische Ermittlung von Querdruckkenngrößen für Brettsperrholz. Diploma
thesis, Graz University of Technology (2008)
4. Salzmann, C.: Ermittlung von Querdruckkenngrößen für Brettsperrholz (BSP). MaVWHU¶VWKHVLV*UD]
University of Technology (2010).
5. Serrano, E. and B. Enquist.: Compression strength perpendicular to grain in cross-laminated timber
(CLT). WCTE 2010. (2010).
6. Brandner, R.: Cross laminated timber (CLT) in compression perpendicular to plane: Testing,
properties, design and recommendations for harmonizing design provisions for structural timber
products. (2018).
7. Thiel, A.: ULS and SLS design of CLT and its implementation in the CLT designer, University of
Bath, Bath (2013).
8. National Design Specification for Wood Construction. American Forest & Paper Association (2018).
9. GB 50005. "Code for Design of Timber Structures." (2017).
10. Gagnon, S., Pirvu, C.: CLT Handbook: cross laminated timber - Canadian edition (2011).
11. Wallner-Novak, M., Koppelhuber, J., Pock, K.: Cross-Laminated Timber Structural Design - Basic
design and engineering principles according to Eurocode, proHolz Austria, Vienna, (2013).
12. Brandner, R.: Cross laminated timber (CLT): overview and development. European Journal of Wood
and Wood Products 74.3: 331-351, (2016).
 A current-state-of-the-art on design rules vs test
resistance of Cross Laminated Timber members
subjected to transverse loading

Xin Li1, Mahbube Subhani1, Mahmud Ashraf1, Bidur Kafle1 and Paul Kremer1, 2
1
School of Engineering, Deakin University, Geelong Waurn Ponds, VIC 3216, Australia
2
XLam Australia and New Zealand
mahmud.ashraf@deakin.edu.au

Abstract. Cross laminated timber (CLT) is an innovative and environmentally

sustainable engineered timber product which has superior in-plane and out-of-
plane bending strength compared to other conventional timber products. CLT
panels can be used as both wall and floor elements in low-rise to mid-rise
construction. Nevertheless, both shear and bending strength (along with other
material properties as well) of CLT panels can vary significantly depending on
the species used to manufacture these products. As a result, in-depth research
with respect to the variations in bending and shear properties of CLT panels
from different parts of the world is required in order to generate a common
standard. However, no standardised design procedures for CLT can be found in
the open literature, at this stage. The main purpose of this paper is to investigate
various available design rules outlined in Eurocode, North American standards,
Chinese standards and handbooks for CLT in regards to in-plane and out-of-
plane bending and shear design of panels subjected to transverse loading. In
addition to the theoretical review, available testing standards to obtain relevant
characteristic properties have been included. Lastly, the ultimate limit states
design method based on different theoretical models available for CLT
elements subjected to transverse loading (such as, gamma method, k-method,
shear analogy method, representative volume sub-element (RVSE) method and
in-plane beam method) are also discussed and summarized at the end.

Keywords: Cross laminated timber, transverse loading, design rules

1 Introduction

Cross laminated timber (CLT) is a relatively new engineering timber product. It

usually composes of an odd number of wood-based layers (at least three), which are
arranged and laminated crosswise to each other using structural adhesives [1]. Since
timber is strong in the grain direction, the cross arrangement of CLT enable it to
withstand considerable stresses both in-plane and out-of-plane. Therefore, they can be
used as both wall and floor elements in mid-rise construction [2].
In general, wall and floor elements are normally subjected to transverse loading
resulting in shear and bending responses in CLT panels. A series of research have

© Springer Nature Singapore Pte Ltd. 2020 185

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_26
186 X. Li et al.

been conducted with special focus on out-of-plane bending and shear [3,4] as well as
in-plane bending and shear [5-8]. Although majority of them are based on European
products, CLT has not been adopted in the current timber design code (Eurocode 5).
Newly released North American standards [1,9] and Chinese standard (GB) [10] include
specific section for CLT, but only regulate limitations related to production and
construction. In short, there is no universally accepted test procedures or design rules
for CLT products.
In this paper, various available design guidance for CLT panels subjected to in-
plane and out-of-plane bending and shear loading are investigated. The general design
concept will be explained as well as a review of characteristic properties and
analytical design of CLT will be thoroughly presented. Finally, conclusions are drawn
at the end.

2 Codified design parameters of CLT

2.1 General design concept

According to strength limit state design, the resistance of Rd should be greater than or
equal to the design action, Nd as shown in Eq. 1:
Nd d Rd (1)
The resistance Rd is determined from the characteristics strength value Rk (fk) by
considering possible influencing factors [11].

2.2 Characteristic values of strength, Rk (fk)

Characteristic value of transversely loaded CLT members can be generally

determined by standard testing or empirical approach. This section will briefly discuss
available testing standards and research outcomes reported recently within European
context giving special emphasis on local wood species (Norway Spruce etc.).

Available testing methods for CLT members subjected to transverse loading to

determine bending and shear strength are summarized in Table 1.
Table 1. Available testing methods for CLT
Out-of-plane In-plane
Shear Rolling Shear Shear Torsional
Standards Bending Bending
(bending) shear (bending) (net section) shear
EN 16351 × × × × × ×
prEN 16351 × × × × × ×
PRG 320 × × × ×
ISO 16696-1 × ×

North American standard PRG 320 [1] provides guidance to test bending and shear
elements through three or four point bending test, while ISO 16696-1 only describes
testing related to out-of-plane bending and shear. To date, EN 16351 is the most
comprehensive standard available for the characterisation of CLT panel. In 2018, the
draft prEN 16351 was released to further normalize the existing test configurations by
A current-state-of-the-art on design rules vs test resistance … 187

the acceptance of using wider samples and the replacement of shear test method for
net cross section.
European empirical approach for evaluating out-of-plane bending strength was
proposed by Jobstl et al. [3], as follows in Eq. 2:
f m , CLT , k
0.8
km, CLT u f c ,0, l , k (2)

where, fm,CLT,k is the out-of-plane bending strength; fc,0,l,k is the tensile strength of
timber lamella parallel to the grain direction; km,CLT is a coefficient related to the
effects at bending, difference between CLT and GLT, and the variation of basic
material [2,3].
For CLT loaded in out-of-plane direction, shear strength is assumed to be equal to
that reported for glue laminated timber (GLT) according to EN 14080 i.e. fv,CLT, k = 3.5
N/mm2 [2].
Out-of-plane bending of CLT element causes shear stresses along the grain
direction for the longitudinal layers, whereas in case of transverse layers, the shear
stresses are developed perpendicular to the grain. The shear stress in transverse layers
that act perpendicular to the grain is also called rolling shear, which highly depends
on wood species, width to thickness ratio (wl/tl ) in a single lamella and annual ring
orientation [3,4]. Ehrhart et al. [4] proposed a bi-linear approach for the estimation of
rolling shear strength, and suggested the typical values of fr,CLT, k = 1.4 N/mm2 and
fr,CLT, k = 0.8 N/mm2 for wl/tl •DQGwl/tl < 4, respectively.
CLT panels subjected to in-plane shear typically demonstrate three types of failure
mechanisms [5-8]: gross shear failure occurs due to the longitudinal shear failure when
the lamellae are not bonded sideways, net-shear failure in case of CLT with gaps and
cracks, and torsion failure at the gluing-interfaces between the orthogonal layers, as
shown in Fig. 1.

(a) Gross-shear failure (b) Net-shear failure (c) Torsion failure

Fig. 1. CLT element subjected to in-plane transverse forces [9]

Based on aforementioned researches, They proposed characteristic values of fv,gross,k

= 3.5 N/mm2, fv,net,k = 5.5 N/mm2 and fv,tor,k = 2.5 N/mm2 for each of those failure
modes, respectively.

2.3 Design value of resistance, Rd

The design value of resistance Rd in Eurocode, defined by considering the partial
safety factor Ȗm and the modification factor kmod, is expressed as:
R
Rd k mod ˜ k (3)
Jm
188 X. Li et al.

where, Ȗm is taken as 1.25 for CLT; kmod considers the effect of load duration and
moisture content and is taken as 0.8 for utilisation classes 1 and 2.
In Chinese standard [10], the design values of CLT are assumed to be equal to those
obtained from the singular lamellae of the outermost layer.
In North American standard, reference design values shall be obtained from the
FURVV ODPLQDWHG WLPEHU PDQXIDFWXUHU¶V OLWHUDWXUH RU FRGH HYDOXDWLRQ UHSRUW
Adjustment factors are considered to determine the bending and shear resistance
including load duration, service environment etc., and are elaborated in NDS [9].

3 Analytical Design of CLT

Experimental methods are the most accurate technique to determine the mechanical
properties of CLT, but lack versatility. Therefore, a number of analytical approaches
are proposed which are able to consider changes in layups and species of the wood.

3.1 Out-of-plane design

For bending elements, Euler-Bernoulli beam theory is the most commonly used
approach, but due to the shear deformation in CLT, this traditional theory cannot be
applied directly. At present, gamma method, k-method and shear analogy method are
available to determine the properties of CLT by knowing the properties of the
lamellae.
Gamma method was originally presented in Eurocode 5 and developed for beams
connected with uniformly spaced fasteners. By modifying and assuming the
ORQJLWXGLQDO OD\HUV RI WKH &/7 SDQHOV DV LQGLYLGXDO ³EHDPV´ DQG LQWURGXFLQJ D
connection efficiency factor to account for the shear deformation of the transverse
layer, it can be applied to CLT. This modified method has been adopted by European
Technical Approvals (ETA) [11]. In Chinese standard GB 50005 [10], similar approach
is accepted for CLT bending elements with a length to height ratio (l/h) greater than
10.
K-method was proposed by Blass and Fellmoser [12] to predict the strength and
VWLIIQHVV RI &/7 E\ FDOFXODWLQJ VSHFLILF IDFWRU ³k´ WR FRQVLGHU WKH HIIHFWV RI &/7
panel subjected to different loading conditions, such as out-of-plane bending as well
as in-plane bending, tension and compression. This method does not account for the
shear deformation in individual layers, but gives reasonable accuracy to the elements
with high span-to-depth ration (l/h •0) [12].
Shear analogy method seems to be the most accurate for CLT in terms of determining
out-of-plane bending and shear properties. It separates the multi-layer CLT into two
virtual beams with equal vertical deformation. Material properties of entire CLT can
be obtained by overlaying the performance of both beams [12]. This theory is not
limited by the number of layers within a panel, and has been adopted by PRG 320 [1].
Due to the page limits, aforementioned theoretical details are not elaborated in this
paper. Table 2 summaries typical experimental research from the literature and
compares relevant EI and ımax with theoretical calculations. All results show good
A current-state-of-the-art on design rules vs test resistance … 189

agreements between different methods. This is in line with previous research that for
span with l/h > 15, the discrepancies among these methods to calculate effective
stiffness and relevant strength are not significant [13].
Table 2. Test vs theoretical calculation
Section, mm EIeff,cal / EIeff,test ımax,cal/ımax,test
Reference L/h
(h × w × L) Gamma SA K Gamma SA K
M. He et al. 2018 175 × 305 × 4875 27.9 0.781 0.838 0.857 1.278 1.233 1.000
D.P. Hindman et al. 2015 175 × 305 × 4720 27.0 0.754 0.825 0.818 1.086 1.061 0.847
K. Sikora et al. 2016 72 × 288 × 1296 18.0 1.035 1.065 1.061 1.073 1.064 1.026
I.P. Christovasilis et al. 2016 132 × 250 × 2340 17.7 1.133 1.162 1.162 1.047 1.038 1.001

3.2 In-plane design

CLT bending elements under axial in-plane loads acts as deep beams for which the
aforementioned k-method can be implemented for design [12]. For in-plane shear
element, representative volume sub-element (RVSE) method [5] and in-plane beam
method [6,7] are developed.

RVSE method was proposed by Moosbruger and Bogensperger et al. [5]. This
analytical model introduced the smallest irreducible unit of CLT element (Fig. 2),
FDOOHG³UHSUHVHQWDWLYHYROXPHVXE-HOHPHQW596(´ to estimate the distinct effect of
the shear behaviours between single layers. Shear stress for each failure mechanism
can then be evaluated.

Fig. 2. Basic elements in RVSE [5] (left); shear stresses under gross shear, net shear and torsion [8] (right)

In-plane beam method was proposed by Flaig and Blaas [6]. This method assumed
transverse layers in CLT are strong enough to act as solid units. Therefore, Bernoulli-
Euler Beam theory can be used to calculate the gross and net shear stress. In this
method, shear distributions in both longitudinal and transversal layer (of three-layer
CLT as an example in Fig. 3) follow the parabolic function calculated with relevant
cross section; The maximum value of gross and net shear strength are able to be
calculated.
For torsion failure, initial theory [6] were further modified by Danielsson and
Serrano [7] by considering the influence of the element layup on the distribution of the
shear stress and uneven distribution of torsional moment and shear stress over the
beam height. Assumed shear stress distributions in the bond area between orthotropic
lamellae can be verified.
190 X. Li et al.

Fig. 3. Illustrations of assumed shear stress distribution for beams with m = 3 longitudinal lamellae in the
beam height direction [7]

4 Conclusions

CLT is an innovative engineered timber product. Its sustainable and superior

characteristics are gaining global interest. The current study aimed at investing the
available testing standards and design rules for CLT panels subjected to transverse
loading. prEN 16351 is found to be the most comprehensive and recommended test
standard. Gamma method, k-method and shear analogy method can be used for
DVVHVVLQJ&/7SDQHO¶VEHQGLQJDQGVKHDUFDSDFLW\XQGHURXW-of-plane loading, whilst
RVSE method and in-plane beaP PHWKRG DUH FDSDEOH RI HYDOXDWLQJ &/7 SDQHO¶V
gross shear, net shear and torsion capacity under in-plane loading.

References
1. PRG 320, Standard for performance-rated cross laminated timber. APA, USA. (2018).
2. Brandner, R., Flatscher, G., et al.: Cross laminated timber (CLT): overview and development.
European Journal of Wood and Wood Products 74.3, 331-351 (2016).
3. Jöbstl, R.A., Moosbrugger, T., et al.: A Contribution to the Design and System Effect of Cross
Laminated Timber. In: CIB W18, 39th Meeting. Florence (2006).
4. Ehrhart, T., and Brandner, R.: Rolling shear: Test configurations and properties of some
European soft-and hardwood species. Engineering Structures 172, 554-572 (2018).
5. Bogensperger, T., Moosbrugger, T., and Silly, G.: Verification of CLT-plates under loads in
plane. In: Proceedings of 11th World Conference on Timber Engineering. Trentino (2010).
6. Flaig, M., and Blaß, H. J.: Shear strength and shear stiffness of CLT-beams loaded in plane.
In: Proceedings of 46th CIB-W18 Meeting, Vancouver. (2013).
7. Danielsson, H., and Serrano, E.: Cross laminated timber at in-plane beam loading - Prediction of
shear stresses in crossing areas. Engineering Structures 171, 921-927 (2018).
8. Brandner, R., Dietsch, P., et al.: Cross laminated timber (CLT) diaphragms under shear: Test
configuration, properties and design. Construction and Building Materials 147, 312-327 (2017).
9. ASD/LRFD, NDS, National Design Specification for Wood Construction. APA, USA (2018).
10. GB 50005. Code for Design of Timber Structures. China (2017).
11. Wallner-Novak, M., Koppelhuber, J., and Pock, K.: Cross-Laminated Timber Structural Design
- Basic design and engineering principles according to Eurocode, proHolz Austria, Vienna
(2013).
12. Gagnon, S., and Pirvu, C.: CLT Handbook: cross laminated timber - Canadian edition,
FPInnovations, Quebec, (2011).
13. Bogensperger, T., Silly, G., and Schickhofer, G.: Comparison of methods of approximate
verification procedures for cross laminated timber, Research report, Graz, Austria, (2012).
 A Further Study on Stay Cable Galloping under Dry
Weather Condition

Vo Duy Hung1, Nguyen Duy Thao2

1,2
The University of DaNang ± University of Science and Technology, 54 Nguyen Luong Bang
Street, Danang city, Vietnam
Email: 1vdhung@dut.udn.vn, 2ndthao@dut.udn.vn,

Abstract. Wind-induced cable vibrations can be classified into several types such
as buffeting due to wind gust, vortex-induced vibration, classical galloping with
iced cables, wake galloping, parametric excitation, Reynolds number related drag
instability, rain±wind induced vibration, high-speed vortex excitation and dry
galloping. Among these vibration types, vortex induced vibration and buffeting
due to wind gust are generally small amplitude while the last three types are
mainly related to stay cables with larger amplitude vibration and the rain±wind
induced vibration is the one most frequently observed on site of bridge.
Furthermore, mechanism of rain±wind vibration has been fully elucidated in
recent years, and some effective control methods have been successfully applied
in practice. In particular, dry galloping is still less understood and it would require
research that is more intensive. Hence, the aim of this paper is to elucidate the
FDEOH YLEUDWLRQ FKDUDFWHULVWLF LQ QR UDLQ FRQGLWLRQ FDOOHG ³'U\ JDOORSLQJ´ by
wind tunnel test. Finally, the detail of its generation mechanism will be
investigated and discussed.

Keywords: stay cable vibration, dry galloping, Scruton number effect,

generation mechanism

1 Introduction

Thanks to advanced construction technology and development of building materials,

cable-stayed bridge length is being broken time by time. Due to the increase of main
span, cable length becomes longer and more vulnerable to wind¶s attack. Common
vibration types are rain-wind-induced vibration, cable galloping in dry weather (dry
galloping-DG), vortex-induced vibration and wake galloping. In general definition, dry
galloping is classified as one of the wind-induced large amplitude vibration phenomena
in no rain condition, usually occurs at high-reduced wind speed. Though there are
reported field observations of large amplitude violent cable motions [1-5], no site
measurement was conducted for further clarification of possible cause. Recently, some
studies showed the existence of dry galloping by wind tunnel test and observation field.
Nevertheless, its characteristics have not been fully understood. Nakamura et al. [6]
indicated that the galloping generation mechanism is interruption of communication
between upper and lower separation flows. Because of communication of two separated

© Springer Nature Singapore Pte Ltd. 2020 191

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_27
192 V. D. Hung and N. D. Thao

flows can tend to make pressure on upper and lower surfaces of cable become no
difference, this lead to change the earodynamic force exciting on cable. In the
experiments of Cheng et al.[7], both divergent type of motion and limited-amplitude
vibration at high-reduced wind speed were also recorded. The former has similar
response as galloping while the latter occurs only in limited range of wind speed with
restricted amplitude. Jakobsen et al. (2012) concluded that inherent flow pattern
unsteadiness in the critical Reynold number range and its interaction with wind
turbulence and inclined angle are significant in forming DG conditions [9]. In the
existing studies, characteristics of cable dry galloping have been elucidated to some
extent. In addition, the flow field around cable for dry galloping still suspected and
unclear. Therefore, to fulfill the literature for dry galloping, further investigation should
be carried out. In this paper, wind-induced response of a circular cable was first
measured at various angles against the wind, the characteristic of cable dry galloping
and its generation mechanism will be elucidated.

2 Wind tunnel test for dry galloping

2.1 Reproduction of dry galloping

Wind tunnel test were carried out under no precipitation to investigate the cable dry
galloping. A cable model was fabricated by a high-density polyethylene tube. Diameter
and length of the cable model are 158mm and 1,500mm, respectively. The cable model
was supported by two coil springs in the vertical 1DOF direction. Cable attitude is
defined by the vertical angleD and horizontal angle E as shown in Figure 1. Test was
conducted with three vertical angles: D = 9, 25 and 40 degrees and horizontal angles:
E = 0, 15, 30, 45, 60 degrees. Other test conditions such as mass, damping, frequency
are shown in Table 1.

Wind E*
flow
D
E

Fig. 1. Set up of cable model

The wind tunnel test results are summarized in Figure 2. Generally, dry galloping took
place in many cases. The arrows in Figure 2 show the diverging characteristic of cable
at a constant wind speed. Divergent vibration occurred only at some wind attack angles
in the subcritical Reynolds number region (6u104 - 1.2u105). This experimental result
is consistent with previous studies [5][8] that dry galloping can occur at wind flow
angle from 15°-60° and high wind speed range above 10m/s. As mentioned above,
A Further Study on Stay Cable Galloping under Dry … 193

current experiment conduct with low Scruton number range from 3.26 to 11.22. In the
other expression, current cable model is sensitive to wind excitation. Therefore, dry
galloping seems to take place easily. From these results, it should pay attention that
cable dry galloping is one of large wind induced vibration, which can cause divergent
galloping.

Table 1. Wind-tunnel test conditions

Parameters Value
Diameter: D (mm) 158
Effective length : L (mm) 1,500
Inclination angle : D (deg) 9, 25, 40
Flow angle : E (°) 0, 15, 30 ,45, 60
Mass: m (kg/m) 9.9
Natural frequency (Hz) 0.78 ± 1.02
Logarithm decrement: G 0.005 - 0.016
Scruton number 3.3 ± 11.5
(2mG/UD2)
Reynolds number ~ 2.1u105
&DEOH¶VVXUIDFH Smooth (normal)

2 DG of smooth surface
D=158mm
Dry condition
1.6 DE

25-0 25-15
25-30 25-45
1.2
25-60 40-0
A/D

40-15 40-30
0.8
40-45 40-60

0.4

0
0 20 40 60 80 100 120 140 160 180

Reduced wind speed (U/fD)

Fig. 2. Dry galloping of smooth surface cylinder

2.2 Sensitivity of dry galloping to Scruton number

In order to investigate sensibility of dry galloping to Scruton number, wind tunnel test
was measured with various cases. The Scruton number was changed by increasing the
mass of model. According to previous study [5], cable DG can occur horizontally in a
range of yaw angle from 30°-60°, therefore current test will carried out with 50° of
yawed angle. The effect of Scruton number was estimated as Figure 3a. Generally,
cable dry galloping is not so sensitive to Scruton number change. Even though when
increases Sc to over 45, divergent galloping still occurred at approximate 15m/s.
However, when Scruton number was risen step by steps, the onset wind speed of dry
194 V. D. Hung and N. D. Thao

galloping a little bit increased. These outcomes totally agreed with Saito et al [3] that it
is quite hard to eliminate dry galloping of cable by adding more external damping or
installing the external dampers. Therefore, the way to increase the stability of stay cable
should be change of the cable cover to harvest higher the aerodynamic damping so that
suppress dry galloping.

2.3 Natural frequency¶V dependence

The parameters of the these tests are wind yawed angle 50 0 and natural frequencies of
the cable model of 1.567Hz, 1.717Hz, 2Hz, 2.233Hz and 2.433 Hz respectively. The
frequency was change by increasing the stiffness of supported springs. Figure 3b
illustrates the vibration amplitudes against wind speed under different natural
frequencies. It is obvious that the increase of the natural frequency of the cable, the
vibration amplitude decreases considerably. Dry galloping occurred at around 12-
15m/s with the low natural frequency cases, say 1.567Hz, 1.717Hz and 2Hz. However,
as natural frequency increased to 2.233Hz and 2.433Hz, dry galloping was vanished up
to 20m/s of wind speed. In the other expression, increase natural frequency of stay cable
can suppress dry galloping.

0.8
Effect of Scruton number
0.8 0.7 Effect of natural frequency
Non-dimmensional amplitude A/D

0.7 D= 86mm 0.6 D = 86mm

Galloping amplitude䠄A/D䠅

Dry condition
0.6 Dry condition
Yaw angle 50ƒ
0.5 Yaw angle 50ƒ
Sructon number
0.5 Natural Frequency
Sc 19.95 0.4 f1=1.567
0.4
Sc 29.24 f2=1.717
0.3 Sc 32.47 0.3
f3=2
Sc 34.98
0.2 0.2 f3=2.333
Sc 46.39
0.1 f4=2.433
0.1
0
0 20 40 60 80 100 120 0
0 20 40 60 80 100 120
Reduced wind speed (U/fD)
Reduced wind speed (U/fD)

a. Effect of Scruton number b. Effect of natural frequencies

Fig. 3. Estimated vibration amplitude with different parameters

2.4 Dry galloping generation mechanism

The fluctuating of wind velocity in the wake of the inclined cable model was recorded
along the model. The circular cylinder was mounted in the wind tunnel with flow angle
E=30° and inclined angle D= 25°. The hot-wire anemometer was set at 2D from the
cable wake and 0.5D from cable axis as shown in Figure 4 and its position along the
cable direction was varied from the upstream side to the downstream side at the
distances from 2D to 7D. By this set up, the fluctuation of wind velocity will be
measured at different wind speeds.

To investigate generation mechanism of dry galloping, normalized power spectrum

density at different wind speeds were analyzed for wind speed fluctuation. Generally,
A Further Study on Stay Cable Galloping under Dry … 195

normalized PSD is obtained by PSD multiplied with frequency and divided by standard
deviation. However, frequency multiplied form is difficult to compare different
frequency region, so normalized PSD will be calculated by PSD divide the standard
deviation of fluctuating wind speed. According to Figure 4, Karman vortex (KV) was
mitigated gradually when wind speed increased. In detail, the normalized PSD was
around 0.17 at wind speed 5m/s and reduced to nearly a half at U=10m/s, and then it
reached around zero at 15m/s. Typically, Karman vortex Strouhal number (fD/U) of
circular is around 0.2 in case cable normal to the wind. However, Strouhal number a
little bit decrease as cable orientation is yawed and inclined to the wind. In this figure,
the Strouhal number was around 0.18, which is in typical range. According to former
literature, it was pointed out that the generation mechanism of galloping is interruption
RI FRPPXQLFDWLRQ EHWZHHQ XSSHU DQG ORZHU VHSDUDWHG ÀRZV %HFDXVH WKH
FRPPXQLFDWLRQRIWZRVHSDUDWHGÀRZVFDQWHQGWRFDQFHOSUHVVXUHGLIIHUHQFHRQXSSHU
and lower surfaces of cylinder. In addition, Karman vortex would be produced by
communication of upper DQGORZHUVHSDUDWHGÀRZVLQDQRWKHUH[SUHVVLRQWKH.DUPDQ
YRUWH[VKHGGLQJVKRXOGSURPRWHWKHFRPPXQLFDWLRQEHWZHHQWZRVHSDUDWHGÀRZVWKH
LQWHUUXSWLRQRIWKLVFRPPXQLFDWLRQEHWZHHQWZRVHSDUDWHGÀRZVVKRXOGEHLGHQWLFDOWR
the interruption of Karman vortex shedding. Therefore, it can be explained that
mitigation of Karman vortex can excite galloping instability [5-8]. In parallel with KV
interruption found here, low frequency component started dominating when incoming
wind increased. Low frequency component contains much of energy can create high
excitation force. There was a relationship between forming of low frequency
component and interruption of Karman vortex, when Karman vortex decreased, low
frequency increased. That means wake flow itself content various latent frequency
(Strouhal number), as Karman vortex frequency suppressed, the other Strouhal
component will be dominant. ,Q VXPPDU\ SRVVLEOH GU\ JDOORSLQJ¶V JHQHUDWLRQ
mechanism can be summarized as follow: when wind speed is high enough, Karman
vortex will be interrupted. At this time, flow pattern will become unstable and its
regularity will be lost. It leads to the flow separation will not synchronize periodically
at some locations. Consequently, low frequency flow/vortices will be appeared with
high energy, which can excite dry galloping.

1
U=5m/s
Wind flow U=10m/s
0.8 U=15m/s
2D

PSD[S /(V)2](s)
5D

0.5D

0.6
3
7D

Hot wire
4

w
5

2D 0.4
6

Cable model
7

D=158mm 0.2

0 -3 -2 -1 0 1
10 10 10 10 10
Reduced frequency(fD/U)

Fig. 4. Normalized PSD (Location 7D, D =158mm, E=30° and D= 25°)
196 V. D. Hung and N. D. Thao

3. Conclusions

In this study, cable dry galloping phenomena was investigated by wind tunnel test. The
characteristics of dry galloping also clarified. Furthermore, the detail of its generation
mechanism was investigated and discussed. Results obtained are as follows:
x Dry galloping has been reproduced successfully in various conditions. Both
restricted response and divergent type galloping were recorded with large amplitude
vibration.
x Dry galloping is less sensitive to the cable damping rather than frequency. Therefore,
it is hard to suppress the dry galloping by installing the external damper.
x The generation mechanism of dry galloping strongly relates to the forming of low
frequency components and interruption of Karman vortex near the wake of cable.

References

1. Ni, Y.Q., X.Y. Wang, and Z.Q. Chen, Ko, J.M., Field observations of rain-wind-induced
cable vibration in cable-stayed Dongting Lake Bridge. J. Wind Eng. Ind. Aerodyn., 2007.
95: p. 303-328.
2. Cheng, S., et al., Experimental study on the wind-induced vibration of a dry inclined cable,
2008. 96: p. 2231-2253.
3. Zuo, D., N.P. Jones, and J.A. Main, Field observation of vortex and rain-wind-induced stay-
cable vibrations in a three-dimensional environment. J. Wind Eng. Ind. Aerodyn, 2008. 96:
p. 1124±1133.
4. Wen-/L&HWDO,QÀXHQFHRI'\QDPLF3URSHUWLHVDQG3RVLWLRQRI5LYXOHWRQ5DLQ± Wind-
Induced Vibration of Stay Cables. Journal of Bridge Engineering, 2013. 18(10): p. 1021-
1031.
5. Matsumoto, M., et al., Dry-galloping characteristics and its mechanism of inclined/yawed
cables. Journal of Wind Engineering and Industrial Aerodynamics, 2010. 98: p. 317-327.
6. Cheng, S., et al. Aerodynamic instability of inclined cables. in The 5th ISCD. 2003. Santa
Margherita.
7. Nakamura, Y., K. Hirata, and T. Urabe, Galloping of rectangular cylinders in the presence
of a splitter plate. Journal of Fluids and Structures, 1991. 5: p. 521-549.
8. J.B. Jakobsen, et al., Wind-induced response and excitation, characteristic of an inclined
cable model in the critical Reynolds number range. Journal of Wind Engineering and
Industrial Aerodynamics, 2012. 110: p. 100-112.
9. Nikitas, N. and J.H.G. Macdonald, Aerodynamic forcing characteristics of dry cable
galloping at critical Reynolds numbers. European Journal of Mechanics B/Fluids, 2015. 49:
p. 243-249.
 Stress-dependent permeability of the fractured rock
masses: numerical simulation based on the Embedded
Fracture Continuum approach
Hong-Lam DANG1,*, Duc-Phi DO2, Dashnor HOXHA2
1
University of Transport and Communications, No.3 Cau Giay Street, Hanoi, Vi-
etnam
2
Univ. Orléans, Univ. Tours, INSA CVL, Lamé, EA 7494, France
*dang.hong.lam@utc.edu.vn

Abstract. In this work, the Embedded Fracture Continuum (EFC) approach will
be used to model the stress-dependent permeability of the fractured rock mass-
es. This novel approach allows incorporating explicitly the fracture networks in
the porous rock mass by using the fracture cell concept which represents the
grid mesh intersected by at least one fracture. More precisely, in the EFC ap-
proach, each fracture cell represents an equivalent porous medium that the con-
cept of continuum model can be straightforwardly applied. The numerical simu-
lations will take into account the closure/opening effect due to the non-linear
behavior of fractures with respect to normal stress as well as the shear-dilation
effect. A damage law of the intact rock that undergoes yield tensile strength is
also considered to model the new flow connections between fractures. Through
some numerical applications, we highlighted the influence of different parame-
ters on the effective permeability of fractured rock masses.

Keywords: stress-dependent permeability, fractured rock, EFC approach, non-

linear behavior, fracture propagation, and Coulomb failure criterion.

1 Introduction

Albeit a huge number of studies on fractured rock masses in the past, their model-
ling is yet a challenging and dynamic research topic, with two principal trends group-
ing continuum and discontinuum approaches [1-2]. Recently, effective properties of
fractured rock masses were studied with the help of embedded fracture continuum
approach [3-5], in which effective properties of fractured rock mass was evaluated.
However, in the previous studies, the propagation of fractures is not considered and
only the linear behavior of fractures was taken into account. In this work, an exten-
sion of the embedded fracture continuum approach will be conducted in which we
account for the non-linear behavior as well as the propagation of fractures. As an
application, this approach will be used to study the stress effect on the effective per-
meability of the fractured rock.

© Springer Nature Singapore Pte Ltd. 2020 197

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_28
198 H.-L. Dang et al.

2 Embedded Fracture Continuum Approach on modeling

fractures

The embedded fracture continuum approach (EFC), recently developed in [3-5],

aims to model explicitly the fractures network on the coupled hydro-mechanical be-
havior of fractured rocks. Follow that, each fracture is modeled explicitly in the dis-
cretized porous medium by a group of grid cells (fracture cells) intersected by this
fracture. In this approach, one distinguishes a fracture cell which represents the grid
cell of the discretized media intersected at least by one fracture with the grid cells that
are not intersected by any fracture (called matrix cell). If the matrix cell represents the
continuum porous matrix owing the physical properties of the intact rock, the fracture
cell properties represent those of an equivalent continuum porous medium counting
for hydraulic and mechanical properties of intact rock and ones of fractures intersect-
ing the cell. Based on the results of the joint model, the simplified poroelastic proper-
WLHV <RXQJ¶V PRGXOXV 3RLVVRQ¶V UDWLR %LRW FRHIIicient and Biot modulus) of the
facture cell is proposed as:
1
­§ 1 cos 2 T .sin 2 T cos 4 T ·
°¨   ¸ , if T  > 0, S / 4@ ‰ >3S / 4, S @ ;
fc °© Em kt .d k n .d ¹
E ® 1
°§ 1 cos 2 T .sin 2 T sin 4 T ·
°¨   ¸ , if T  >S / 4, 3S / 4@ ;
¯ © Em kt .d k n .d ¹
(1)
fc E fc (1  2Q m ) 1 3(1  2Q m )b fc
Q Q m ; b fc 1 ; ;
Em (1  2Q fc ) M fc Em
°h. cos T
­ if T  > 0, S / 4@ ‰ >3S / 4, S @ ;
d ®
°̄ h.sin T if T  >S / 4, 3S / 4@ ;
In Eq. (1), Em, Qm are the <RXQJ¶V PRGXOXVDQG3RLVVRQ¶VUDWLRRI the intact ma-
trix ; kn, kt indicate the normal and shear stiffness of the fracture, T is the inclined
angle of fracture with respect to the horizontal direction while h is the cell size (see
Fig. 1 on the right). By neglecting the permeability of the matrix (which is so small in
comparison with the permeability of the fracture), the corresponding permeability of
the fracture cell is calculated as:
­ w
° K cos 2 (T ) if T  > 0, S / 4@ ‰ >3S / 4, S @ ;
K fc ° f h (2)
®
°K w
sin 2 (T ) if T  >S / 4, 3S / 4@;
°̄ f h
In Eq. (2), the permeability of the fracture Kf is calculated IURPWKHIUDFWXUH¶V hy-
draulic aperture that in 2D case is equal to Kf =w2/12.

3 Non-linear behavior of a single fracture

It has been known since the long time that the fracture behavior is non-linear with
respect to the normal and shear stresses. For example, with respect to the normal
Stress-dependent permeability of the fractured rock masses... 199

compressive stress, the closure of fracture follows the hyperbolic law and the most
commonly formulae used today is the Barton-Bandis function [6-7]:
V n' (3)
G ; V n' V n  b. p
ini V n'
kn 
w
where knini is the normal stiffness at the initial state (without applied stress) of fracture.
The corresponding normal stiffness kn can be calculated by taking the derivative of
the normal effective stress with respect to the normal closure:
2
wV n' V n' (4)
ini ª º
kn kn «1  ini ' »
;
wG ¬ w.kn  V n ¼
Correspondingly, fracture slips when the shear stress reaches the critical value,
usually determined by the Coulomb failure criterion. The shear displacement of frac-
ture induces an increase of facture aperture (shear dilation effect) which is calculated
as function of the dilation angle \f:
­W  W cri
­U s .tan\ f if U s d U sc ; ° K if W t W cri V n' tan I f  C f (5)
G dil ® Us ® t
¯ 0 if U s ! U sc ; ° 0 '
¯ if W  W cri V tan I f  C f
n

In Eq. (5), the two parameters If and Cf present respectively the friction angle and
cohesion of the fracture following the Coulomb failure criterion. The parameter Usc
represents the critical shear displacement at which the shear dilation effect vanishes.
To take into account the non-linear behavior of fracture in the continuum medium,
the algorithm presented in [8] can be used. However, in the present work, the model-
ing is conducted by using the commercial code Flac3D (Itasca) in which the shear
behavior of fracture in the fracture cell can be accounted for by using the available
ubiquitous model. This last model allows to simulate the non-linear behavior (follow-
ing the Coulomb failure criterion) of both the matrix as well as the fracture in which
the fracture is considered as the weaken plane in the matrix.

4 Propagation modeling of fracture

Simulating correctly the propagation of fractures is one of the most challenge tasks
of the fractured medium. Due to the high concentration of stress at tip, the propaga-
tion of fracture initiates from this place. However, for the fracture modeling based on
the continuum approach in which fracture has the stair-step form (Fig. 1), one states
that the propagation can initiate from any point of the fracture if the tip is not well
defined. To distinguish the fracture cell representing the tip of fracture with any other
fracture cells of the same fracture, in this work we distinguish two types of matrix
cell. The first ones (type 1) presents the matrix cell owing an elastic behavior of intact
matrix while the second ones (type 2), owing an elasto-plastic behavior, is used to
model the matrix cell that can be damaged due to the propagation of fracture. The
matrix cell of type 1 is used throughout the medium of the intact matrix (to prevent
the propagation of fracture which can initiate at not the tip of fracture) while the ma-
200 H.-L. Dang et al.

trix cell of type 2 is defined only around the two tips of each fracture at which the
propagation can initiate. This matrix cell of type 2 will be damaged when its stress
state yields the failure strength. Particularly, when one matrix cell of type 2 is dam-
DJHGLWVQHLJKERUV¶PDWUL[FHOORItype 1 becomes matrix cell of type 2 meaning that
they can potentially be damaged after that.

Fig 1: Fracture modeling through the fracture cell (grid cell in blue); Two types of matrix cell:
elasto-plastic matrix cell (type 2) around tip of fracture and elastic matrix cell (type 1); fracture
cell intersected by one inclined fracture.

In this work, the degradation model initially presented in [9] will be used to model
the damage behavior of the matrix cell of type 2. Follow that, an improvement of the
strain-softening constitutive model by considering the deformation modulus degrada-
tion was proposed. More precisely, if the strain-softening model accounts only for the
degradation of shear and tensile strengths, one can also introduce a function to present
the degradation of the elastic modulus as function of the plastic strain after yielding.
A linear function with respect to the equivalent plastic strain was proposed in [9] to
model the gradual decrease of the elastic modulus that changes from the initial value
Em to the residual value Elim. Note that, a similar idea was used in [10] to model the
degradation of the stiffness and strength of the failure element of the intact rock by
tension. By considering that the failure by tension is the principal mechanism in the
intact matrix, the decrease of elastic modulus with respect to the tensile plastic strain
is expressed in this work as follow:
­ Em if Ht  Ht0
°
° Ht  Ht0
E ® Em  ( Em  Elim ) if H t 0 d H t d H tc (6)
° H tc  H t 0
° Elim if H t ! H tc
¯
The two limit values et0, etc of the tensile plastic strain in Eq. (6) can be deter-
mined from a back-analysis process based on the deformation data obtained from the
laboratory or on-site testing. Concerning the residual elastic modulus Elim, the follow-
ing value is proposed:
(7)
1 1 1
 ini
Elim Em kn .h
Stress-dependent permeability of the fractured rock masses... 201

…
5 Stress-dependent permeability of fractured medium

Fig. 2a is presented the model of a fractured rock under uniaxial loading. This
model of 1m in dimension is discretized in 10000 uniform grid cells. For the simplici-
ty purpose, one considers that all the fractures have the same initial aperture of
w=100(ȝm). The mechanical properties of the fractures used in these simulations
consist of: knini=100 (GPa/m), kt=100 (GPa/m), If=29.4°, Cf=0.5 (MPa), Rft=0.1
(MPa), \f=5°. The shear and tensile strengths of the intact rock are characterized by
the following parameters I=40°, C=20 (MPa), Rt=4 (MPa) while the limited parame-
ters are chosen Usc=3(mm), et0=10-6, etc=10-3 respectively.

(a) (b) (c)

Fig 2: Uniaxial loading of the fractured rock (a); contour of the permeability in the fractured
rock masses at the initial state (b) and at 80% of the peak stress (c).

Fig. 3: Ratio of effective permeability with respect to the initial state (a) and anisotropic degree
of the effective permeability as function of the uniaxial stress applied on the fractured rock (b).

After each uniaxial stress state, the effective permeability of the fractured rock is
calculated by imposing a gradient of pressure on the two opposite sides of the medi-
um. Figs. 2b and 2c are captured the contour of the permeability in the fractured me-
dium at the initial state and at the stress state corresponding to 80% of the peak stress.
At this last stress state, it highlights that the propagation of fracture creates new con-
nections in the fracture network. Consequently, as shown in Fig. 3, the effective per-
meability increases in the fractured medium as function of the applied uniaxial stress
particularly for the stress state superior than 80% of the peak stress. A higher permea-
bility in the vertical direction, which is initially smaller than that in the horizontal
direction, is represented by an increase of the anisotropic degree (ratio of Ky/Kx) ver-
202 H.-L. Dang et al.

sus the uniaxial stress. Among different mechanisms considered in this work, the
propagation of fracture presents the most significant effect on the variation of perme-
ability when one observe only the moderate effect of the normal closure or dilation
effect at the very early stress state.

6 Conclusions

The extension of embedded fracture continuum approach was conducted in this

work by taking into account the non-linear behavior and propagation of fractures. By
applying this approach to study the effective permeability of fractured rock masses
under stress, we investigated the influence of different parameters. The results showed
that fracture propagation plays significantly role while only the moderate effect of
normal closuring or dilation on the variation of permeability of fractured rock.

References
1. Marmier, R.: Changement d'échelle dans les modèles hydromécaniques couplés des réser-
voirs fracturés. PhD thesis, Lorraine University, France (2007).
2. Min, K.B., Rutqvist, J., Tsang, C.F., Jing L.: Stress-dependent permeability of fractured
rock masses: a numerical study. Int. J. Rock Mech. Min. Sci., 41, 1191-1210 (2004).
3. Dang, H.L.: A hydro-mechanical modeling of double porosity and double permeability
fractured reservoirs, PhD thesis, University of Orleans, France (2018).
4. Dang, H.L., Do, D.P., Hoxha, D.: Fractured Reservoirs Modeling by Embedded Fracture
Continuum Approach: Field-Scale Applications. Lecture Notes in Civil Engineering, vol
18. Springer, Singapore (2019).
5. Dang, H.L., Do, D.P., Hoxha, D.: Fracture modelling by an embedded fracture continuum
approach: Application in the investigation of permeability and elastic compliance tensor of
fractured rock masses with high contrast of permeability. Advance in Civil engineering
(2019).
6. Bandis, S.C., Lumsden, A.C., Barton, N.: Fundamentals of rock joint deformation. Int. J.
Rock Mech. Min. Sci. Geomech. Abstr., 20:249-268 (1983).
7. Barton, N.R., Bandis, S.C., Bakhtar, K.: Strength, deformation and conductivity coupling
of rock joints, International Journal of Rock Mechanics and Mining Sciences and Geome-
chanics, Abstract. 22 (3), 121-140 (1985).
8. Hurley, R.C., Vorobiev, O.Y., Ezzedine, S.M.: An algorithm for continuum modeling of
rocks with multiple embedded nonlinearly-compliant joints. Comp. Mech., 60, 235-252
(2017)
9. Li, Y., Zhang, D., Fang, Q., Yu, Q., Xia, L.: A physical and numerical investigation of the
failure mechanism of weak rocks surrounding tunnels. Comp. Geotech., 61, 292-307
(2014).
10. Figueiredo, B., Tsang, C.F., Rutqvist, J., Niemi, A.: A study of changes in deep fractured
rock permeability due to coupled hydro-mechanical effects. Int. J. Rock Mech. Min. Sci.,
79, 70-85 (2015).
 Improved Rigid-Plastic Method for Predicting the
Ultimate Strength of Concrete Walls Restrained on Three
Sides

J.H. Doh1( )
, N.M. Ho2 and T. Peters3
1
School of Engineering and Built Environment, Griffith University, QLD 4222, Australia
2
EDGE Consulting Engineers, EDGE Tweed Heads, NSW 2486, Australia
3
EDGE Consulting Engineers, EDGE Gold Coast, QLD 4212, Australia
j.doh@griffith.edu.au

Abstract. The modern-day popularity of tilt-up construction, shear walls and

concrete cores in multi-storey buildings means that the construction of concrete
walls with various boundary conditions and higher slenderness ratios, along with
the presence of openings, has become common. The design of such elements un-
der eccentric axial loads, however, could be outside the restrictions of current
major design codes such as the Eurocode 2 (EC2-2004), the American Concrete
Institute Code (ACI318-2014) and the Australian Concrete Standard (AS3600-
2018). There have been many experimental and numerical studies on the behav-
iour of both one-way action walls and two-way action walls supported on four
sides, with and without openings in the range of high slenderness ratios (up to
50). Efforts have also been made to develop design models capable of predicting
the axial load capacity of such walls. However, research into the behaviour of
two-way action walls supported on three sides (TW3S walls) is still relatively
unexplored and further studies in this area are needed. Recent research has
demonstrated that a rigid-plastic approach could be used to describe the behav-
iour of TW3S walls with and without openings. Although predictions obtained
using the rigid-plastic approach showed reasonably good agreement with exper-
imental test data, the scope of the analysis approach is considered limited. In this
study, a validated finite element method, using the ABAQUS program, was em-
ployed to improve the rigid-plastic model, covering a broader spectrum of de-
signs for axially-loaded TW3S walls. The reliability of the modified model was
confirmed through comparisons with the available test data.

Keywords: Concrete Walls, Axial Load, Rigid-Plastic Method.

1 Introduction

Cast-in-situ or precast reinforced concrete (RC) walls are commonly used in multi-sto-
rey buildings to withstand gravitational and lateral loadings. Walls restrained along the
top and bottom edges by floors, with free vertical edges, subjected to in-plane axial
loads behave in one-way action (OW). Axially-loaded walls can also behave in two-
way action with lateral support on three sides (TW3S) or four sides (TW4S), formed

© Springer Nature Singapore Pte Ltd. 2020 203

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_29
204 J. H. Doh et al.

by floors and intersecting walls, when they are combined to form an isolated box, a
bundled box, a coupled core, or a geometric, U-, T- or L-shape [1]. In many circum-
stances, walls are pierced with openings due to architectural requirements or functional
modifications of the structures. However, these openings are a source of weakness and
can size-dependently reduce the stiffness and load-bearing capacity of the structure [2].
The crack patterns developed at failure of the two-way action walls (TW walls) ax-
LDOO\ORDGHGZLWKDQHFFHQWULFLW\DWWKHµNHUQ¶RIDVHFWLRQtw/6, were similar to the yield
lines of members under transverse loading, as observed in previous research results [3-
10]. This similarity suggested that a rigid-plastic approach could be used to describe
the biaxial effects on such TW walls. Popescu et al. [11] and Ho and Doh [9] proposed
analytical models, based on the rigid-plastic approach, to evaluate the failure loads of
axially-loaded TW4S and TW3S walls, respectively. Although these researchers illus-
trated that predictions using the rigid-plastic approach showed reasonably good agree-
ment with their experimental test data, the scope of the analysis approach is considered
limited. In this study, a validated finite element method (FEM), using the ABAQUS
program, was employed to improve the rigid-plastic model, covering a broader spec-
trum of designs for axially-loaded TW3S walls. The reliability of the modified model
was confirmed through comparisons with the available test data.

2 Improved rigid-plastic method

The rigid-plastic approach for TW3S walls was initially developed by Ho and Doh [9].
Details of the full derivation can be found in their publication [9]. In this paper, the
improved rigid-plastic method proposed by Ho [12] will be briefly presented.
The external and internal work can be obtained using a general work equation de-
veloped by Popescu et al. [11]:

 ³ (nux ,uyG dx, y )each boundary ¦ ( ³ m T ds)

c each yield line
(1)

where nux, nuy are the uniform in-plane compressive force per unit length applied in the
x (horizontal) and y (vertical) directions, respectively; į is the virtual displacement; ș
is the rotation of the region about its axis of rotation; and mc is the membrane moment.
The equation that defines the membrane moment is:

1
mc Q 1 f c' (tw  G p )2
4 (2)

where I¶c is the compressive strength of concrete; tw is the wall thickness; įp is the max-
imum displacement at peak load. Ȟ1 is the effectiveness factor and is defined as Ȟ1 =
ȘfcȘİ; Șfc is the strength reduction factor reflecting the brittleness of concrete, and is
given by Șfc = (fc0/I¶c)1/3 ”ZLWKfc0 = 30 MPa; Șİ is the strength reduction factor reflect-
ing the influence of transverse cracking, and is taken as 0.55.

The yield-line patterns and the maximum displacement at peak load are determined
as follows:
Improved Rigid-Plastic Method for Predicting the Ultimate Strength … 205

2.1 Yield-line patterns

It is assumed that the patterns of yield lines include both real and fictitious yield lines.
These proposed patterns were justified by the crack patterns observed in the experi-
mental tests [4, 6, 7, 9, 10] and by those obtained from the parametric studies using the
FEM [12]. The FEM, using the ABAQUS program, was developed and validated in the
previous study by Ho and Doh [9] for predicting the behaviour of RC walls with various
support conditions under eccentric axial loading. Details of the crack patterns of the
FEM results can be found in the PhD Thesis [12]. Figs. 1 and 2 show the proposed
simplified yield-line patterns for solid TW3S walls and TW3S walls with openings,
respectively. The procedure to determine the yield-line patterns for TW3S walls with
and without openings is provided by Ho [12]. Additionally, the geometric compatibility
suggested by Baker et al. [13] is adopted for the proposed patterns.

(a) Walls with Hw/Lw ” (b) Walls with Hw/Lw > 1.4
Fig. 1. Proposed yield-line patterns for TW3S solid walls.

2.2 Maximum out-of-plane deflection at peak load

The maximum deflection at peak load, įp, is calculated as:

įp ȜHa (3)

in which ea is the lateral deflection for TW3S walls defined in AS3600 [14]; Ȝ is the
fine-tuning parameter, and is defined as:

A ª §e t · H º
O B
u «1  ¨  w ¸ u w u C » for solid walls
 Hw / tw  ¬« © tw 6 ¹ Lw ¼» (4)

A ª §e t · H º D
O B
u «1  ¨  w ¸ u w u C » u E
for walls with openings
 Hw / tw  «¬ © tw 6 ¹ Lw »¼ Kox / Koy 
(5)

where Hw is the floor-to-floor height of the wall; Lw is the wall length; Șox and Șoy are
the distance of opening centre to the edge in the horizontal and vertical directions,
206 J. H. Doh et al.

respectively. A, B, C, D and E factors were calibrated from the FEM results using re-
gression analysis for walls with tw/6 ”e ”tw/3; and A = 43.6, B = 1.04, C = 4.5, D =
0.93 and E = 0.65. Further information concerning the lower limit imposed at e = tw/6
is given by Ho [12].

(a) Pattern 1 (b) Pattern 2

(c) Pattern 3 (d) Pattern 4

(e) Pattern 5 (f) Pattern 6

Fig. 2. Proposed yield-line patterns for TW3S solid walls.

Improved Rigid-Plastic Method for Predicting the Ultimate Strength … 207

3 Comparative study

Table 1 compares the failure loads predicted by the proposed rigid-plastic model with
the available test results [4, 7, 9, 10]. The ratio of the predicted strength to actual
strength varied from 0.6 to 1.34, with an overall mean of 0.95 and a standard deviation
of 0.15. For the solid walls, the average ratio of the predicted strength to actual strength
was 1.00, with a standard deviation of 0.09. Concerning the walls with openings, the
model provided a good conservative mean of 0.91, but a relatively high standard devi-
ation of 0.18. Notably, the model safely and reasonably predicted the failure loads for
wall panels, except the WD2 panel, which had a great discrepancy in the predicted
value. It should also be noted that the predicted value from the proposed rigid-plastic
model for the WD2 panel was 440.8 kN, comparatively close to the predicted FEM
value of 406.1 kN [12]. If the WD2 value was removed, the mean and the standard
deviation would be 0.87 and 0.14, respectively.

Table 1. Comparison of test results and predicted failure loads.

Panel des- Nuw,Test Eq. (1)solid Eq. (1)opening Eq. (1)solid Eq. (1)opening
Ref.
ignation (kN) (kN) (kN) Nuw,Test Nuw,Test
TSNO 502.2 514.6 1.02
N/A# N/A#
TSHO 809.3 705.3 0.87
TSNC 353.2 404.8 1.15
[4]
TSNR 421.1 424.6 1.01
N/A# N/A#
TSHC 715.2 554.9 0.78
TSHL 668.1 402.1 0.60
[7] TW3S-NF 450.0 N/A# 380.4 N/A# 0.85
WS1 620.0 603.4 0.97
WS1a 499.1 496.2 0.99
WS1b 450.6 482.9 1.07
WS1c 398.4 377.7 0.95
WS2 500.1 569.4 N/A# 1.14 N/A#
WS2a 410.3 402.3 0.98
WS2b 355.9 401.7 1.13
WS3a 471.7 413.8 0.88
[9,10] WS4 475.7 476.3 1.00
WL1 529.5 409.5 0.77
WC1 512.5 455.8 0.89
WR1 480.5 477.7 0.99
WL2 358.5 311.6 0.87
N/A# N/A#
WC2a 449.0 395.2 0.88
WC2b 414.9 371.5 0.90
WR2 486.7 389.7 0.80
WD2 329.8 440.8 1.34
Mean 1.00 0.91
Standard deviation 0.09 0.18
#
Not applicable
208 J. H. Doh et al.

4 Conclusions

In this paper, the rigid-plastic model for evaluating the ultimate load of TW3S walls
was improved, which covers a vast range of conditions encountered in practice. The
modified model was established based on several assumptions from the experimental
and numerical observations, and was calibrated with the FEM results. This model has
shown good predictions and correlations with the experimental results. In future, when
further research is undertaken, the proposed model could be further validated and en-
hanced, improving its reliability and practicality.

References
1. Menegon, S.J., Wilson, J.L., Lam, N.T.K., Gad, E.F.: RC Walls in Australia: Reconnais-
sance Survey of Industry and Literature Review of Experimental Testing. Australian Journal
of Structural Engineering 18(1), 24±40 (2017).
2. Seddon, A.E.: The Strength of Concrete Walls under Axial and Eccentric Loads. In: Sym-
posium on the Strength of Concrete Structures, London, UK (1956).
3. Doh, J.H., Fragomeni, S.: Evaluation of Experimental Work on Concrete Walls in One and
Two-way Action. Australian Journal of Structural Engineering 6(1), 37-52 (2005).
4. Doh, J.H., Lee, D.J., Guan, H., Loo, Y.C.: Concrete Wall with Various Support Conditions.
In: 4th International Conference on Advances in Structural Engineering and Mechanics
$6(0¶.RUHD
5. Lee, D.J.: Experimental and Theoretical Studies of Normal and High Strength Concrete Wall
Panels with Openings. PhD Thesis, Griffith University, Australia (2008).
6. Doh, J.H., Loo, Y.C., Fragomeni, S.: Concrete Walls with and without Openings Supported
on Three Sides. In: Incorporating Sustainable Practice in Mechanics and Structures of Ma-
terials, 209-214 (2010).
7. Lima, M.M., Doh, J.H., Miller, D.: Experimental Study of RC Walls with Openings
Strengthened by CFRP. In: 23rd Australasian Conference on the Mechanics of Structures
and Materials (ACMSM23), Byron Bay, Australia (2014).
8. 3RSHVFX&6DV*6DEăX&%ODQNVYनUG7: Effect of Cut-out Openings on the Axial
Strength of Concrete Walls. Journal of Structural Engineering 142(11), (2016).
9. Ho, N.M., Doh, J.H.: Experimental and Numerical Investigations of Axially Loaded RC
Walls Restrained on Three Sides. The Structural Design of Tall and Special Buildings 27(7),
1-23 (2018). doi: 10.1002/tal.1459
10. Ho, N.M., Doh, J.H.: Prediction of Ultimate Strength of Concrete Walls Restrained on Three
Sides. Structural Concrete, (2019). doi: 10.1002/suco.201800188
11. Popescu, C., Schmidt, J.W., Goltermann, P., Sas, G.: Assessment of RC Walls with Cut-out
Openings Strengthened by FRP Composites Using a Rigid-plastic Approach. Engineering
Structures 150, 585-598 (2017).
12. Ho, N.M.: Experimental and Theoretical Studies on Three-side Restrained Reinforced Con-
crete Walls. PhD Thesis, Griffith University, Australia (2018).
13. Baker, C., Chen, B., Drysdale, R.: Failure Line Method Applied to Walls with Openings. In:
10th Canadian Masonry Symposium, Alberta, Canada (2005).
14. AS3600: Concrete Structures. Sydney, NSW, Australia: Standards Australia International
(2018).
 Flexural Capacity Accounting for SBHS500 Steel of
Composite Bridge Girders

Dang Viet Duc 1 , Yoshiaki Okui2

1
ThuyLoi University, 175 Tay Son, Dong Da, Hanoi, Vietnam,
2
Saitama University, 255 Shimookubo, Sakura Ward, Saitama, 338-8570, Japan
dangvietduc@tlu.edu.vn

Abstract. The positive bending moment capacity of composite steel girders is

examined through study employing elasto-plastic FE analyses. The web slen-
derness limits of section classification for homogeneous and hybrid girders with
bridge high performance steel are explored. Besides, the effects of initial bend-
ing moment due to unshored construction method on the web slenderness limit
are investigated. It is shown that the noncompact web slenderness limits in con-
ventional design standards are conservative for composite sections

Keywords: SBHS steel, hybrid factor, composite girders, hybrid girders

1 Introduction

The Steels for Bridge High Performance Structures (SBHS) have been standardized
by [1]. SBHS steels present the advantages such as good weldability and high yield
strength, but it has different inelastic behavior, such as almost no yield plateau, small-
er ductility, and greater yield ratio comparing to conventional grade steels. The hybrid
girders are considered to be an economical solution for composite girder bridges. [2]
employs the hybrid factor Rh to calculate the yield moment Myf of a hybrid section
from My of the relevant homogeneous section. The effect of initial bending moment
(M1) due to the un-shored construction method should be considered; hence a new
hybrid factor formula will be developed. To investigate the flexural capacities of
composite girder with applying SBHS500 steel, the section classification should be
studied. By comparing the ultimate bending moment Mu to quantities of Mp and My,
which explained in Fig.1a and b, the section of composite girders is classified as
compact, non-compact and slender. The current codes as [2] and [3], by evaluating
the web slenderness, specify the section classification as summarized in Table 1.
[4] presented the effects of M1 on the web slenderness limit (WSL) for section
classification of composite girders and reported the significant effect of M1 on
noncompact-slender WSL. However, the study only considered composite girders
with homogeneous steel section and conventional steel grade SM490Y. Hence, the
first is to derive the improved hybrid factor with considering the effect of M1. The
second is the assessment of section classification of composite girders applying
SBHS500 steel to the case of steel girder homogeneous and hybrid sections.

© Springer Nature Singapore Pte Ltd. 2020 209

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_30
210 D. V. Duc and Y. Okui

Table 1. Web slenderness limit given by Eurocode (CEN, 1996) and AASHTO (2005)

Web slenderness limit

Section class Flexural resistance
AASHTO Eurocode
­ 41.5H
2 Dcp E ° D d 0.5
Mu t M p 䍺3.76 s bw tw d ® D
Compact tw fy 456H
° D . ! 0.5
¯13D  1
­ 42H
2 Dc E ° \ t 1.0
Noncompact M p t Mu t M y  5 .7 s bw tw d ® 0.67  0.33\
tw fy °62H 1  \   \
¯ \ d 1.0
Slender M y t Mu Otherwise Otherwise

where Mu is ultimate flexural strength; Mp and My are the plastic moment and the
yield moment, respectively. The symbols used in Table 1 are defined in Fig.1.

0.85f ¶cd ɍG
- D cp = Șbw
 tw Ș' b w = D c
bw tw bw
+
ɒɍG
 fy

Fig. 1a Stress distribution at full plastic state Fig. 1b Stress distribution in noncompact and
in compact section slender sections of homogeneous girders

2 Proposal of hybrid factor

2.1 Derivation of hybrid factor

In AASHTO, the hybrid factor Rh is expressed as

M yf 12  E (3D  D 3 ) f yw 2 Dn t w
Rh where D E (1)
My (12  2E ) f yf A fn

My and Myf stand for yield bending moments of homogeneous and hybrid girders,
respectively; fyf and fyw stand for the flange and web yield strengths respectively; Dn
and Afn are the distances from the elastic neutral axis to an inside face of a flange and
the area of the flange where yielding occurs first. In unshored construction method,
first M1 due to the dead load at construction stages is applied to steel sections alone,
M2 due to the live load is done to composite sections, as shown in Fig.2. The Rh is
developed for positive bending moment only, and for sections where bottom web
yields first. The following assumptions are made in the derivation of the hybrid fac-
Flexural Capacity Accounting for SBHS500 Steel … 211

tor: Linear distribution of strain in a section, perfect elastic-plastic behavior of steel

materials, flange thickness is negligible compared with a beam depth.

Vc
Vc
Vt1 +Vt2
Vt1 Vt2
ttf
My
M2
M1 yc
M¶ y
ys
tbf b
Vb1 Vb2
Vyw
VyfV
Vyw

Vb1 +Vb2 = Vyf

Fig. 2 Assumption on stress distributions for the estimation of hybrid yield moment; (a) Initial
bending moment on steel section alone, (b) Bending moment on composite section, (c) Super-
posed stress distribution

Myf = My ± 0¶ 01+M2 ± 0¶ (2)

where My = M1+M2 is the yield moment of the homogeneous steel section

2
­ ½
° °
° h °
(2  D )(1  D ) 2 ® ¾ tw
°1  ª)  §¨1  )y s ·§ h ·º °
¸¸¨¨  1¸¸»
° « ¨ h y °
yielded § 1
¨ 1 ·¸ ¯ ¬« © s ¹© yc ¹¼» ¿ (3)
Rh )S1.tf  1
¨ S 2.bf S1.bf ¸ 3S
© ¹ 2.bf

2.2 Examination with FEM analysis results

Non-linear elasto-plastic FEM model. A half of the steel-concrete composite I-
girders are meshed based on symmetry rule. The magnitudes of the initial bending
moment are assigned to M1 = 0.2; 0.4; 0.6 Mys, where Mys is the yield bending mo-
ment of steel girder section only. The displacement control method is used in the in-
cremental nonlinear analysis. The elasto-plastic behavior is modeled by the von Mises
yield criterion. The uniaxial multi-linear stress-strain curves of steel grade SM490Y
for the web plate and SBHS500 for the flange plates are shown in Fig.3. Concrete in
compression is modeled as linear-elastic, plastic hardening material and Mohr-
Coulomb yield criterion. The uniaxial stress-strain curve for concrete under compres-
sion is shown in Fig.4 (Standard [5]).
The studied girders are assigned with length = 9m, web depth = 3m and slab thick-
ness = 300cm. The pure bending moment is induced incrementally by displacement
method as presented in Fig.5. The initial geometrical imperfection in the web plate is
considered as a sinusoidal form with the maximum level of bw/250 according to [6].
The web and flange plates of the steel girder are meshed by four-node quadrilateral
212 D. V. Duc and Y. Okui

shell element (Q20SH), and the concrete slab part is meshed solid element (HX24L),
from DIANA software library. The effect of M1 is considered by application of phase
analysis, as showed in Fig.6.

SBHS500 40
600

Compressive stress V c (N/mm2 )

30
V(N/mm )
2

400 SM490Y
20

200
10

0
0
0 0.001 0.002 0.003 Hcu 0.004
0 0.05 0.1 0.15 0.2
strain H compressive strain H c

Fig. 3 Uniaxial stress-strain relation applied Fig. 4. Compression uniaxial stress-strain

for steel material models curves applied for concrete material model

Z axis
compression

M
Y axis

X axis

Pure bending
4.5 m moment

Fig. 5 Boundary condition of the model Fig. 6. FE models for phase analysis

M 1=0 M 1=0.4Mys
1 1.01
1
0.995
0.99

0.99 0.98
0.97
Rh
Rh

0.985
0.96

0.98 0.95
Proposal 0.94 Proposal
0.975 FEM results FEM results
0.93
AASHTO AASHTO
0.97 0.92

0.4 0.6 0.8 1 1.2 1.4 0.4 0.6 0.8 1 1.2

E E

Fig. 7 Comparison of hybrid factors for Fig. 8 Comparison of hybrid factors for the case
M1=0 with initial bending moment M1 = 0.4 My

Results and discussion. The case in which M1=0 is shown in Fig.9, almost all pro-
posal hybrid factors are lower than those obtained from FEM analysis results. The
Flexural Capacity Accounting for SBHS500 Steel … 213

lower bound of proposal and FEM analysis results are similar. The proposal hybrid
factors are slightly lower than those of AASHTO specifications; the differences are
about 0.5 %. Fig. 8 represents results with considering initial bending moment. The
results of hybrid factor obtained from FEM analysis results are lower than those of [2]
specifications, which shows the un-conservativeness of AASHTO specifications for
the case of applying high level of M1.

3 Web slenderness limits in design of composite girders

3.1 SBHS500 homogeneous section

250 250
Width-to-thickness ratio bw /tw

200 Width-to-thickness ratio bw /tw 200

m Gupta et al., 2006 Gupta et al., 2006o
150 150
m AASHTO Compact AASHTO Noncompacto
EC Class 2 o
100 100
EC Class 1 o EC Class 3 o
50 50
Compact Non-compact
Non-compact Slender
0 0
0 0.1 0.2 0.3 0.4 0.5 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7
D or D c D or D c

Fig. 9. Compact-noncompact limit of homoge- Fig. 10. Noncompact-slender limit of homoge-

neous SBHS500 steel sections (M1=0) neous SBHS500 steel sections (M1=0.4Mys)

The following figures will present the FE analysis results of SBHS500 homogeneous
girders along with the WSL of [2], [3] and [4]. In Fig.9, the circle symbols stand for
the numerical results judged as compact sections, while the square ones for
noncompact sections. The SBHS500 homogeneous sections for M1=0 present signifi-
cant greater WSL than those of [2], [3] and proposed in [4]. Owing to the smaller
yield plateau of SBHS500, it can sustain a greater plastic local bucking strength than
conventional steel. Fig. 10 show the results for the WLS of noncompact-slender
boundary for of M1 = 0.4 Mys,. With increasing M1, the WSL of noncompact-slender
boundary increases, which presents the same trend as reported by [4]
3.2 SBHS500-SM490Y hybrid sections
Fig.11 shows comparison between FE results for SBHS500-SM490Y hybrid sections
and the compact-noncompact WSL of [2] and [3]. The FE results present the larger
compact-noncompact WSL than those of [2] and [3]. Fig.12 shows the noncompact-
slender WLS with M1 = 0.4Mys. The WSL for hybrid sections is lower than that of
homogeneous sections as well. The difference is about 10%.
214 D. V. Duc and Y. Okui

250 250

Width-to-thickness ratio bw /tw

Width-to-thickness ratio bw /tw

200 200
m Gupta et al., 2006 Gupta et al., 2006o
150 150
m AASHTO Compact AASHTO Noncompacto
EC Class 2 o
100 100
EC Class 1 o EC Class 3 o
50 50
Compact Non-compact
Non-compact Slender
0 0
0 0.1 0.2 0.3 0.4 0.5 0.6 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7
D or D c D or D c

Fig. 11. Compact-noncompact limit of hybrid Fig. 12. Noncompact-slender limit of compo-
SBHS500-SM490Y steel sections (M1=0) site SBHS500-SM490Y hybrid steel sections
(M1=0.4Mys)

4 Conclusions and Remarks

Proposed hybrid factor: Considering effect of M1 specification of hybrid factors [2]

shows the un-conservativeness.
Application of SBHS500 steel to composite girders: Applying SBHS500 steel to
both homogeneous and hybrid sections can extend significantly the WSL for section
classification, especially for hybrid sections. For composite girders with hybrid
SBHS500-SM490Y steel section and applying the M1=0.2; 0.4; 0.6 Mys, there is no
slender section in practical range of D or D¶

References
1. Japanese Industrial Standard. JIS G 3140, Higher yield strength steel plates for bridges,
2008
2. American Association of State Highway and Transportation Officials. AASHTO LRFD
bridge design specification ± Fourth edition, 2007
3. CEN Eurocode 4. Design of Composite Steel and Concrete Structures, Part 2, Bridge 1,
Draft, 1996.
4. *XSWD9.2NXL<DQG1DJDL0³'HYHORSPHQWRIZHE6OHQGerness Limits for Com-
posite I-*LUGHUVDFFRXQWLQJIRU,QLWLDO%HQGLQJ0RPHQW´'RERNX*DNNDL5RQEXQVKXX$
Vol.62, No.4, 854-864, 2006.
5. Committee on Steel Structures of Japanese Society of Civil Engineers Standard. Specifica-
tions for Steel and Composite Structures, (General Provision, Basic Planning, Design),
2007.
6. Japan Road Association: Specifications for Highway Bridges, Part II, Steel Bridges, Maru-
zen, 2002 .
 The inadequacies of the existing Structural Health
Monitoring systems for cable stayed bridges in Vietnam

Luong Minh Chinh 1[0000-0001-8552-170X]

1 Transportation Department, Civil Engineering Faculty, Thuy Loi University, Hanoi, Vietnam.
chinhlm@tlu.edu.vn

Abstract. Since 2000 when the My Thuan Bridge, the first cable-stayed bridge
in Vietnam, was put into operation, and now Vietnam has more than 20 types of
cable-stayed bridges. Therefore, the Structural health monitoring (SHM) system
is gradually being designed and installed for cable stayed bridges to ensure eco-
nomic exploitation and safety. Due to the limited of financing sources, these sys-
tems are very limited, and their quality have a lot to be desired. Also, due to the
lack of appropriate classification personnel with experience in the SHM system,
these systems encountered a lot of problems. In this article author will deeply
analyze the mistakes and problems of these SHM systems to find solutions for
the future.

Keywords: SHM system, concrete cable-stayed bridge, AE sensors, cracks.

1 INTRODUCTION

Structural Health Monitoring (SHM) system for bridges has recently been considered
for installation on a few cable-stayed bridges in Vietnam due to their sensitivity to the
structural load and live load [1] [2]. The reality is that the SHM systems installed in
Vietnam are quite diverse and currently there are no common regulations, requirements
for them from the regulatory authorities. SHM systems are implemented by many dif-
ferent vendors with devices from various suppliers with different number of sensors,
quality and cost. The purposes of system design are not consistent. During the exploi-
tation phase of each bridge, the operating unit develops a maintenance manual or
maintenance manual, regulations on data, however, the reporting regime is still unclear,
lacks of unity and the maintenance requirements are not detailed that leads to difficul-
ties in usage and operation of these SHM systems [3].

2 EXISTING MONITORING SYSTEMS IN VIETNAM

2.1 Bai Chay Bridge SHM system

During the construction of the Bai Chay Bridge, a monitoring system has been installed
for the construction phase (by the Contractor Shimizu). However, the system is no

© Springer Nature Singapore Pte Ltd. 2020 215

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_31
216 L. M. Chinh et al.

longer active (due to a fire inside the bridge box girder). During the construction phase,
the data has been sent directly to Research Centre of Shimizu in Japan for analyzing
and making necessary adjustments for the construction work. New monitoring system
for the operation phase has been installed by ADVITAM, completed and activated since
2014, but due to limited budget, the system is only installed on one side of the P3 pier
(see Fig.1). Therefore, the assessment of the overall condition of the structure faces a
lot of difficulties. The data collection maintenance is implemented by ADVITAM. All
system modifications must be made by ADVITAM - this is a restriction that leads to a
lack of flexibility of the system.

Fig. 1. Weather station install in main desk and on the top of pylon anchor of Bai Chay Bridge

The bridge management unit concurrently operates the monitoring system, however, it
is incapable of analyzing and evaluation of data for assessment and maintenance of the
bridge. Also, there is no analysis of the data reported.

2.2 Binh Bridge SHM system

In 2010, an incident occurred to Binh Bridge in Hai Phong. The CONSON hurricane
broke off the anchors of three large ships anchored at Bach Dang Shipyard, about 500m
from the bridge and they all went downstream, crashing into Binh bridge girder causing
damage at main girder (see Fig. 2) and two cables, leading to stopping the traffic for
vehicles over 3.5 tons until the repair works completed.

Fig. 2. Main girder deformation after the ship collision with bridge during CONSON hurricane.

In 2012, repair and rehabilitation work was carried out and a monitoring system was
established for the purposes: regular monitoring at the operation phase, support mainte-
nance work for early warning of any abnormal states, as well as to make accurate rec-
ommendations on upgrade, maintenance. Due to limited budget, the minimum option
for system has been selected. The system will be combined with regular inspection and
monitoring on an annual basis to evaluate the current situation of Binh Bridge after
The inadequacies of the existing Structural Health … 217

rehabilitation. SHM system of Binh Bridge has some advantages and disadvantages as
follow:

x SHM system of Binh Bridge is very simple with few sensors to provide some basic
measurements, which did not help to the assessment current status of the structures.
x The supplier has an office in Vietnam, which allows a quick technical support; mod-
ern equipment and software have a Vietnamese interface with full basic information.
x System management unit is incapable for in-depth analysis of data in order to make
assessment and maintenance of the bridge and have no data for analysis reports.

Due to the complexity of Binh bridge structure, that previously has had an incident and
the bridge was once repaired, in addition to the regular inspection and monitoring sys-
tem, other periodical monitoring is required to fully assess the current status of the
structures in order to avoid future incidents.

2.3 Rao Bridge II SHM system

A monitoring system has been installed for Rao Bridge II in Haiphong in 2012, at the
end of construction works to monitor bridge in the operation phase (see Fig.3).

Fig. 3. Temperature sensor install on main girder of Rao II Bridge during inspection.

This is a relatively complete system and if integrated with proper analytical software
can help assess the structural status. The cost of the system is lower than some similar
systems in Vietnam. But at the time of the survey the system was not active, the main
sensors did not transmit the signal to the center. The supplier is abroad so should be
late in technical support, maintenance; the warranty period is expired and no solution
have been taken. System management unit is incapable for in-depth analysis of data in
order to make assessment for bridge maintenance and inspection to troubleshoot. It has
no data for analysis report. Then the SHM system needs to be repaired and put back
into operation as soon as possible.

2.4 Can Tho Bridge SHM system

Can Tho Bridge installed a SHM system – BRIMOS [1], in operation phase from the
end of 2013 to provide data for analysis and to evaluate the structural condition through
the behavior of the bridge structure. Use of monitoring data to manage (control) traffic
safety and flow in abnormal conditions as well as provide design check data.
218 L. M. Chinh et al.

During the survey, the monitoring equipment was transmitting raw data to Can Tho
bridge maintenance department. The SHM system send daily basic reports, but doesn’t
have detailed analysis of the data due to the incapable of management unit to analyze
and evaluate the data. It happened that due to the high frequency of data collection, that
leads to memory overflows and the system automatically erase old data, leading to the
loss of essential information. The SHM system of Can Tho Bridge has some disad-
vantages as follows:

x The initial data of status "0" is missing.

x The supplier is abroad so there will be difficulties in technical support, maintenance.
x No incapable for in-depth analysis of data to make assessment for maintenance.

To overcome the disadvantages mentioned above, there should be consider the ability
to upgrade the software especially for the processing of GPS data to exclude unneces-
sary duplication of raw data in order to effectively evaluate structure status.

2.5 Nhat Tan Bridge SHM system

Nhat Tan Bridge monitoring system has been installed in construction phase in 2015,
which has been handed over and put into operation (see Fig. 4) [4].

Fig. 4. Installation of SHM system of Nhat Tan Bridge (left) and seismic sensor (right)

Nhat Tan Bridge SHM system has been installed in the construction phase, with a
large number of sensors, some highly effective measuring cumulative values of con-
struction, such as cable tension and deformation sensors. The system has the most num-
ber of sensors. But adjusting and defining the initial value along with warning thresh-
olds is difficult. At the time of survey, the monitoring equipment was working nor-
mally. System management unit is incapable for in-depth analysis of data in order to
make assessment for bridge maintenance and inspection to troubleshoot. There is no
data for analysis report.

3 OVERALL ASSESSMENT ON CURRENT SHM SYSTEMS

SHM system designed and installed in Vietnam are quite diversified. They were pro-
vided by various consultants and came with different quality and cost from many ven-
dors. The determination of the purpose and system building method, analysis and eval-
uation and use of data are not regulated and specifically guided from state management
agencies, so, the comparison and evaluation are relatively complicated. Almost all
The inadequacies of the existing Structural Health … 219

SHM system have ambiguous purpose and the level of monitoring to be achieved has
not been determined. The SHM system often includes a wide range of categories and
various measurement sensors, at very high cost, often at the maximum price. The sur-
veyed systems are only at very basic level, that means, only some basic raw information
are provided with some additional warning messages, but warning thresholds are not
accurately and clearly defined [3][5].
Ability of provision of information of SHM system installed is also very diverse.
The accuracy of the equipment is also a matter for periodic inspection and calibration.
Analysis ability of the current software limits the processing capabilities and further
analysis. Management staff do not have an in-depth knowledge of data analysis and
assessment, so the daily reports are mostly just the current status of the operation of the
equipment (Table 1).

Table 1. Evaluation of installed SHM system by monitoring level

Bridge SHM System Level Ability to improve to level 3, 4
Need to complete software and calculation
model with an update of monitoring data. The
Bai Chay In operation stage Level 1
monitoring system installed only on the P3
tower is a big downside of this system
Need to complete software and calculation
Can Tho In operation stage Level 1
model with an update of monitoring data
Binh In operation stage Level 1 Need to supplement in second stage
Rao II In construction stage Level 1 The system has stopped, need to be rectified
In construction and Need to complete software and calculation
Nhat Tan Level 1
operation stage model with an update of monitoring data
The bridge cameras are one of the most effective items in monitoring of traffic flow,
accident handling, bridge protection, with sufficient number that can scan throughout
the bridge. It is necessary to supplement monitoring radar and traffic count equipment
to measurement number, type of vehicles and speed, etc., and combine with WIM that
can weigh and determine the load, to restrict overload vehicles.
Information about temperature, humidity, wind direction and wind speed are fully
provided. However, the association of these parameters with structural behavior analy-
sis due to the influence of ambient temperature has not been fully conducted. SHM
system installed in Vietnam have set up operational warning systems. However, the
determination of values beyond the threshold (warning value) is difficult and in fact,
these thresholds are not working effectively.
Ability to determination of damaged location, abnormal status of the bridge - These
are the highest range of level 4 (state control) and 5 (defect detection) that SHM system
can achieve. Other applications such as Acoustic Emission (AE) should be applied to
detect damages and damaged locations, especially for rein-forced concrete bridges [5].
Ability to forecast, assessment of structure working life - This is the highest thresh-
old that the monitoring system can achieve. The system can forecast in order to support
to determine the remaining life of the structure, forecast damage and abnormal status,
support to make timely maintenance or preservation.
The installed SHM system have initially supported maintenance work, even reduced
many maintenance costs, and removed periodical inspection work. However, it should
220 L. M. Chinh et al.

be noted that the SHM system cannot replace the maintenance work, and should be
combined with other inspection work. [6].
Durability and working life of the system - Sensors have a relatively good working
life span (over 5 years), but problems often occur with data transfer. This is noteworthy
as it will affect the continuity and accuracy of the data. Currently, according to the
world's statistics, the current SHM sys-tem cost is about 0.3% - l.5% of total investment
cost and cost of SHM system in Vietnam are very high [3].

4 CONCLUSION
The study has proposed criteria and specification for technical and economic assesment
on the basis of Vietnam conditions. SHM system have been installed on 05 typical cable
stayed bridges. SHM system of each bridge has been analyzed to show the
disadvantages and propose specific rectification solutions to help the management units
to learn from experience and have timely rectification. However, SHM system in
Vietnam still has many problems and is not managed consistently. The installation of
the SHM system has been individually and spontaneously carried out in a number of
projects, largely dependent on the finance source and subjective opinions of consultants
and contractors. The owner and management unit do not have much experience in this
sector. The management, analysis, processing and storage of data and parameters
collected from monitoring systems in the exploitation process in many works is
perplexed, not effective and the number of experts is limited.

References
1. L. M. Chinh (2014). BRIMOS – SHM system for cable-stayed Can Tho Bridge, Inżynieria
I Budownictwo – ISSN 0021-0315. Poland. No 7/2014:.397-400.
2. L. M. Chinh. G. Swit. (2016). A Prototype System for Acoustic Emission-Based Structural
Health Monitoring of Mỹ Thuận Bridge. Prognostics & System Health Management Con-
ference—Chengdu (PHM-2016 Chengdu, China).
3. L. M. Chinh (2014). Long term structural health monitoring system for cable stayed bridge
in Vietnam. Tạp chí Khoa học kỹ thuật thủy lợi và Môi trường - ISSN: 1859-3941. No. 44
(3/2014)
4. K. Matsuno & N. Taki (2014) - Construction of the Nhat Tan (Japan Vietnam Friendship
Bridge) Superstructure. Bridge and Foundation Engineering Vol.48: 02–12.
5. L. M. Chinh, G. Swit, (2015). Application of the acoustic emission method of identification
and location of destructive processes to the monitoring of the technical state of pre-stressed
concrete bridges. Hội nghị khoa học Công nghệ Giao thông vận tải lần thứ III, năm 2015.
6. L. M. Chinh, G. Swit, A. Adamczak, A. Krampikowska. (2016). Nhat Tan Bridge –The
biggest cable-stayed bridge in Vietnam. Symposium WMCAUS 2016. Czech. Procedia En-
gineering 161 (2016): 666 – 673.
 Riding comfort assessment of high-speed trains based on
vibration analysis

Van Nguyen Dinh1,a, Ki-Du Kim2,b and Dinh Tuan Hai3,c

1
MaREI Centre, University College Cork, Beaufort Building, Ringaskiddy, Co. Cork, Ireland
2
Konkuk University, 120 Neungdong-ro, Gwangjin-gu, Seoul 05029, Korea
3
Department of Urban Management, Hanoi Architectural University, Ha Noi, Vietnam
a b
nguyen.dinh@ucc.ie (correspondence) kimkd@konkuk.ac.kr
c
dinhtuanhai@yahoo.com

Abstract. This paper proposes a method to assess riding comfort of high-speed

multi-car trains and railway systems by using three-dimensional train-bridge in-
teractive models. Each train car is modeled as a multi-axle double-layer of
mass-spring-damper system having 27 degrees of freedom and the bridge mod-
eling is adaptive to various finite elements. The case study considers a real train
of fifteen cars travelling at a speed range of 50-400 km/h over a railway section
having a bridge and rail irregularities of U.S. Federal Railroad Administration
class 6. The riding comfort of each train car is found to reduce considerably
when it is crossing the bridge. The car body accelerations and bridge deflection
are compared with the allowable values specified in the design codes for high-
speed railways in Korea, Japan and Taiwan. The assessment shows that riding
comfort of the considered railway section is poor and that the design of the
railway section or the car body suspensions need to be revisited.

Keywords: Riding Comfort, High-speed Train, Car-body Acceleration, Train-

Bridge Dynamic Interaction.

1 Introduction

High-speed train systems have been operated in several countries including Japan,
France, Germany, Belgium, The Netherlands, Spain, Italy, South Korea, Taiwan and
China. The pre-feasibility study of the North-South High-Speed Railway project in
Vietnam has been conducted where the preliminary estimation of the total investment
is US$58.26 billion [1]. Transportation service quality is one of the major factors to
attract passengers required to profit from such huge capital investment. Besides trav-
eling speed and carrying volume, riding comfort is important to the service quality.
Riding comfort of a vehicle indicates the how much comfortable the passenger
feels during vehicle riding. Various qualitative and quantitative factors have been
quantified and found to be related to the ride comfort of the Korean high speed trains
(KTX) [2]. It is most widely indicated by the level of accelerations produced by vehi-
cles or felt by passengers. Since a passenger body has its own mass and inertia mo-
ment, under action of vehicle accelerations, the forces are consequently induced.

© Springer Nature Singapore Pte Ltd. 2020 221

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_32
222 V. N. Dinh et al.

While the general tendency in the railway transport is to increase velocities, maintain-
ing the level of riding comfort becomes increasingly important [3]. Particularly, when
a train is moving across a bridge or an elevated railway section, the bridge vibration
and rail irregularities, in turn, affect the train vibration and the riding comfort.
The investigations of Taiwan high-speed railway (THSR) riding comfort were car-
ried out by using car body vertical accelerations and the Sperling’s ride index [4].
Vehicle vertical accelerations were calculated from a bridge-single railway vehicle
interactive system [5]. Vertical vibrations of the railway car-body were examined by
using a two-dimensional (2-D) model and passive suspension damping of was select-
ed for riding comfort enhancement at high velocities [3]. A 2-D model of ten degrees-
of-freedom (DOFs) and several vertical profiles of rail irregularities were used to
investigate the vehicle’s vertical vibration and riding comfort [6].
In practice, the riding comfort is indicated by not only vertical acceleration of car
bodies but also by their lateral, rolling and yawing accelerations. In addition, the rid-
ing comfort of a car is interactively affected by other cars in the train. Three-
dimensional (3-D) finite element models were developed for trucks, road surface and
the composite girder bridge itself for studying bridge–vehicle interaction and the re-
sultant perceptible vibration [7] and KTX [8]. Ride comfort of elevated high-speed
trains was preliminary investigated basing on maximum vertical and lateral accelera-
tions in the train’s cars and the ratio of bridge deflection to the span length [9].
In this paper, riding comfort of high-speed railway is assessed by using a 3-D mul-
ti-car train-bridge system that enables computation of vertical, lateral, pitching and
rolling accelerations of all car bodies. A real train of fifteen vehicles over a slab-track
railway section having a bridge is investigated. The train car accelerations increase
largely when they are crossing the bridge. Based on the investigation results, im-
provement of the railway section design or the car-body suspensions may be required.

2 Modeling of a high-speed train system

The dimensions and parameters of a train car of Shinkansen (SKS) type are shown in
Fig. 1, where the letters Z, RX, RY, Y and RZ stand respectively for lateral, rolling,
yawing, vertical and pitching motions of a vehicle rigid body. The superscripts “c”,
“b” and “w” denote for car-body, bogie and wheel-set, respectively. n = 1,..., N where
N is the number of cars. Total DOFs of a train car is 27. It is assumed that: (i) Each
car consists of one car-body, two bogies and four wheelsets. These bodies are consid-
ered rigid, neglecting their elastic deformations during vibration and each wheelset
consists of two side wheels. (ii) The vertical or lateral suspension is idealized by two
identical linear springs of stiffness and two alike viscous dashpots of damping. (iii)
The motions of a car-body or a bogie are described by five DOFs, its mass and mass
moments. (iv) The motions of a wheel-set are represented by three DOFs, its mass
and mass moment. (v) Local deformations in the wheelsets under contact forces are
allowed. The 3-D contact model between a wheel-set and the rail is mechanically
presented by vertical contact stiffness kwV, and lateral contact stiffness kwH and damp-
ing cwH those can be approximated to be constants at a speed, and are zero values
Riding comfort assessment of high-speed trains … 223

when this wheel-set is jumping. (vi) The train and the bridge are considered as two
separate subsystems interacting with each other only at the wheel-rail interfaces.

Fig. 1. 3-D model, dimensions and parameters of a train car

The motion equations of each car-body are formulated by using the dynamic equi-
librium method with assumption of small vibration amplitudes [10]. For computation-
al efficiency, all the terms containing wheel-rail contact stiffness and damping are
moved to the right-hand side of the motion equations. The forces acting into the kth
wheel-set of the jth bogie due to rail responses and irregularities are expressed as

$%& & &

𝑃"# = 2𝑘$" *𝑧-# + 𝐼𝑅1,-# − 𝑟$ 5𝜃7,-# + 𝐼𝑅81,-# 9:
(1)
+ 2𝑐$" 5𝑧̇-# − 𝑟$ 𝜃̇7,-# 9
& &

$%& & &

𝑀7,# = 2(𝑘$? 𝑏AB + 𝑘$" 𝑟$B)5𝜃7,-# + 𝐼𝑅81,-# 9 − 2𝑘$" 𝑟$ 5𝑧-# + 𝐼𝑅1,-# 9 (2)
B B )𝜃̇ & &
+ 2(𝑐$? 𝑏A + 𝑐$" 𝑟$ 7,-# − 2𝑐$" 𝑟$ 𝑧̇-#
$%& & &
𝑃?# = 2𝑘$? 5𝑦-# + 𝐼𝑅1,-# 9 + 2𝑐$? 𝑦̇ -# (3)

& & &

In Eqs. (1) - (3), the variables 𝑦-# , 𝑧-# and 𝜃7,-# are the mean vertical, lateral,
& &
torsional displacements of the two rails at wheel position xnjk. The variables 𝑦̇ -# , 𝑧̇-#
̇ &
and 𝜃7,-# are the corresponding velocities. The terms IRy, IRz and IRyz are the eleva-
tion, alignment, and angular irregularities of the rail, respectively.

3 Modeling of bridge and substructure

The substructure including rail, track and bridge in this study is modelled separately
from the train cars, and adaptive to any types of finite elements. The choice of such an
element type depends on each bridge configuration. The ballastless prestressed con-
crete (PC) slab-track is adopted. Neglecting the small height of rail, the bridge re-
sponses at deck are approximately equal to that of the rail at every contact point. The
moving train and the rail of the substructure are interacting though their wheel-rail
interface, which must satisfy geometrical and equilibrium compatibilities.
224 V. N. Dinh et al.

The dynamic interaction forces acting onto the rail from the kth wheel-set are of
the same magnitudes and opposition direction to the forces acting onto the wheelset.
When the bridge is idealized by finite beam elements, the sectional eccentricity is
included. The forces acting on the bridge under each wheel position are concentrated
and consist of wheel-rail contact forces, gravity force and eccentric moment. The train
cars and bridge systems are separately and iteratively integrated.

4 Case study: Riding comfort of a fifteen-car train

A real high-speed train of fifteen identical vehicles passing over a railway section
consisting of a simply supported bridge at a constant speed shown in Fig. 2 is consid-
ered in this case study. The dimensions and dynamic properties of a train car are listed
in Table 1. The lateral and vertical eccentricities are 2.35 m and 1.2 m for the whole
bridge. The frequencies of the first vertical and lateral free vibration modes of the
bridge are 4.02 Hz and 12.387 Hz, respectively and those of the train cars are 0.723
Hz and 0.642 Hz. The differences in properties of soil that results in time-variations in
both amplitudes and frequency content [11] are not taken into account in this study
but in the future research. The random rail irregularities with the shortest wavelength
of 0.5 m and the longest wavelength of 300 m corresponding to the moderate rail
quality defined by the U.S. Federal Railroad Administration class 6 [12] is used.

Fig. 2. Geometry of train, bridge, railway section and rail irregularity profiles

Table 1. Mechanical properties and dimensions of a train car

Description & Name Unit Values (Cars numbers 1-15)

Mass: Mc, Mb, Mw ton 41.75, 3.04, 1.776
2
c c c b
J ,J ,J ,J ,J , J ,J
X Y Z X
b
Y
b
Z
w
X
ton.m 23.20, 2080.0, 2100.0, 1.580, 2.344, 3.934, 1.138

Stiffness: k1H, k1V, k2H, k2V kN/m 2350.0, 590.0, 176.0, 265.0
Damping: c1H, c1V, c2H, c2V kN.s/m 58.86, 19.62, 39.2, 45.12
b0, b1, h3, t, rw, b2, h2, h1 m 0.75, 1.0, 0.2, 1.25, 0.455, 1.23, 0.42, 0.75
Riding comfort assessment of high-speed trains … 225

(a) Moving time of wheel (s) (b) Moving time of wheel (s)

Positon of wheel (m) Positon of wheel (m)

Fig. 3. Vertical accelerations (a) and lateral acceleration (b) of the 1st car-body at V = 362 km/h

The simulated time-histories of the vertical and lateral accelerations of the first car-
body are shown in double abscissa- the moving time and wheel position in Fig. 3.
Due to the affection of bridge vibration under moving train, both accelerations of the
car-body are shown to increase significantly when it is crossing the bridge.
Table 2 presents the computed extreme responses and the corresponding allowable
values specified in the design codes used for modern high-speed railways such as the
KTX, the SKS [4] and the THSR [4]. The riding comfort criteria for pitching and
rolling accelerations of car-bodies have not been reported in the literature.
The comparisons show that the largest vertical acceleration of the 7th car-body oc-
curred at speed of 340 km/h is still acceptable for SKS system but exceeds the limits
of KTX and HST in criteria (1). Its largest lateral acceleration occurred at the speed of
160 km/h is also above the limit of HST in criteria (2). The largest lateral acceleration
experienced in the first car-body at speed of 400 km/h which is much larger that the
limit of HST in criteria (2). The ratio of bridge maximum vertical deflection (Δmax) to
its span length is always below the limit in criteria (2). The ride comfort of the con-
sidering railway section should therefore be improved by maintenance.

Table 2. Comparison of maximum computed responses with allowable values

Specification, g = 9.81 m/s2 This paper results (maxima)

Criterion Taiwan SKS KTX 1st car- 7th car- 14th car-
THSR (Japan) (Korea) body body body

(1) Max. vertical 0.1g 0.2g 0.05g 0.012 g 0.094 g 0.011 g

acceleration in (at 360 (at 340 (at 360
car-body km/h) km/h) km/h)

(2) Max. lateral 0.05g - - 0.08 g 0.053 g 0.03 g (at

acceleration in (at 400 (at 160 260 km/h)
car-body km/h) km/h)

(3) Δmax/L - 1/1800 1/1700 4.42 mm (1/7000) at 360 km/h

5 Conclusions

A method to assess riding comfort of high-speed railway by using 3-D multi-car train-
bridge interactive system has been proposed. A case of fifteen-car train travelling at
226 V. N. Dinh et al.

speed range 50-400 km/h over a slab-track railway section having a bridge and rail
irregularities FRA class 6. The computed car body accelerations and bridge deflection
are compared with the corresponding allowable values specified in the design codes
used for modern high-speed railways. Due to the bridge vibration under moving train,
the ride comfort of each train car reduces considerably when it is crossing the bridge.
The assessment shows that riding comfort of the train-railway system is poorer than
the limits and that the design of the railway section needs to be improved.

Acknowledgement

This study has been partially funded by Konkuk University, Seoul, Korea. The first
author (V.N. Dinh) has been funded by Science Foundation Ireland (SFI) Research
Centre: MaREI - Centre for Marine and Renewable Energy (12/RC/2302).

References
1. A. Minh: Selecting the direction for North-South high-speed railway. Vietnam Investment
Review, 27.02 (2019).
2. J. H. Lee, B. S. Jin, Y. Ji: Development of a structural equation model for ride comfort of
the Korean high-speed railway. Int. Journal of Industrial Ergonomics, 39, 7-14 (2009).
3. M. Dumitriu: Ride comfort enhancement in railway vehicle by the reduction of the car
body structural flexural vibration. IOP Materials Science and Engineering, 227 (2017).
4. Y. B. Yang, J. Yau, Y. Wu: Vehicle-bridge interaction dynamics: with applications to
high-speed railways, Singapore: World Scientific Publishing (2004).
5. P. Lou, Q. Y. Zeng: Formulation of equations of motion for a simply supported bridge
under a moving railway freight vehicle. Shock and Vibration, 14, 429–446 (2007).
6. K. Youcef, T. Sabiha, D. E. Mostafa, D. Ali, M. Bachir: Dynamic analysis of train-bridge
system and riding comfort of trains with rail irregularities. Journal of Mechanical Science
and Technology, 27(4), 951-962 (2013).
7. H. Moghimi, H. R. Ronagh. Development of a numerical model for bridge–vehicle
interaction and human response to traffic-induced vibration. Engineering Structures,
30(12), 3808–3819 (2008).
8. V. N. Dinh, K. D. Kim, J. S. Shim, E. S. Choi, S Suthasupradit: Development of three-
dimensional dynamic analysis model of high speed train-bridge interaction. Journal of the
Korean Society of Steel Construction (2008).
9. V. N. Dinh, K. D. Kim, P. Warnitchai, M. D. Dinh: Evaluation of passenger ride comfort,
noise and operation safety of elevated high-speed trains. In Proceedings of ISSOT-12,
International Association of Traffic and Safety Sciences, Bangkok, Thailand (2006).
10. V. N. Dinh: Dynamic analysis of bridge-train interaction through wheel-rail interfaces,
Bangkok: Ph.D Dissertation, Asian Institute of Technology (2009).
11. V. N. Dinh, B. Basu, R. B. J. Brinkgreve: Wavelet-based evolutionary response of multi-
span structures including wave-passage and site-response effects. Journal of Engineering
Mechanics, 140(8) (2014).
12. L. Frýba, Dynamics of railways bridges, Czech Republic: Thomas Telford House (1996).
 Assessment of methods of riprap size selections as scour
countermeasures at bridge abutments and approach
embankments

Huy Quang MAI1 and Noi Thi DOAN1

1
The University of Transport and Communications (UTC), Hanoi, Vietnam
mqhuy@utc.edu.vn

Abstract. In flood season, abutment scour is one of the main causes of bridge
damages or even bridge collapse resulting in the interruption of traffic and pos-
sibility death. The most commonly employed method of protecting bridge
abutments against scour is the application of bank-armoring method. Riprap is
the most common armoring scour protection method used at bridge abutments
and approach embankment. The selection of riprap size is based on the assump-
tion of riprap layer failure mechanisms. Despite the widespread use of riprap
protection in Vietnam, the guidelines for its design at bridge abutments and ap-
proach embankment are based on limited research. This article presents assess-
ment of methods of riprap size selections as scour countermeasures at bridge
abutments and approach embankments and procedure of design guidelines for
their uses.

Keywords: Abutment scour, countermeasure, riprap.

1 Introduction

Abutment is a part of the substructure of the bridge over the river, which is often di-
rectly or indirectly related to the flow. Abutment scour is a common cause of major
bridge damages during floods, often resulting in interruption to traffic flow, vehicle
damages, and even loss of life. For example, damage in the abutment of the Suoi Doi
II Bridge, Yen Bai province after the historical flood in October 2017 shown in Fig 1.
Bridge-scour countermeasures are methods to protect bridges from scour and channel
instability (Fig.2). Protection of bridge abutments and approach embankments from
scour includes countermeasures that alter flow and scour patterns and those that armor
the bed, bank, floodplain, and embankment slopes, such as riprap and cable-tied
blocks.
Rock riprap is one of the most common types of scour protection used at bridge
abutments and approach embankments. While the Federal Highway Administration
(FHWA) does not recommend riprap as a design countermeasure for piers on new
bridges, it can be used for new bridge abutments. Rock riprap can also be used as a
countermeasure for scour at piers and abutments for existing bridges. With a large
weight of riprap stones, they can resist the increased turbulence of the flow due to the

© Springer Nature Singapore Pte Ltd. 2020 227

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_33
228 H. Q. Mai and N. T. Doan

presence of the abutment in the flow range, effectively armoring the underlying sedi-
ments. Riprap stones are usually arranged on the slope of the embankment. In addi-
tion, riprap stones are also arranged at the toe of the abutment, usually as horizontal
apron around the abutment. The stones from the apron fall into the edge of the devel-
oping scour hole, deflecting the scour hole away from abutment and reducing the
scour depth, so the abutment foundation would be protected from scour.

Fig.1. Abutment scour at SuoiDoi II bridge, Fig.2. Example of using riprap stones as
YenBai province after the historical flood in 10/ protective method at the bridge abutment
2017

Despite the widespread use of riprap protection in Vietnam, the guidelines for its
design at bridge abutments and approach embankment are based on limited research.
In Vietnam, empirical studies on this issue are almost absent. Meanwhile, the riprap
stone size calculated according to the formulas of different authors gives much dis-
persed results, making it difficult for engineers to practice design. This article pre-
sents assessment of methods of riprap size selections as scour countermeasures at
bridge abutments and approach embankments and procedure of design guidelines for
their uses.

2 Riprap failure mechanisms

Riprap failure mechanisms depend on where they are arranged with respect to the
bridge abutments. Riprap that is placed in an apron at the foundation of an abutment
may be subjected to shear failure, edge failure, winnowing failure, and bed-form de-
stabilization [1, 3, 4]. Shear failure occurs where the individual riprap stones are not
large enough to resist entrainment by the flow. Scour of the riprap stones at the edges
of the riprap layer is termed edge failure, while winnowing describes the erosion of
the finer bed material through voids in the riprap layer. Filter layers are often placed
beneath the riprap to prevent winnowing failure. Shear failure may be triggered at the
edges of the riprap layer, i.e., shear and edge failure are often linked and are both a
consequence of undersized riprap. Bed-form undermining occurs under mobile-bed
conditions due to the migration of the troughs of large bed forms through the bridge
section. The riprap stones settle into the troughs of the passing bed forms, which may
destabilize the riprap layer. In general, riprap size selection can be based on stability
Assessment of methods of riprap size selections … 229

against shear and edge failure, as long as the other possible modes of failure are also
addressed appropriately.
The four possible failure mechanisms are shown schematically in Fig. 3 [2], which
also shows zones of riprap stability. In the diagram, d is bed sediment size d, u* is
shear velocity, while the subscripts c and r VLJQLI\ ³FULWLFDOFRQGLWLRQV IRUVKHDUHn-
WUDLQPHQW´DQG³ULSUDSVWRQHV´UHVSHFWLYHO\$VGHSLFWHGLQWKHGLDJUDPVKHDUIDLOXUH
occurs only when a riprap threshold condition, which can be measured in terms of
u*/u*cr, has been exceeded. Melville and Coleman [2] give an approximate value of
u* /u*cr=0.35 for the riprap entrainment threshold. As depicted, bed-form destabiliza-
tion occurs under mobile-bed conditions when u*/u*c >1, whereas the other failure
mechanism occurs in all flow conditions. The likelihood of winnowing failure in-
creases with increasing u*/u*c and at larger relative riprap sizes D/d while edge fail-
ure is more likely once the riprap threshold condition has been exceeded.

Fig.3. Failure mechanism of riprap protection aprons [2]

3 Determination of the riprap size

A key of element of the design of riprap countermeasures is rock sizing which is

based on equations generally derived from simplified laboratory experiments. The
increased weight of the riprap stones enables them to resist the increased velocities
and turbulence caused by the presence of the abutment in the flow, and thereby pro-
vide an armor layer protection to the underlying sediments. Interlocking forces be-
tween adjacent stones also act to stabilize the riprap layer. Typically, the riprap is
placed directly on the embankment slopes to protect the sediment from scour.
Various equations for sizing riprap at bridge abutments are listed in Table 1. These
equations can be rewritten in the general form:
230 H. Q. Mai and N. T. Doan

dr X
D
Fr E (1)
y  Ss  1 K slG

where dr is riprap size; X, D , E and G are coefficients given in table 1; Ss is specific

gravity if riprap stone; y is flow depth and Fr = V/(gy)0.5 is Froude number; V is ap-
proach flow velocity or contracted section velocity, depending on the author; g is
acceleration of gravity; K sl is the embankment slope factor given by the well-known
Lane (1955a) equation [2]:
sin 2 D
K sl 1 (2)
sin 2 T
where D is slope angle and T is angle slope of repose of riprap stones. Inherent in
the derivation of the above equation is an assumption that the flow at the bank surface
is in the direction of the channel axis, rather than oblique to the bank.

Table 1. Equations for sizing riprap at bridge abutments.

Standard format
No Reference Equation Comments
(for comparison)
1 Simons and 0, 4Vr2 dr 0,67 Spill-through
Lewis (1971) K Frr2 abutments
 s  1 gd r
S y  Ss  1
2 Croad (1989) dr 50 0,025Vb2 Ksl1 dr 50 0,91 Spill-through
Frr2 abutments
y  Ss  1 K sl
3 Brown and 0,006V 3 § S f ·
1,5
d r 50 0, 295S 1,5
f
Clyde (1989) d r 50 ¨ ¸ 1,5
Frr3
yK sl1,5 © 1, 2 ¹ y  Ss  1 K sl1,5

4 Pagan-Ortiz § 1,064V22 y20,23 ·

0,81
d r 50 1,05 Vertical wall
(1991) d r 50 Fr21,62 abutment
¨¨ ¸¸ y2  Ss  1
0,81

©  S s  1 g ¹

0,535V22 dr 50 0,535 Spill-through

d r 50 Fr22 abutments
 Ss  1 g y2  Ss  1
5 Austroads dr 50 1,026 dr 50 1,026
(1994) Fr 2 Fr 2
y  Ss  1 y  Ss  1
dr 50 Ks dr 50 Ks
Fr22 Fr22
y2  Ss  1 y2  Ss  1 Fr2 d 0,8
Richardson and
6
Davis (1995) dr 50 Ks dr 50 Ks
Fr20,28 Fr20,28
y2  Ss  1 y2  Ss  1 Fr2 ! 0,8
Assessment of methods of riprap size selections … 231

The graph on Fig.4 shows that the variation relations give a wide range of estimated
riprap stone size. The equation given by Croad (1989) is conservative with respect to
the other equations and is likely therefore to lead to conservative riprap size.

4 Example and Discussion

Given data: A bridge on National Highway number 32 over Suoi Doi spring has verti-
cal wall abutments. Hydraulic, foundation and scour data for bridge are as follow:
flow depth y0 = 1,87 m; flow velocity corresponding with design flow V1% = 1,22 m/s;
specific gravity of stone, Ss = 2,24; width of abutment footing 0,3m; width of abut-
ment 2,4 m; design scour depth at abutment ysmax = 1,68m.
Design of riprap countermeasure for abutments. Using these equations mentioned
in the table 1. Result of calculation of riprap stone size is given in Table 2. The result
of the stone size calculation shows that The Croad (1989)'s equation gives the greatest
result, while Pagan-Ortiz (1991)'s equation gives small value. The formula of Rich-
ardson and Davis (1995) gives the average value, often used in design.

0,30
Simons and Lewis (1971)
Croad(1989)
0,25
Brown and Clyde (1989)
Pagan-Ortiz(1991)
0,20
Austroad(1994)
Dr/y
Richardson and Davis (1995)
0,15

0,10

0,05

0,00
0,00 0,10 0,20 0,30 0,40 0,50 0,60
Fr1

Fig 4. Variation of Dr/y with Froude number of approach flow

Table 2. Equations for sizing riprap at bridge abutments.

Reference Dr unit
Simons and Lewis (1971) 0,14 m
Croad (1989) 0,23 m
Brown and Clyde (1989) 0,12 m
232 H. Q. Mai and N. T. Doan

Pagan-Ortiz (1991) 0,11 m

Austroads (1994) 0,21 m
Richardson and Davis (1995) 0,18 m

5 Conclusion

Using riprap to prevent scour at bridge abutments is a common method, low cost.
However, the selection of the right size of the riprap stone should consider the failure
mechanism of the stone to determine the rock's most dangerous working conditions.
There are many equations used to calculate the size of riprap stones, the results are
also much dispersed. In the future, more experimental research is needed, as well as
an assessment of the actual working conditions of riprap stones.

References:
1. Chiew, Y.M.: Mechanism of Riprap failure at Bridge Piers. Journal of Hydraulic Engineer-
ing, ASCE, 121(9),635-643 (1995)
2. Bruce W. Melville, Stephen E. Coleman.: Bridge Scour. Water Resources Publication, 550
pages. (2000)
3. Melville, Bruce & van Ballegooy, Sjoerd & E. Coleman, Stephen & Barkdoll, Brian.:
Riprap Size Selection at Wing-Wall Abutments. Journal of Hydraulic Engineering-ASCE -
J HYDRAUL ENG-ASCE. 133. (2007)
4. Parker, G., et al.: &RXQWHUPHDVXUHVWRSURWHFWEULGJHSLHUVIURPVFRXU8VHU¶VJXLGH1a-
tional Cooperative Highway Research Program, Transportation Research Board, National
Research Council, Washington, D.C. (1998)
 A review on protection methods against debris
accumulation for bridge in mountain areas

Huy Quang MAI1 and Phong Dang NGUYEN1

1
University of Transport and Communications (UTC), Hanoi, Vietnam
huymq@utc.edu.vn

Abstract. In the mountainous region of Vietnam, in the flood season, the phe-
nomenon of debris is very common, causing danger to bridges. Debris affects
the scour morphology at bridge piers, thus increasing the bridge failure poten-
tial. During the flood in October 2017, the Thia Bridge collapsed in Yen Bai,
Vietnam. After the survey, the large trees attached to the piers are believed to
be the main cause of this bridge collapse. Debris accumulation countermeasures
can be classified into two main categories, including structural and non-
structural methods. So, this paper presents details of these methods, and their
applicability in Vietnam.

Keywords: Bridge collapse, Pier scour, Debris.

Introduction

The scour problem caused by debris is dangerous not only in Vietnam, but also
through the world. Debris in particular tree trunk is often transported to stream in
heavy rainfall event. If debris reaches a bridge pier, it can accumulate on the pier and
creates a large obstacle of material to flow normally referred to as debris rafts. The
flow velocity will increase for that decrease in area of the waterway opening. The
additional flow obstruction generally causes local scour depth in excess of depth un-
der conditions without debris accumulation.
Vietnam is located in Southeast Asia with a total land area of 329,314 square kilome-
ters and a coastline of approximately more than 3,260 kilometers. Vietnam is prone to
QDWXUDOGLVDVWHUVLQFOXGLQJW\SKRRQVVWRUPVIORRGVGURXJKWVODQGVOLGHV«Especial-
ly, in the Northern mountainous region of Vietnam, the watersheds with steep slope,
debris such as wood, bamboo trees from upstream area of forested or wooded water-
sheds is often transported to stream during flood season. Therefore, the debris prob-
lem is one of the main causes threatening the bridge in this region during the rainy
and flood season. The presence of large accumulation of debris was a main reason in
the failure of Thia Bridge in Yen Bai province, Vietnam during the historic flood in
October 2017 (Fig.1).
In this area there are many old bridges, with weak bridge pier structures based on
shallow foundation. Bridge collapses such as Thia Bridge may continue to occur in
the future. Although debris accumulation at bridge has been recognized as a danger-

© Springer Nature Singapore Pte Ltd. 2020 233

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_34
234 H. Q. Mai and P. D. Nguyen

ous problem, in Vietnam, almost no research has been done to supply cost-effective
method to protect bridge piers from debris accumulations. This article focuses on
analyzing several protective methods for bridge pier against debris accumulation be-
ing used in the world, as well as, proposes suitable methods that can be applied in
Vietnam conditions.

Fig. 1. Thia Bridge collapsed during the historic flood, 10/2017

Debris accumulation countermeasures

Debris accumulation countermeasures can be classified into two main categories,

including structural methods such as debris sweepers, debris fins, and deflector, as
well as, non-structural methods, including watershed improvement.

2.1 Structural methods

Structures designed to deflect or redirect debris so that it travels between piers have
been investigated in a series of researches [1] [2] [3]. Structures of these types include
debris fins, river training structures, hydrofoils, debris deflectors and debris sweepers.
These structures are described in detail below.

Debris fins are walls built in the stream channel immediately upstream of the bridge
to align large floating trees so that their length is parallel to the flow, enabling them to
pass under the bridge without incident. This type of measure is also referred to as a
³SLHU QRVH H[WHQVLRQ´ These structures are recommended for control of medium to
large debris, require moderate levels of maintenance, and have relatively low envi-
ronmental impact when properly designed and installed [Bradley et al, 2005]. Exam-
ples of debris fin deflectors are provided in fig.2.

In channel debris basins are structures placed across well-defined channels to form
basins which impede the stream flow and provide storage space for deposits of detri-
tus and floating debris. These structures can be expensive to construct and maintain
A review on protection methods against debris … 235

(the basin must be cleaned out between storm events) and prevent up-
stream/downstream movement of aquatic organisms. Due to maintenance and ecolog-
ical concern, in channel debris basins are not a practical solution to debris problems in
majority of locations.
..

Fig. 2. Timber debris fin installation with sloping leading edge [1]

Fig. 3. In channel debris basin [1]

River-Training Structures are structures placed in the river flow to create counter-
rotating stream wise vortices in their wakes to modify the near-bed flow pattern to
redistribute flow and sediment transport within the channel cross section. Examples of
this type of structure include Iowa vanes, and impermeable and permeable spurs.

Crib Structures are walls built between open-pile bents to prevent debris lodging
between the bents. The walls are typically constructed out of timber or metal material.
Crib structures are relatively inexpensive to install, and the environmental impact of
the structures is minor. A moderate continuing investment is required to maintain the
structures [1].

Debris Deflectors are structures placed upstream of the bridge piers to deflect and
guide debris through the bridge opening. They are normally "V"-shaped in plan with
236 H. Q. Mai and P. D. Nguyen

the apex upstream. An example of this type of structure is shown in Fig.4. A special
type of debris deflector is a hydrofoil. Hydrofoils are submerged structures placed
immediately upstream of bridge piers that create counter-rotating streamwise vortices
in their wakes to deflect and divert floating debris around the piers and through the
bridge opening. Unfortunately, no hydrofoils have been implemented within the field.
They have only been tested within a physical model study

Fig. 4. Debris deflectors installed in an Indiana [Bradley 2005]

Debris Sweeper is a polyethylene device that is attached to a vertical stainless steel

cable or column affixed to the upstream side of the bridge pier. The polyethylene
device travels vertically along the pier as the water surface rises and falls. It is also
rotated by the flow, causing the debris to be deflected away from the pier and through
the bridge opening. This type of device is shown in Fig.5.

Fig. 5. Close up of a debris sweeper installed in a Tennessee river [4]

Design Features are structural features that can be implemented in the design of a
proposed bridge structure. The first feature is freeboard, which is a safety precaution
of providing additional space between the maximum water surface elevation and the
A review on protection methods against debris … 237

low chord elevation of the bridge. The second feature is related to the type of piers
and the location and spacing of the piers. Ideally, the piers should be a solid wall type
pier that is aligned with the approaching flow. They should also be located and spaced
such that the potential for debris accumulation is minimized. The third feature in-
volves the use of special superstructure design, such as thin decks, to prevent or re-
duce the debris accumulation on the structure when the flood stage rises above the
deck. The last feature involves providing adequate access to the structure for emer-
gency and annual maintenance.
Additional techniques including construction of booms, debris basins, and/or crib
structures at bridges prone to debris accumulation may help alleviate the problem [1].
An appropriate method of protecting a specific bridge should be selected from these
options based on the suitability of a structure for debris transport and channel form
characteristics at the site of interest, construction and maintenance costs, risk associ-
ated with failure of the structure(s), and aesthetic considerations.

2.2 Non-structural Methods

There are generally two types of non-structural measures available for bridge struc-
tures. The first type of non-structural measure is emergency and annual maintenance.
Emergency maintenance could involve removing debris from the bridge piers and/or
abutments; placing riprap near the piers, abutments, or where erosion is occurring due
to flow impingement created by the debris accumulation; and/or dredging of the
channel bottom. Annual maintenance could involve debris removal and repair to any
existing structural measures.
The second type of non-structural measure is management of the upstream watershed.
The purpose of this measure is to reduce the amount of debris delivered to the struc-
ture by reducing the sources of debris, preventing the debris from being introduced
into the streams, and clearing debris from the stream channels. The type of manage-
ment system implemented varies depending on the type of debris. For organic floating
debris, the management system could involve removing dead and decayed trees,
and/or debris jams; providing buffer zones for areas where logging practices exist;
implementing a cable-assisted felling of trees system; and stabilizing hillside slopes
and stream banks.

Applicability in Vietnam conditions

In order to select appropriate countermeasures to reduce debris accumulation at exist-

ing bridges and to assess the likelihood of debris accumulation at new bridges in
Vietnam conditions, we need to consider the following factors: (1) Stream and water-
shed characteristics influencing debris supply; (2) Stream characteristics influencing
debris transport; (3) Bridge characteristics affecting debris accumulation. These pa-
rameters should be surveyed in the field for each specific bridge, from which the most
appropriate method will be selected.
238 H. Q. Mai and P. D. Nguyen

Conclusion

A review of existing countermeasure to reduce debris accumulation at existing bridg-

es was presented, as well as, assesses the reliability and sustainability of selected
methods. These factors that need to be considered when selecting suitable measures
in the Vietnam conditions were also indicated.

References
1. Bradley, J.B., Richards D.L., and Bahner C.D.: Debris control structures - evaluation and
countermeasures. Federal Highway Administration Report No. FHWA-IF-04
016,Hydraulic Engineering Circular No. 9, 179 p (2005)
2. Richardson E.V. and Davis S.R.: Evaluating scour at bridges: Federal Highway Admin-
istration Report No. FHWA-IP-90-017, Hydraulic Engineering Circular No. 18, (3rd ed.),
204 p (1995)
3. Saunders S. and Oppenheimer M.L.: A method of managing floating debris. Hydraulic
Engineering 1993, ASCE 1373-1378 (1993).
4. Sheeder S. A. and Johnson P. A.: Controlling Debris at Pennsylvania Bridges. Report No.
FHWA-PA-2008-011-PSU 011, 219 p (2008)
 Numerical modelling of thermo-mechanical performance
of small-scale CFRP reinforced concrete specimen using
near surface mounted reinforcement method
Phi Long NGUYEN1, Xuan Hong VU2, Emmanuel FERRIER2
1
Ho Chi Minh City University of Transport, Faculty of Civil Engineering, Number 2, Vo Oanh
St. (Precedent name: D3 St.), Ward 25, Binh Thanh Disct., HCMC, Vietnam.
2
Université de Lyon, Université Lyon 1, Laboratory of Composite Materials for Construction
(LMC2), 82 bd Niels Bohr, F-69622 Villeurbanne, France.
(Mail: long.nguyen@ut.edu.vn; xuan-hong.vu@univ-lyon1.fr, emmanuel.ferrier@univ-lyon1.fr)

Abstract. Carbon fibre reinforced polymer (CFRP) is commonly used in civil

engineering in strengthening concrete structure such as slabs, beams, and col-
umns using externally bonding reinforcement method (EBR) or near surface
mounted method (NSM). Under thermo-mechanical condition that is close to
fire case condition, CFRP reinforced structures are under actions of mechanical
load and elevated temperature at the same time. In the literature, the evaluation
of the thermo-mechanical performance (such as fire performance) of CFRP-
reinforced concrete structures requires complicated-and-expensive experimental
works on full scale or large scale structures and therefore this may exceed time
limit and financial budget for multi-variables observation. This paper introduces
a hybrid method (numerical based and experimental validated method) to
evaluate the fire performance of CFRP reinforced concrete structure over a
small scale laminate-CFRP reinforced concrete specimen which is then capable
to evaluate fire performance of more general CFRP reinforced concrete struc-
tures.
Keywords: numerical modelling, carbon fibre reinforced polymer (CFRP), ele-
vated temperature, thermo-mechanical performance, small scale CFRP rein-
forced concrete specimen, near surface mounted reinforcement method.
1 Introduction
In civil engineering, CFRP is commonly used in reinforcing, strengthening, and retro-
fitting structures such as beams, slabs, columns due to its advantages in mechanical
properties as well as workability, and economic cost. The material can reinforce struc-
ture interior or exterior by being directly bonded to concrete surface (Externally
Bonding Reinforcement method - EBR) or via dipping CFRP in to trenches on con-
crete surface with epoxy adhesive (Near Surface Mounted reinforcement - NSM). The
implemented method depends on the structure type and site condition. However, fire
performance of CFRP reinforced structure is always an important concern because it
involves complicated scenario of elevated temperature and mechanical loads. The
consequence of failure, if happens, can cause loss of human and severe damage to
whole structure. Among the literature, material properties at different temperature
levels of constituent materials have been presented in the design standard for con-
crete, steel [1], as well as synthesized and reported for CFRP and adhesive material
[2±4]. For fire test at structure scale, several observations have been conducted to

© Springer Nature Singapore Pte Ltd. 2020 239

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_35
240 P. L. Nguyen et al.

evaluate the fire performance of CFRP reinforced concrete structures. This included
the performance of CFRP reinforced concrete beams and columns using EBR rein-
forcement method [5±7] or NSM reinforcement method [8]. Other residual tests on
NSM single lap specimen using epoxy and cement-based adhesives at elevated tem-
perature are also reported [9]. Two reinforcement methods (EBR and NSM) have
been compared for mechanical performance at room temperature condition [10, 11].
The results show that the NSM reinforcement method allows higher failure strain and
also increases the service duration [11]. Regarding fire-concerned issues, few numeri-
cal simulations were conducted: on bond performance of NSM CFRP repaired con-
crete prism [9], or insulated EBR CFRP strengthened T-beam subjected to mechanical
loading [12]. In this study, the authors would like to introduce a numerical model
which is able to characterize the thermo-mechanical performance of CFRP reinforced
concrete structure with the validation from experimental test. This model is extend-
able to characterize the behaviours of CFRP reinforced concrete structure at several
thermo-mechanical conditions including close-to-fire conditions.
2 Experimental Approach
In this study, the thermo-mechanical system [3] was used to experimentally study the
performance of CFRP reinforced concrete specimen at different temperature-
mechanical conditions. The specimen and testing procedure are detailed in the follow-
ing subsections.
2.1 Specimens and Materials

Fig. 2 Specimen extension under tem-

perature and mechanical load.
Fig. 1 Design of the experimental sample
Specimen design: To study the bond behaviour of CFRP-concrete interface, sev-
eral methods were used in the literature such as single lap, double lap [13, 14] or
even beam test [15] to determine the shear capacity of the bond. In this study, the
standard double shear method [16] was developed to study the bond behaviour of
CFRP-concrete interface under thermo-mechanical condition. The double lap testing
specimen includes two concrete blocks connected by two laminate CFRP strips using
near surface mounted reinforcement method (NSM) (Fig. 1). The dimension and ma-
terial grain sized were modified to meet the capacity of apparatus. The steel system in
the core was designed to connect the specimen with two loading heads and also to
reduce tensile effects on concrete as well as to better exploit compressive strength of
concrete. The CFRP-concrete bonded areas were designed to be equal on both sides
(10mm x 70mm for observed block and 10mm x 90mm for tie block).
Materials: The gravels have been sieved with net size of 6mm to meet the re-
quirement for the maximum size of the CFRP/concrete specimen tested in the furnace.
The concrete was mixed with ratio presented in Table 1 to obtain class C25/30 after
28 days. The bonded CFRP is a uni-directional prefabricated laminate with 68% fi-
Numerical modelling of thermo-mechanical performance of small-scale … 241

bres in volume. Its Young modulus is 165 GPa and its ultimate strength is 2800 MPa
according to VXSSOLHU¶V datasheet. A two-component high-temperature-performance
adhesive, filled with Al2O3 and metal, has been used to bond CFRP to concrete
blocks. Its tensile strength at 20°C is about 81.4 MPa. Its temperature range can reach
350°C for short term and 320°C for long term application according to suppliHU¶V
datasheet.
Table 1. Concrete mixing ratio (for 1m3 of concrete)
Cement, 32.5 (kg) Water (kg) Gravel (G”PPNJ 6DQGG”PPNJ
336 180 1262 622
2.2 Test procedures
In this experiment, the specimen was tested following a thermo-mechanical regime
with constant load condition [3]. Firstly, the specimen was imposed with a load called
applied force (Fw). Afterwards, the temperature surrounding sample increases with the
ramp rate up to 30°C/min from ambient temperature until its rupture temperature
(maximum 1100°C due to safety recommendation of the furnace). The temperature, at
which the specimen collapses, is identified as the failure temperature (T r) under the
load level Fw. The duration, from when temperature starts increasing until when
specimen collapses, is identified as the exposure duration under applied load level F w.
During the test, the temperatures along the vertical side of specimen were measured
via three thermocouples. During the experimental test, when the applied mechanical
load (Fw(t)) is controlled stable and temperature T(t) is increased, the total specimen
length (L(T(t),Fw(t)), Fig. 2) at different temperature (T(t)) and mechanical load
(Fw(t)) are monitored. In this experimental study, two mechanical load cases have
been applied at at Fw = 840 N and at Fw = 1400 N. In the following sections, the re-
sults, which are later compared with numerical model, are presented and discussed.
3 Numerical Investigation
The finite element model has been generated to study the heat transfer effect within
the CFRP reinforced concrete specimen at different temperature-mechanical condi-
tions that were presented in the experiment. In this study, the numerical model fo-
cuses on the simultaneous effect of elevated temperature and mechanical loads on the
tested specimens via coupled-field analysis: thermal analysis and mechanical analysis.
The temperature history recorded in the experimental test was regenerated to obtain
the history result of temperature distribution across the specimen. The mechanical
load is then applied on the specimen to obtain the result of simultaneous combined-
effects at different points of time on the model.
3.1 Numerical model
Due to the symmetry of loading, boundary conditions and material, a fourth model
was generated and analysed using ANSYS APDL (Fig. 3a). This can accelerate the
computation by reducing elements while maintain its reliability. The element type for
the heat transfer analysis is SOLID70 (3D 8-nodes thermal solid) and SOLID45 for
stress analysis. It is because the failure mode of three experiments shows that CFRP,
concrete and steel are little affected in comparison with the used adhesive. The CFRP-
concrete bond is modelled based on ³cohesive bilinear zone material´ model intro-
242 P. L. Nguyen et al.

duced by Alfano et. al. [17] with coefficients that vary as temperature proposed by
Arruda et. al.[18]. According to this model, the behaviour of the CFRP-concrete bond
depends on both temperature and mechanical load. The meshed is finer at the contacts
between CFRP and concrete, steel and concrete and coarser in concrete material (Fig.
3b).

Fig. 3 FE model: a) 3D model; b) Meshed elements;

3.2 Material properties
In the literature as well as design standards, there is available data for thermal and
mechanical properties of concrete, steel and laminate CFRP. Therefore, as summa-
rized and reported in the literature, the material properties at ambient and elevated
temperatures in this FE model are adopted from Eurocode 2 [19] for concrete, from
Eurocode 3 [1] for steel and from experimental results obtained by the previous stud-
ies of the authors [3]. The thermal and mechanical boundary conditions of FE model
are based on the exterior temperature evolution monitored by three thermocouples
along the side of specimens and force sensor (or load cell) during the experiment.
4 Results and discussion
This section presents experimental and numerical results and then the discussion.
4.1 Experimental results

Fig. 4 Force-time and temperature-time Fig. 5 Total displacement- time curves at

curves at different mechanical loads different mechanical loads

Fig. 4 and Fig. 5 display the evolution of temperature, mechanical load and total
displacement as function of time. The value of temperature and mechanical load are
recorded by thermo-couples and load sensor while the total displacement of specimen
in the following sections is determined as  7RWDO GLVSODFHPHQW ǻ/ /[T(t),
Fw(t)=const] - L[T(t=0))= Room Temperature, Fw(t=0)=const]. According to the re-
sult (Fig. 4), the average temperature rate is about 24.4°C/minute and the furnace
temperature reaches 750°C after 30 minutes. The test is terminated when mechanical
load drops and the service duration is calculated based on this time point. As also
shown in the figure 5, under mechanical load 840N, the service duration of specimen
is about 18-26 minutes while at higher mechanical load 1400N, the service duration is
Numerical modelling of thermo-mechanical performance of small-scale … 243

about 8-12 minutes. As shown in Fig. 5, under elevated temperature and mechanical
load actions, the specimen is extended. The total displacement of specimen gradually
increases at the beginning but becomes more rapid before final failure. As the applied
mechanical load (Fw) increases, the total displacement reduces significantly.
4.2 Numerical results and discussions
Fig. 6 displays the total-displacement vs. time relation obtained in the numerical
solution and experimental result at two mechanical load cases. In each numerical
model, the thermal-stress analysis results are exported at different point of time, cor-
responding to the progress of experimental test. As can be seen in the figure 6, the
development of total displacement obtained from numerical result well fits with total
displacement curves obtained from experimental data. The harmony between numeri-
cal simulation and experimental result shows that the thermo-mechanical behaviour of
CFRP reinforced concrete structure can be numerically investigated via couple-field
analysis in finite element method. Further analysis can be considered to better under-
standing the properness of numerical model in characterizing the thermo-mechanical
conditions even which are close-to-fire case.

Fig. 6 Total displacement vs. time curves at two mechanical loads (obtained by nu-
merical modelling (*.N) and experiment (*.E))
5 Conclusion
This study has successfully investigated the performance of CFRP reinforced concrete
specimen, which is under service load and elevated temperature condition, with the
combined experiment and numerical solutions. The actual temperature evolution on
the experiment has been considered and then reproduced in FE models. The appropri-
ateness of simulation to experiment shows that with adopted material data, the FE
model is able to estimate the thermo-mechanical performance of NSM CFRP rein-
forced concrete specimen under simultaneous elevated temperature and mechanical
load condition. This allows to predict the thermo-mechanical response of CFRP rein-
forced concrete structures under complicated thermal and mechanical loads and thus
feasible for fire-performance analysis in civil engineering.
6 Acknowledgment
This research is performed with the financial support of the LMC2 and from doctoral
scholarship by the Ministry of Education and Training of Vietnam for the first author.
244 P. L. Nguyen et al.

7 References
1. EN 1993-1-2: Eurocode 3: Design of steel structures - Part 1-2: General rules
2. Firmo, J.P., Correia, J.R., Bisby, L.A.: Fire behaviour of FRP-strengthened reinforced
concrete structural elements: A state-of-the-art review. Composites Part B: Engineering.
80, 198±216 (2015).
3. Nguyen, P.L., Vu, X.H., Ferrier, E.: Characterization of pultruded carbon fibre reinforced
polymer (P-CFRP) under two elevated temperature-mechanical load cases: Residual and
thermo-mechanical regimes. Construction and Building Materials. 165, 395±412 (2018).
4. Nguyen, P.L., Vu, X.H., Ferrier, E.: Elevated temperature behaviour of CFRP applied by
hand lay-up (M-CFRP) under simultaneous thermal and mechanical loadings: Experi-
mental and analytical investigation. Fire Safety Journal. 100, 103±117 (2018).
5. 7XUNRZVNL 3 àXNRPVNL 0 6XOLN 3 5RV]NRZVNL, P.: Fire Resistance of CFRP-
strengthened Reinforced Concrete Beams under Various Load Levels. Procedia Engi-
neering. 172, 1176±1183 (2017).
6. Bisby, L.A., Green, M.F., Kodur, V.K.R.: Response to fire of concrete structures that
incorporate FRP. Progress in Structural Engineering and Materials. 7, 136±149 (2005).
7. Firmo, J.P., Correia, J.R., Pitta, D., Tiago, C., Arruda, M.R.T.: Experimental characteriza-
tion of the bond between externally bonded reinforcement (EBR) CFRP strips and concrete at
elevated temperatures. Cement and Concrete Composites. 60, 44±54 (2015).
8. Al-Abdwais, A., Al-Mahaidi, R., Al-Tamimi, A.: Performance of NSM CFRP strength-
ened concrete using modified cement-based adhesive at elevated temperature. Construc-
tion and Building Materials. 132, 296±302 (2017).
9. Jadooe, A., Al-Mahaidi, R., Abdouka, K.: Modelling of NSM CFRP strips embedded in
concrete after exposure to elevated temperature using epoxy adhesives. Construction and
Building Materials. 148, 155±166 (2017).
10. Firmo, J.P., Arruda, M.R.T., Correia, J.R.: Contribution to the understanding of the
mechanical behaviour of CFRP-strengthened RC beams subjected to fire: Experimental
and numerical assessment. Composites Part B: Engineering. 66, 15±24 (2014).
11. Kotynia, R.: Bond between FRP and concrete in reinforced concrete beams strengthened
with near surface mounted and externally bonded reinforcement. Construction and
Building Materials. 32, 41±54 (2012).
12. Hawileh, R.A., Naser, M., Zaidan, W., Rasheed, H.A.: Modeling of insulated CFRP-
strengthened RC T-beam exposed to fire. Engineering Structure 31, 3072±3079 (2009).
13. Camli, U.S., Binici, B.: Strength of carbon fiber reinforced polymers bonded to concrete
and masonry. Construction and Building Materials. 21, 1431±1446 (2007).
14. Mazzotti, C., Savoia, M., Ferracuti, B.: An experimental study on delamination of FRP
plates bonded to concrete. Construction and Building Materials. 22, 1409±1421 (2008).
15. Chen, J.F., Teng, J.G.: Proceedings of International Symposium on Bond Behaviour of
FRP in Structures (BBFS 2005). International Institute for FRP in Construction (2005)
16. Ferrier, E., Quiertant, M., Benzarti, K., Hamelin, P.: Influence of the properties of exter-
nally bonded CFRP on the shear behavior of concrete/composite adhesive joints. Com-
posites Part B: Engineering. 41, 354±362 (2010).
17. Alfano, G., Crisfield, M.A.: Finite element interface models for the delamination analy-
sis of laminated composites: mechanical and computational issues. International Journal
for Numerical Methods in Engineering. 50, 1701±1736 (2001).
18. Arruda, M.R.T., Firmo, J.P., Correia, J.R., Tiago, C.: Numerical modelling of the bond
between concrete and CFRP laminates «. Engineering Structures. 110, 233±243 (2016).
19. EN 1992-1-2: Eurocode 2: Design of concrete structures - Part 1-2: General rules. CEN.
Experimental characterization of multi-full-culm bamboo
to steel connections

Nischal P. N. Pradhan1, Themelina S. Paraskeva2, Elias G. Dimitrakopoulos1,3

1 Department of Civil and Environmental Engineering, The Hong Kong University of Science
and Technology, Clear Water Bay, Kowloon, Hong Kong
2
School of Engineering, Edith Cowan University, Joondalup WA 6027, Australia
3 ilias@ust.hk

Abstract. The present research examines the performance of newly developed

multi-full-culm bamboo to steel connections under monotonic axial loading. The
culms are of Kao Jue (Bambusa pervariabilis) bamboo species. Findings reveal
that the plain (unreinforced) connections fail early by undesirable brittle longitu-
dinal splitting of bamboo culms. The confinement provided by hose-clamps in-
hibits this brittle failure mode, and with sufficient end-lengths, drastically in-
creases the strength and ductility of the connection. Compared to the hollow-
section connections with hose clamps, adding mortar infill further increases the
strength. However, it also restricts bolt-deformation and thus diminishes the duc-
tility. More importantly, the European Yield Model (which refers to dowelled
timber connections) can analytically estimate the obtained experimental yield
loads with satisfying accuracy. This is a promising direction towards a more ra-
tional and safer structural design of bamboo structures.

Keywords: bamboo structures, full-culm bamboo, round bamboo, bamboo con-

nections, European Yield Model.

1 Introduction

Bamboo is a sustainable construction material [1] with remarkable mechanical proper-

ties. The full-culm or round bamboo is among the oldest material for traditional housing
applications, scaffoldings and bridges [2]. However, the shortage of design codes and
lack of reliable practical engineered connections means the contemporary bamboo
structural design and construction employ mostly ad-hoc or experience-based ap-
proach. This has among other reasons, limited a more widespread utilization of bamboo
as a structural material. Therefore, there is a need of design frameworks [3,4] with re-
liable engineered joints or connection systems [5] to enable rational engineered design
of bamboo structures. This study focuses on the said challenges, and proposes an engi-
neered solution utilizing full-culm bamboo members for truss structures.
The present research utilizes connections with metal fasteners [6±8] as they are ubiq-
uitous, versatile and easy to use [9], and evaluates the performance under monotonic
axial loading. The motivation originates from the need to provide practical engineered

© Springer Nature Singapore Pte Ltd. 2020 245

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_36
246 N.P. N. Pradhan et al.

connections of predictable performance as a sustainable solution for the developing

areas, e.g. [10]. The goal is to characterize experimentally the performance of the con-
nections under monotonic loading.

2 Methodology

Fig. 1 shows experimental test specimens with three types of connections. All bamboo
culms used in this study have external diameter ranged within 40-60 mm and were
screened for imperfections in their geometry. Selecting culms with small variations
along their profiles reduces the difficulty in fabricating and assembling the test speci-
mens. The culms have average external diameter 47.42 mm (SD 3.90) and average wall
thickness 6.05 mm (SD 1.45) measured at the culm-ends as per [11]. The average mois-
ture content is 10.41% (SD 1.64), where SD stands for standard deviation. All steel
components are stainless steel of grade SS304, with the steel plates of thickness 3 mm
and the hose-clamps of size 40-63 mm. The fully-threaded bolts and the nuts are of type
A2-70 and size M6. The oversized bolt-holes in the culm-walls and steel plates provide
1 mm of clearance. The center-to-center bolt-spacing are at 50 mm distance, while the
end-lengths are either 50 mm or 80 mm center-to-center. The compression specimens
have only 50 mm end-lengths.
All tests were conducted at the Structural Engineering Laboratory of The Hong Kong
University of Science and Technology using the MTS 810 universal testing machine.
The nomenclature used for the specimens is similar to that in [12] starting with "A",
"B" or "C" for connection type followed by "50t" or "80t" tension loading

movable steel hose grouted

steel plate crosshead clamps mortar
50 mm
or
80 mm
c/c
800 mm

bamboo
M6
culms
bolts
Z 50
mm
Y
c/c
X
movable
crosshead XZ YZ XZ YZ XZ YZ
(a) (b) (c) (d)

Fig. 1. (a) Type A (plain), (b) Type B (confined with hose-clamps), and (c) Type C (confined
with hose-clamps and infilled with mortar); with (d) experimental setup
Experimental characterization of multi-full-culm bamboo … 247

with 50 mm or 80 mm end-lengths or "c" for compression loading, and ending with the
specimen numbers 01 to 05. The rate of loading is set to 0.01 mm/s.

3 RESULTS

3.1 Specimen yielding and damage modes

The Type A connections (plain) exhibit European Yield Model (EYM) Modes IIIs
(hollow section) [13] yielding after which they fail by brittle longitudinal splitting of
culm walls (Fig. 2a). The Type A connections show this failure mode regardless of
increase in the end-lengths (e.g. A80t) or change in the loading directions (e.g. Ac).
The Type B connections also display EYM Mode IIIs (hollow section) yielding but
are protected against longitudinal splitting of culms through transverse confinement
from hose-clamps. After yielding, the deforming bolts transversely compress and
ovalize the culm-ends. This results in crushing and row-shearing (e.g. B50t specimens,
Fig. 2b), causing connection failure in B50t01-03 specimens. Under further loading the
bolts embed extensively (Fig. 2c) into the culms and eventually fracture, resulting in
connection failure for the remaining Type B50t specimens. The B80t specimens due to
their longer end-lengths and Bc specimens due to the reversed loading direction, resist
crushing and do not suffer row-shearing ² they fail exclusively by bolt-fracture.
The Type C connections have mortar infill in addition to confinement from hose-
clamps. Within the end-length zones, the infill debonds easily and allows the outer bolts
to yield by Mode IIIs (hollow section) mechanism. In contrast, the infill restricts move-
ment of inner bolts, forcing them to yield by an intermediate mode in-between Mode
IIIs and Mode IV. For Type Cc specimens, both the inner and outer bolts display the
intermediate yielding mode due to the reversed loading direction (Fig. 2d). The deform-
ing bolts push the infill outwards and the outside-facing culm-walls develop multiple
longitudinal cracks (Fig. 2e). The longer end-lengths in Type C80t specimens prevent

Fig. 2. (a) splitting failure in Type A connections; (b) crushing and row-shearing, and (c) em-
bedment with bolt-fracture in Type B connections; (d) embedment and bolt-fracture, and (e) in-
fill push-out and multiple cracking of outer walls in Type C connections
248 N.P. N. Pradhan et al.

(a) (b)

(c) (d)

Fig. 3. Box plots of (a) load-carrying capacity, (b) displacement at load-carrying capacity, (c)
normalized load-carrying capacity and (d) ductility ratio

this multiple cracking effect. The stiff infill prevents the culm-ends from crushing lo-
cally.

3.2 Performance of the connections

Fig. 3 presents the performance of the examined connections. The significant im-
provement of the axial response of Type B over Type A specimens indicate the high
impact of steel hose clamps on the capacity of the examined connections. Transverse
confinement significantly increases the maximum load or the failure load that a con-
nection can attain (represented by load-carrying capacity, Fig. 3a). This improves the
efficiency by increasing the maximum developed axial stresses within the culms (rep-
resented by the normalized load-carrying capacity, Fig. 3c). This also translates to dras-
tic gains in obtainable axial displacements at load-carrying capacity (Fig. 3b), thus con-
tributing to superior ductility of Type B specimens compared to Type A (Fig. 3d). In-
filling with mortar leads to additional increase in strength over Type B (Fig. 3a and c),
but it restricts bolt-deformation and thus reduces displacement at failure (Fig. 3b) and
ductility (Fig. 3d).

3.3 Connection yield loads

This section adopts the European Yield Model (EYM) [13,14], first implemented in
Experimental characterization of multi-full-culm bamboo … 249

Table 1. Analytical yield loads of the examined specimens (in kN) using EYM

No. A50t Ac A80t B50t Bc B80t C50t Cc C80t

1 8.22 9.12 9.34 7.36 8.63 8.01 11.71 11.18 12.63
2 8.40 8.30 9.09 7.58 12.02 7.04 11.57 11.62 12.52
3 8.58 9.71 9.39 8.13 6.81 9.50 12.04 12.09 12.01
4 8.99 7.94 8.03 8.03 9.63 7.98 11.84 11.42 12.28
5 8.55 9.35 10.15 7.59 7.25 8.83 12.24 12.65 11.89

Table 2. Experimental yield loads of examined specimens (in kN), using 5% offset method

No. A50t Ac A80t B50t Bc B80t C50t Cc C80t

1 7.50 7.85 7.75 7.39 9.92 8.56 9.08 16.17 12.55
2 6.99 8.31 7.86 10.47 12.96 8.51 11.78 12.77 11.90
3 6.30 11.82 6.75 6.70 7.61 9.90 10.76 12.65 8.99
4 10.54 7.14 7.18 8.89 11.27 9.01 10.38 9.90 13.55
5 7.88 9.48 9.99 9.72 6.97 9.03 9.79 12.66 9.44

multi-culm bamboo to steel connections by [12]. The EYM describes multiple possible
modes of yielding that depends upon the geometry of connection components ( [12]),
embedment strength of bamboo (i.e., fes = 51.99 MPa and fes = 48.69 MPa for hollow
and solid sections respectively [12]) and the yield moment of the bolts (My = 14880.54
Nmm [12]). To that end, the Type A and B specimens yield by Modes IIIs hollow sec-
tion. The analytical yield loads of those connections are twice the Mode IIIs (hollow
section) yield load of a single bolt. For Type C specimens, the analysis assumes that
the inner bolts yield by pure Mode IV (solid section) while the outer bolts yield by pure
Mode IIIs (hollow section). Accordingly, the yield load of Type C connections is esti-
mated by summing the computed EYM yield loads of the inner and outer bolts. Results
(Table 1) show that the yield loads of examined specimens calculated using the EYM
approach agrees well with the experimental yield loads (Table 2) obtained using the 5%
offset method [15].

4 Conclusion

Plain bolted connections display poor performance due to premature failure by longi-
tudinal splitting of the culms. For this reason, their use should be discouraged and the
culms transversely confined. Providing transverse confinement through hose-clamps
prevent this failure mode and brings significant improvements in performance. Com-
bining this with sufficient end-lengths increase the resistance of the culms to crush-
ing/row-shearing and further improves the ductility. Mortar infill also reinforces the
bamboo culms against crushing and row-shearing, and increases the overall strength of
250 N.P. N. Pradhan et al.

the connections. However, the infill reduces ductility by restricting bolt deformation.
Importantly, the study shows that the European Yield Model (that refers to dowelled
timber connections) can analytically predict the yield loads obtained experimentally
with satisfying accuracy. This indicates a path towards a more rational engineered de-
sign of bamboo to steel connections. Lastly, despite the gains in performance of the
reinforced connections over the plain, there is still ample headroom in further improve-
ment in design in terms of greater strength and efficiency.

References
1. E.Z. Escamilla, G. Habert, J.F.C. Daza, H.F. Archilla, J.S.E. Fernández, D. Trujillo,
Industrial or traditional bamboo construction? Comparative life cycle assessment (LCA) of
bamboo-based buildings, Sustain. 10 (2018). doi:10.3390/su10093096.
2. G. Minke, Building with Bamboo, Birkhäuser, Basel, 2012. doi:10.1515/9783034611787.
3. A. Gatóo, B. Sharma, M. Bock, H. Mulligan, M.H. Ramage, Sustainable structures: bamboo
standards and building codes, Proc. Inst. Civ. Eng. - Eng. Sustain. 167 (2014) 189±196.
doi:10.1680/ensu.14.00009.
4. K.A. Harries, B. Sharma, M. Richard, Structural Use of Full Culm Bamboo: The Path to
Standardization, Int. J. Archit. Eng. Constr. 1 (2012) 66±75. doi:10.7492/IJAEC.2012.008.
5. J.J.A. Janssen, Technical Report 20: Designing and building with bamboo, International
Network for Bamboo and Rattan, 2000.
6. M. Sassu, A. De Falco, L. Giresini, M.L. Puppio, Structural solutions for low-cost bamboo
frames: Experimental tests and constructive assessments, Materials (Basel). 9 (2016).
doi:10.3390/ma9050346.
7. Morisco, F. Mardjono, Strength of filled bamboo joint, in: I.V.R. Rao, C.B. Sastry, P.M.
Ganapathy, J.J.A. Janssen (Eds.), Proc. Vth Int. Bamboo Work. IV Int. Bamboo Congr.,
International Development Research Centre, Bali , Indonesia, 1995.
8. A. Awaludin, V. Andriani, Bolted bamboo joints reinforced with fibers, Procedia Eng. 95
(2014) 15±21. doi:10.1016/j.proeng.2014.12.160.
9. D.J.A. Trujillo, D. Malkowska, Empirically derived connection design properties for
Guadua bamboo, Constr. Build. Mater. 163 (2018) 9±20.
doi:10.1016/j.conbuildmat.2017.12.065.
10. T.S. Paraskeva, G. Grigoropoulos, E.G. Dimitrakopoulos, Design and experimental
verification of easily constructible bamboo footbridges for rural areas, Eng. Struct. 143
(2017) 540±548. doi:10.1016/j.engstruct.2017.04.044.
11. ISO 22157-1:2004(E), Bamboo ² Determination of physical and mechanical properties ²
Part 1: Requirements, International Organization for Standardization, Genveva, Switzerland,
2004.
12. T. Paraskeva, N.P.N. Pradhan, C.D. Stoura, E.G. Dimitrakopoulos, Monotonic loading
testing and characterization of new multi-full-culm bamboo to steel connections, Constr.
Build. Mater. 201 (2019) 473±483. doi:10.1016/j.conbuildmat.2018.12.198.
13. Technical Report 12, General Dowel Equations for Calculating Lateral Connection Values
with Appendix A, American Wood Council (AWC), Leesburg, VA, 2015.
14. W.R. Parsons, Energy-based modelling of dowel-type connections in wood-plastic
composite hollow sections, Washington State University, 2001.
15. F1575-03, Standard Test Method for Determining Bending Yield Moment of Nails, ASTM
International, West Conshohocken, PA, 2003. doi:10.1520/F1575-03R08.
 FE modelling of RC frames with Link Column Frame
System under in-plane loading
J. Joel Shelton 1*, G. Hemalatha2 and V. Venkatesh3

1*
Assistant Professor, Department of Civil Engineering, RGM College of Engineering and
Technology, Kurnool, joelsheltonj@gmail.com
2
Associate Professor, Department of Civil Engineering, Karunya Institute of Technology and
Sciences, Coimbatore, hemalathag@karunya.edu
3
$VVLVWDQW 3URIHVVRU 'HSDUWPHQW RI &LYLO (QJLQHHULQJ 6W 0DU\¶V ,QWHJUDWHd Campus,
Hyderabad, venkat@karunya.edu.in

Abstract. The purpose of the link column frame is to provide acceptable collapse prevention
performance and also easily repaired following a moderate earthquake. Linked column frame
system (LCF) for steel structures was proposed by Peter Dusicka et.al (2009) with the objective
of utilizing replaceable components that are strategically placed to protect the gravity load
carrying system. In this paper the concept is extended to Reinforced concrete structures. The
analytical studies were presented in this paper mainly focuses on the behavior of normal and link
column RC frames with different connection configurations. Quasi-static cyclic load analytical
results are presented and discussed for three 1:3 scaled RC frames with and without link column.
Link column with various connections between the main beam and the link column i.e. rigid and
hinged connection as per IS 12303-1987. With respect to experimental results, by using ANSYS
software, the finite element model related to these frames is made and calibrated, and then
analysis under cyclic static loading are performed. The test results showed a significant increase
in the energy dissipation with a decrease in relative storey drifts for the link column frames with
hinged connection. Greater amounts of energy were dissipated by the link column frame which
has a hinge connection designed according to IS 12303-1987.

Keywords. Storey drifts, gravity load, Link column frame, hinged connection, Energy
Dissipation.

1 Introduction

To perform acceptable collapse prevention performance during the seismic action the
structures must have adequate strength, stiffness, redundancy, ductility and adequate
energy dissipation [1]. During major earthquake, high energy dissipation and high
elastic stiffness were achieved by this EBFs.[2]. Using ANSYS software, the
Reinforced concrete braced frame was modelled and analyzed for its nonlinear behavior
which was obtained from previous experimental results by Maheri and Memarzadeh
(2001) [3]. Dusicka and Iwai [4] in the year 2007 discovered a new seismic load
resisting system for the steel structures which is called as Link Column Frame system
(LCF). In the RC framed structure, the link column frame (LCF) system was studied
analytically in this paper its elevation was shown in Fig.1. The work of the link beam
is to yield in shear which is placed in between the link column and create a plastic

© Springer Nature Singapore Pte Ltd. 2020 251

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_37
252 J. J. Shelton et al.

mechanism to save the nearby gravity frame in which the main beam was flexibly
designed i.e. one end is fixed and another end is hinged. The link beam will dissipate
energy and provide ductile behavior during lateral loading by sacrificing itself by
yielding and reduces the nonlinear behavior in the adjacent main frame. Analytical
investigations on the full scale models has been carried out by the authors and the
performance of the system was satisfactory [5].

Fig. 1. Elevation of the Link Column Frame System.

2 Model for the Study

For Analytical investigation single bay frame was considered. 1:3 ratio scaled down RC
frame specimens were cast and later on two of the specimens were made of linked
column with different connections. The static cyclic load tests were performed. The
nominal capacity and link beam length of LCF system which should be yield in shear
based on the UBC codal provisions [6]. By using Vierendeel column approach and
Cantilever column approach, the section of the link and link column was designed [7].
The reinforcement details and dimensions of a bare frame(M1), LCF with the rigid
connection(M2) and LCF with hinged connection shown in Figure 2. Hinge connection
were made by using Mesnager hinge as per IS 12303:1987 [8].

Figure 2a. Reinforcement details of bare frame M1(All dimensions are in mm).
FE modelling of RC frames with Link Column Frame … 253

plastic hinge length Lp in the Reinforced concrete, Mattock (1967) [9] has derived the
equation and it was given in Eqn. (1).

‫ܮ‬௣ ൌ ͲǤͲͷ‫ ݖ‬൅ ͲǤͷ݀ (1)

where z = distance from the critical section to the point of contraflexure, d = effective
depth of the beam, the plastic hinge was found to be 50 mm.

Figure 2b. Reinforcement details of M2 and M3

3 Analytical Studies Performed on Normal and Link Column

Frame
3.1 3D Finite element PRGHO¶V development
For the finite element analysis of the LCF system, ANSYS software was used. For the
concrete modelling, Solid65 the eight-node element was used. This solid65 element
will have three degrees of freedom at each of their eight nodes i.e. translation in all the
three directions in the nodes and it is having capability to model creep, plastic
deformation, crushing and cracking. For the reinforcement modelling, Link8 element
will be used i.e. both the transverse and longitudinal reinforcement in the models. This
Link8 element will have three degrees of freedom at each of their nodes i.e. translation
in all the three directions in the nodes. Link element will act as a truss element i.e.
subjected to both compression and tension. From the experimental material properties
test results the values were assigned in the analysis i.e stress strain values. In the
specimen, the base of the column in the frame was fixed and at the top of the beam
column joint the Quasi- Static cyclic loading was applied. The medial axis theorem
was used for the structural meshing. The cyclic loading of 15 kN was given to all the
specimens in the finite element analysis. In order to consistent with the experimental
254 J. J. Shelton et al.

results the load-controlled type of loading was given instead of displacement-controlled

type. Figure 3,4 shows the solid and link element assigned in ANSYS and loading
configuration

Fig. 3. Solid 65 and Link 8 element

Fig. 4. Loading configuration

3.2 Analysis of Finite element models
For these scaled-down analytical models, under the cyclic load, the load-deflection
behavior and lateral stiffness results were compared with the experimental results. Also,
the stress distribution for all the models from ANSYS are discussed in following
sections.
Stress Distribution and Validation of the Results
From the comparison of stress distribution in all the specimens, the link column frame
with hinge connection between the main beam and link column shows better connection
in terms of transferring stresses to the main RC frame. From the analysis of the frame
under cyclic loads, it was found that by increasing the load, stress intensity is increased,
and the maximum stress is indicated by red colour. Based on the Finite element results
presented in this paper, it is concluded that the link column frame which is having hinge
connection can successfully be applied in order to study the behaviour of a link column
frame under load -controlled in-plane loading. In this specimen, after the failure of the
link element only, the plastic hinges begin to start in the moment frame. For the normal
and the hinged link column frame; initial lateral stiffnesses are 0.47 kN/mm and 2.73
kN/mm. The considerable decrease in storey drift shown in specimen linked column
with hinged connection can prevent damage to moment frame of the building in
moderate earthquake events. Because of the flexible connection in these specimens, it
FE modelling of RC frames with Link Column Frame … 255

will dissipate more energy than the other models. Fig.5 shows the stress distribution
and the comparison of load deflection graph of all the specimens with experimental.

Fig. 5a. Stress distribution and load deflection graph of Normal frame

Fig. 5b. Stress distribution and load deflection graph of Rigid Linked Column frame

Fig. 5c. Stress distribution and load deflection graph of Hinged Linked Column frame
256 J. J. Shelton et al.

4 Conclusions
This paper presents numerical research on the link column RC frames. A total of 3
specimens with different connections of link column frames and normal frames are
tested, and the seismic behaviour is intensively evaluated. The major conclusions can
be drawn as follows:

By comparing with the reference specimen i.e. normal frame, there is increase in
rigidity and strength was shown by all the link column frame specimens. In terms of
post hardening behaviour and strength, the hinged link column frame which was
designed as per IS 12303-1987 shows an improving performance when compared with
other specimens.

From the finite element analysis in ANSYS, the lateral load carrying capacity of the
link column frame was analytically studied. The hinged link column frame as per IS
12303-1987 shows an improved performance when compared with the normal frame.
The study shows that installing hinged connection in the link column frame crucial in
improving the structural performance.
The shear link in the linked column is more reliable and effective in dissipating a large
amount of energy. Results from the cyclic loading show that the cracks initially occur
at the link beam. After the yielding of link beam only the cracks are begins to occur at
the main frame. For the new earthquake resistant construction, the link column frame
system was used.

References
1. Park, Y. J., and Ang, A. H. S., ³0HFKDQLVWLF VHLVPLF GDPDJH PRGHO IRU UHLQIRUFHG
FRQFUHWH´ Journal of structural engineering, 111(4), 722-739 (1985).
2. *KREDUDK $ DQG (OIDWK + $³5HKDELOLWDWLRQ RI D UHLQIRUFHG FRQFUHWH IUDPH XVLQJ
HFFHQWULFVWHHOEUDFLQJ´ Engineering structures, 23(7), 745-755 (2001).
3. 050DKHUL 30HPDU]DGHK ³1RQOLQHDU DQDO\VLV RI VWHHO EUDFHG FRQFUHWH IUDPHV´ LQ
proceedings of 1st international conference on concrete and development, Tehran, Iran,
(2001)
4. Dusicka, P., and Iwai, R. ³'HYHORSPHQWRIOLQNHGFROXPQIUDPHV\VWHPIRUVHLVPLFODWHUDO
ORDGV´,Q Structural Engineering Research Frontiers, pp. 1-13 (2007).
5. Shelton, J.J. and +HPDODWKD*³%HKDYLRURI/LQNHG-Column System subjected to Seismic
)RUFH´,QGLDQ-RXUQDORI6FLHQFHDQG7HFKQRORJ\
6. Uniform Building Code ³8%&-,Q ,QWHUQDWLRQDO &RQIHUHQFH RI %XLOGLQJ 2IILFLDOV´
Whittier, California (1997).
7. Lopes, A., Dusicka, P., and Berman, J., ³Lateral Stiffness Approximation of Linked Column
6WHHO)UDPH6\VWHP´,Q Structures Congress 2015, 2408-2420 (2015).
8. %,6 ³,6  ²Indian Standard Criteria for Design of RCC Hinges: [CED 2:
&HPHQWDQG&RQFUHWH@´%XUHau of Indian Standards, New Delhi (1987).
9. 0DWWRFN$+³'LVFXVVLRQRI5RWDWLRQDOFDSDFLW\RIUHLQIRUFHGFRQFUHWHEHDPVE\:*
Corley. Journal of the Structural Division, 93(2), 519-522 (1967).
 MECHANICAL BEHAVIOUR OF FOUR-LEG BASE
CONFIGURATION ON COLD-FORMED STEEL
LATTICE COLUMN

Mohd Syahrul Hisyam Mohd Sani1 and Fadhluhartini Muftah2

1,2
Faculty of Civil Engineering, Universiti Teknologi Mara Cawangan Pahang, 26400 Bandar
Jengka, Pahang, Malaysia
1
msyahrul210@uitm.edu.my, 2fadhlu@uitm.edu.my

Abstract. Steel lattice structure has often been used as column or girder, which
is connected to each other by bolting, screwing or welding. Steel lattice is
normally made using hot-rolled steel (HRS) and rarely used cold-formed steel
(CFS). CFS is becoming popular in the building and construction sector due to
less maintenance cost and ease of fabrication. The stability of the steel lattice
structure either by using HRS or CFS is necessary to be studied and
investigated, which is by checking their mechanical behaviour for the overall
structure and focusing on the base configuration. The CFS lattice structure,
member section, configuration, connection or fastener, end support and span
must be checked to determine the mechanical behaviour. However, the failure
of CFS, such as buckling, must be revised to ensure that the CFS lattice column
is stable. The effect of CFS lattice structure mainly fails or become unstable due
to the slender section of the structure, especially on the compression member.
The slender section of the structure must be intentionally replaced with a short
section or get added by bracing. There are four types of four-leg base
configuration that have been tested. From the testing, the B.CFS 4 specimen
showed the highest value of ultimate load among the specimens.

Keyword: Mechanical Behaviour, Lattice, Cold-formed Steel, Buckling

1. Introduction
Steel lattice normally exists as 2-dimensional or 3-dimensional steel structure, which
is taken from the truss structure principle. Steel lattice is utilised as a beam or column
in buildings, advertisement boards, and performance stages. Axisa et al. (2017) have
reported that steel lattice utilised as a tower is generally seen at telecommunication
and electricity transmission tower, whereas minor sources are found in the wind
energy field. The steel lattice structure is categorised as a strong and durable structure
due to the main section being joined with steel bracing in three axes. Steel lattice is
recognised as an efficient and effective structure for carrying certain load, especially
static and dynamic loads, which emerges from self-weight and environmental
impacts. Furthermore, Zhan et al. (2015) have reported that the steel lattice is
acknowledged as a versatile structure due to its ability to generate harmful chemical
exposure or environmental problems, such as seawater, industrial effluents and

© Springer Nature Singapore Pte Ltd. 2020 257

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_38
258 M. S. H. M. Sani and F. Muftah

ground water. There are a lot of advantages of steel lattice structure, which can be
divided into several factors, such as design criteria, production cost, fabrication,
transportation, erection, and environmental-friendliness. The main section in steel
lattice is bracing, either in diagonal or horizontal section and the main column.
Bracing or sometimes recognised as lacing section, while main column or known also
as the chord or limb.
There are some failure issues that must be checked for their connection behaviour
that create problems, such as rotational stiffness, initial deformation of the section,
bolt connection slippage, and eccentricities at the connection (Langlois et al., 2016).
Ann et al. (2019) have investigated the connection failure and found that failure
always exist on the main column of the steel lattice on the tensile part, which shows
that the bolt is cut-off due to wind action. Normally, full loading on the high steel
lattice structure establishes the specific deformation shapes and failure mode locations
(Ramalingam, 2017). The deformation and failure locations are dependent on the
section properties and connections, which concentrated on safety, security and
reliability requirements by ensuring the strength limit state of specific sections and
connections (Ramalingam, 2017). There are several configurations used in lattice
structure, such as z-configuration with z-shaped arrangement of bracing and attached
to the main column. The z-shaped configuration is popular for steel composite
structure (Yang et al., 2019). Yang et al. (2018) reported that the three-leg CFST
lattice section is stiffer, more stable, and economical compared to the four-leg CFST
lattice section, and thus appropriately utilised as the corner or edge column for a
building. Dar et al. (2018) have studied the utilisation of four equal angle CFS section
as the main column and joined with single steel plate bracing, which is situated at an
inclined angle of 450 on the longitudinal axes. The main column made of a hollow
section is recommended to have high strength-to-weight ratio, strength in all axis, and
buckling resistance when compared to the open section (Luo et al., 2019). The
connection used for joining between the diagonal bracing section and the main section
is realised by using bolt, nuts, and gusset plates. Lee and McClure (2007) have
discussed about the normal steel lattice by using angle and connector of bolts.
Cold-formed steel (CFS) is used as the structural element in buildings refers to its
advantages, such as higher strength-to-weight ratio, fast production and ease in
transportation. Zeynalian et al. (2018) have mentioned the advantages of CFS, such as
lightweight, high quality, and ease of construction, which have shifted the use of
conventional material to CFS section in buildings or houses. They have studied the
lateral characteristics of lattice CFS structure by establishing five full-scale specimens
under static and dynamic loadings. They have used the angle section for the main
column and bracing, and also utilised welding and bolting for the end support and
section connector. Meanwhile, Dar et al. (2018) have studied the structural
performance of built-up CFS column, which is appropriately accepted as CFS lattice
column under axial compression load. It has a variety of variables, such as width-to-
thickness ratio, slenderness ratio between main column and bracing section, and
height of column. The local buckling and local-flexural buckling are both noted
currently and thus analysed, showing that failure is started by flexural buckling on the
main column when the specimen classified has lower width-to-thickness ratio (Dar et
al., 2018). Xu et al. (2014) have reported the result of CFS concrete-filled tube
column with lengths ranging from 500 mm to 3500 mm. They have mentioned that
Mechanical behaviour of Four-Leg base Configuration … 259

the lack of knowledge regarding the behaviour and strength of the composite main
column section utilised in steel lattice structure.
Reviews and observations show that the CFS lattice structure is either performing
as the column or beam, with limited mechanical behaviour information. Besides, the
design of the CFS lattice structure refers to the design of trusses structure and did not
have any appropriate design. But the individual section of either the bracing or main
column is required to design the column and beam. The experimental process and
method of CFS lattice structure are inadequate and still heavily researched especially
its support and loading condition. From the observation, much research has shown
that the bracing utilises the angle section, while the main column uses the angle or
hollow section. However, there is no information on discussing or fabricating the
lattice structure using CFS channel section. The failure of local buckling is solved if
the CFS channel section with web stiffener is used as the bracing and box-up CFS
section as the main column. The main objective of this study is to determine the
mechanical behaviour of the base configuration for CFS lattice column. It will also
determine the best configuration that promotes the highest value of ultimate load and
less of buckling failure.

2. Material and Method

The CFS channel section with double web stiffener was selected and its section
properties are shown in Table 1. The CFS channel section was cut into 250 mm for
the main column and horizontal bracing for every configuration. There were four
types of base configuration for the testing, whereby all configurations were of four-
leg main column. There were two main column configuration samples with four
single channel sections, as well as two other samples with four built-up CFS main
columns. All specimens were fastened by using self-drilling screw and spot welding.

Table 1. The section properties of CFS channel section

Properties and Symbols Value Properties and Symbols Value
Web, D 75 mm Area, A 148 mm2
Flange, F 34 mm Strength, fy 550 MPa
Lipped, L 8 mm Moment of Inertia, Ixx 0.135 x 106 mm4
Thickness, t 1 mm Moment of Inertia, Iyy 0.025 x 106 mm4

Specimen labels and information are tabulated in Table 2. All specimens were
located on the steel frame structure and attached with load cell and hydraulic jack at
the bottom side. The load pace of the testing was reported at 1mm/min for
determining the mechanical behaviour of all specimens, especially the compressive
strength behaviour. Then, the failure mode of all specimens was observed and the
result of testing was analysed and compared with each other. From the result, the
four-leg base configuration with reasonable value of ultimate load and less failure was
chosen for further study. Lastly, the mechanical behaviour of the connection test was
completed to determine the relationship of the self-drilling screw in the pull-out test
and shear test, which concentrated on the two and four self-drilling screws,
respectively. The result of the connection test was then analysed.
260 M. S. H. M. Sani and F. Muftah

Table 2. The specimen labels and basic information

Dimension
Specimen Description
& Mass
S.CFS 1 Single four-leg base configuration with total single CFS channel section 320 mm x
of 12 numbers. All bracing is placed in the horizontal direction on all 320 mm x
sides. All bracing are located outside of the main column. One side of the 250 mm &
bracing is fastened by two self-drilling screws to flange element and 3.75 kg
another side of the bracing is joined by four self-drilling screw to web
element as shown in Figure 1 (a).
S.CFS 2 Single four-leg base configuration with total single CFS channel section 320 mm x
of 12 numbers. Two side directions of bracing are placed inside by 250 mm x
joining on the web element of the main column, while another two side 250 mm &
directions are located outside by joining on the flange element of the 3.75 kg
main column. Two and four self-drilling screws are situated on the flange
and web, respectively. The four-leg base configuration is illustrated in
Figure 1 (b).
B.CFS 3 Eight single CFS channel sections are used as the bracing and four built- 320 mm x
up CFS channel section are fabricated by using the spot weld on the top 320 mm x
and bottom side, forming a hollow section. So, 16 single CFS channel 250 mm &
sections are used. All bracing are situated outside of the main column 5.00 kg
with four self-drilling screws. The specimen is shown in Figure 1 (c).
B.CFS 4 A total of 16 single CFS channel sections are utilised for this specimen. 250 mm x
Two sides of bracing are situated inside of the main column and another 320 mm x
two sides of bracing are located outside of main column. All bracing are 250 mm &
connected to the main column by self-drilling screw horizontally, as 5.00 kg
shown in Figure 1 (d).

(a) (b) (c) (d)

Fig. 1. Top view of (a) S.CFS 1, (b) S.CFS 2, (c) B.CFS 3 and (d) B.CFS 4 specimen

3. Result and Discussion

From the compressive strength testing, the result of the ultimate load for the four
specimens is tabulated in Table 3. The table also states the failure mode of all
specimens. The difference of percentage for ultimate load of S.CFS 1 and S.CFS 2
were noted as 9.01%, while B.CFS 3 and B.CFS 4 were recorded with 6.48%. From
the result, the inside bracing was an important solution to minimise the failure of
buckling on the main column, either in single or built-up section. Furthermore, the
built-up CFS channel section on the main column played the role of solving the
buckling failure, rather than by using the single CFS channel section as the main
column. This could be proved as the ultimate load was increased about 42.30% and
40.70% when the single CFS channel section on main column shifted to built-up CFS
channel section in similar configuration. The built-up CFS channel section with
symmetrical section was able to protect the specimen from twisting or buckling. The
Mechanical behaviour of Four-Leg base Configuration … 261

percentage difference of ultimate load between the lowest value of S.CFS 1 against
the highest value of S.CFS 4 was 46.04%. From the testing, the appropriate four-leg
base configuration with the highest value of ultimate load and less failure was S.CFS
2 for single CFS channel section and B.CFS 4 for built-up CFS channel section.

Table 3. The result of ultimate load and failure mode of the specimen
Ultimate
Specimen Failure Mode
Load (kN)
The bottom bracing failed due to local buckling at the middle part between
S.CFS 1 108.82 the two main columns. The main column was observed to have local
buckling at the middle span.
The bottom outside bracing was noted to have local buckling and not
S.CFS 2 119.60 showing any buckling failure at the bottom and top inside bracing. The
main column was reported to have the local buckling at the middle span.
No local buckling occurred on the main column and top bracing, but the
B.CFS 3 188.60
failure of yielding existed at the middle span of bottom bracing.
No local buckling occurred on the main column supported with bracing.
However, a small local buckling failure existed on the main column
B.CFS 4 201.67
unattached with bracing. The bracing at either top or bottom and situated
on the middle structure established that the specimen became stable.

The result of the connection test for the two and four self-drilling screws used as
fastener is tabulated in Table 4. Both specimens were reported to fail due to the tilting
of the screw. The percentage difference between two and four self-drilling screws for
shear and pull-out connection test was recorded as 46.08% and 76.96%, respectively.
The four self-drilling screw was thus appropriate to be utilised on the large cross-
section, especially on the web element. Meanwhile, the two self-drilling screws were
suitable to be located on the flange element having small cross-section. Thus, the two
or four self-drilling screws were suitable to be used in joining the main column with
bracing in which the load distribution was not so large. The load was transferred to
each main column and bracing to provide the stability and safety requirement.

Table 4. The result of shear connection test for two and four self-drilling screw
Ultimate load (kN)
Specimen
Shear Connection Test Pull-out Connection Test
Two Self-drilling Screw 13.05 2.26
Four Self-drilling Screw 24.20 9.81

4. Conclusion
The paper summarised the recent research on steel lattice of either hot-rolled steel or
cold-formed steel for column and beam application. From the observation and
analysis, there is still limited knowledge that needs to be addressed, such as procedure
of mechanical behaviour testing like compression, main column and bracing
configuration, and design flowchart of CFS lattice column. Besides, the ultimate load
percentage difference of the single and built-up CFS channel section as the main
column was reported to be in the range of 45% to 75%. Then, the ultimate load of
single and built-up CFS channel section as main column were increased
262 M. S. H. M. Sani and F. Muftah

approximately by 40% when the outside bracing was changed location to the inside
bracing. Finally, further study must be done to calculate the CFS lattice structure,
either beam or column, which used the other shapes of configuration or cross-section
of bracing, or the main column section.

Acknowledgements
The authors gratefully acknowledge the financial support from the Ministry of Higher
Education, Malaysia under the Fundamental Research Grant Scheme (FRGS) for Universiti
Teknologi Mara (UiTM) Pahang. Heartfelt thanks are also extended to the Faculty of Civil
Engineering of UiTM Pahang, Jengka Campus for providing the laboratory equipment.

Reference
1. Axisa, R., Muscat, M., Sant, T. and Farrugia, R.N.: Structural assessment of a lattice tower
for a small, multi-bladed wind turbine. International Journal Energy Environmental
Engineering 8, 343-358 (2017).
2. Zhan, Y., Wu, Gang and Yang, L-S.: Experimental investigation of the behaviour of a
lattice steel column repaired with pultruded GFRP profiles. Journal of Performance of
Constructed Facilities 29 (4), 04014094-1-13 (2015).
3. /DQJORLV 6 3UXG¶KRPPH 6 Legeron, F. and Pourshargh, F.: Review of Advanced
Modelling Methods for Lattice Steel Towers. 2016 CIGRE-IEC Colloquium, May 9-11,
Montreal, Canada (2016).
4. Ann, L., Wu, J. and Jiang, W.: Experimental and numerical study of the axial stiffness of
bolted joints in steel lattice transmission tower legs. Engineering Structures 187, 490-503
(2019).
5. Ramalingam, R.: Failure analysis of lattice tower like structures. IOP Conference Series:
Earth and Environmental Science 80, 1-8 (2017).
6. Yang, Y-F., Liu, M., Hou, C. and Bie, X-M.: Behaviour of four-legged square CFST
latticed members under lateral cyclic loading. Journal of Constructional Steel Research
156, 54-74 (2019).
7. Yang, Y-F., Liu, M. and Fu, F.: Experimental and numerical investigation on the
performance of three-legged CFST latticed columns under lateral cyclic loadings. Thin-
walled Structures 132, 176-194 (2018).
8. Dar, M.A., Sahoo, D.R., Pulikkal, S. and Jain, A.K.: Behaviour of laced built-up cold-
formed steel columns: Experimental investigation and numerical validation. Thin-walled
Structures 132, 398-409 (2018).
9. Luo, F.J., Huang, Y., He, X., Qi, Y. and Bai, Y.: Development of latticed structures with
bolted steel sleeve and plate connection and hollow section GFRP members. Thin-walled
Structures 137, 106-116 (2019).
10. Lee, P-S. and McClure, G.: Elastoplastic Large deformation analysis of a lattice steel
tower structure and comparison with full-scale tests. Journal of Constructional Steel
Research 63, 709-717 (2007).
11. Zeynalian, M., Bolkhari, S. and Rafeei, P.: Structural performance of cold-formed steel
trusses used in electric power substations. Journal of Constructional Steel Research 147,
53-61 (2018).
12. Xu, F., Chen, J. and Jin, W-L.: Experimental investigation of thin-walled concrete-filled
steel tube columns with reinforced lattice angle. Thin-walled Structures 84, 59-67 (2014).
 Finite Element Simulation of Member Buckling of Cold-
rolled Aluminium Alloy 5052 Channel Columns

Ngoc Hieu PHAM1, Cao Hung PHAM1 and Kim J.R. RASMUSSEN1
1 The University of Sydney, New South Wales, Australia

ngochieu.pham@sydney.edu.au;caohung.pham@sydney.edu.au;
kim.rasmussen@sydney.edu.au

Abstract. This paper presents the numerical investigation of cold-rolled channel

columns made of 5052 aluminium alloy subjected to member buckling. A de-
tailed Finite Element (FE) model using software ABAQUS v 6.14 was developed
to simulate an experimental program recently performed at the University of Syd-
ney. In the FE simulations, actual measured properties and geometric imperfec-
tions were incorporated into the FE models. The FE results are compared and
validated against the experimental results. These accurate and reliable FE models
will be used for the future parametric study to extend the data range for the de-
velopment of the design guidelines for the cold-rolled aluminium alloy channel
columns subjected to member buckling.

Keywords: Cold-rolled Aluminium Alloy, Numerical Study, Member Buck-

ling, Channel Columns.

1 Introduction

Aluminium alloy members have been increasingly used in structural applications. The
conventional method to produce aluminium alloy members is extrusion. In recent years,
BlueScope Permalite [1] in Australia has used the existing rolling system for cold-
formed steel sections to successfully roll-form aluminium sections. The roll-forming
process seems to be more cost effective and faster than the extrusion process. Therefore,
the cold-rolled aluminium alloy sections have been promoted to be widely used due to
their competitive advantages. In numerical simulation, Finite Element (FE) software
packages have been commonly used for structural analysis. In the past, Zhu and Young
[2, 3] provided complete details of the FE models to investigate the buckling modes of
extruded tubular and circular hollow section columns. To date, accurate FE simulations
especially for cold-rolled aluminium alloy channel columns have not been reported.
This paper presents the detailed FE models using finite element software ABAQUS
6.14 [4] to simulate cold-rolled aluminium alloy 5052 channel columns undergoing
member buckling. The experimental data published in a separate paper by the authors
[5] are used for validation of the finite element models. The accurate FE results and
reliable models will be consequently used to perform parametric studies to extend the
data range on the basis of commercially available sections from the catalogue “Roll-
Formed Aluminium Purlins Solutions” [1]. All test and numerical data of the ultimate

© Springer Nature Singapore Pte Ltd. 2020 263

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_39
264 N. H. Pham et al.

strengths of the cold-rolled aluminium columns subject to member buckling will be

used for calibration and proposal of new design equations in an ongoing research pro-
ject at the University of Sydney.

2 Finite Element Models

2.1 Material Properties And Element Type

Material properties including the full stress-strain curves are obtained from coupon test
results reported in Huynh et al. [6]. Based on the measured stress and strain from flat
and corner coupon tests, true stress ıt and true plastic strain İtp used in ABAQUS were
calculated as follows:

Vt V (1  H ) (1)

Vt
H tp ln(1  H )  (2)
E
where ı and İ are the measured stress and strain, respectively, obtained from the tensile
coupon tests [6].
Four-node reduced integration shell element, type S4R, with three translational and
three rotational degrees of freedom at each node, was used to model the aluminium
cold-rolled channel sections. 3D-deformable solid element, type C3D8R, was used to
simulate the other test rig parts at two ends in FE model. A mesh convergence analysis
conducted in Huynh et al. [7] demonstrated that the mesh size of 10×10 mm for sections
is adequate for further modelling of cold-rolled channel sections under compression
and was therefore used in this study.
2.2 Boundary Conditions
Finite element model was illustrated in Fig. 1. Two configurations of end boundary
conditions were designed. In the first configuration, the end cross-sections were re-
strained from rotation about the minor axis, and freely rotated about the major axis.
Conversely in the second configuration, free rotation and prevented rotation conditions
were used for the major and minor axes, respectively (see Fig. 2)
2.3 Initial Geometric Imperfections
Actual initial imperfections were measured using the laser scanning method fully pre-
sented in Pham et al. [5], and were used for generating imperfections in the FE models.
The procedure to introduce the measured imperfections into FE models was reported in
Pham et al. [8]. An example is shown in Fig. 3 including local imperfections d1 of the
web, distortional imperfections d2 of the flanges, and initial twist G3 at two ends.
Finite Element Simulation of Member Buckling … 265

(a) Test rig (b) Finite element model

Fig. 1. Test rig and finite element model

(a) Configuration 1 (b) Configuration 2

Fig. 2. End boundary conditions

Fig. 3. Actual imperfect specimen

3 Model Validation

3.1 Ultimate capacities

Ultimate loads from FE models were summarized in Table 1, and compared with the
experimental results. Good agreements between modelling and test results have been
achieved, with means of 1.014 and 1.009 and standard deviations of 0.017 and 0.012
for the first and second configurations respectively.
266 N. H. Pham et al.

Table 1. Comparison between test and FE results

Peak load (kN)
 Specimens PFE/Ptest
Ptest PFE
C10030-2.0m-1 91.25 91.83 1.006
C10030-2.0m-1e 81.00 82.00 1.012
C10030-2.0m-2e 81.38 81.59 1.003
C10030-2.5m-1 70.64 70.84 1.003
C10030-2.5m-2e 60.10 61.02 1.015
C10030-2.5m-3e 61.60 61.08 0.992
Configuration 1

C10030-3.0m-2 49.45 49.78 1.007

C10030-3.0m-1e 46.76 46.24 0.989
C10030-3.0m-3e 44.65 45.10 1.01
C25025-3.0m-3 88.20 89.30 1.012
C25025-3.0m-1e 89.30 91.70 1.027
C25025-5.0m-2e 55.23 57.15 1.035
C25025-5.0m-3 56.25 58.67 1.043
C25025-7.0m-1 34.55 35.18 1.018
C25025-7.0m-2e 33.15 34.60 1.044
Mean 1.014
Standard deviation 0.017
C25025-2.0m-1 60.75 61.31 1.009
C25025-2.0m-2e 59.40 59.97 1.01
C25025-3.0m-1 36.95 37.04 1.002
C25025-3.0m-2 34.65 35.97 1.038
C40030-7.0m-1 37.45 38.30 1.023
Configuration 2

C40030-7.0m-2e 36.55 37.10 1.015

C40030-7.0m-3e 36.10 36.19 1.002
C40030-5.0m-1 60.35 61.13 1.013
C40030-5.0m-2e 57.60 57.30 0.995
C40030-5.0m-3e 58.95 58.89 0.999
C40030-3.0m-1 104.25 105.50 1.012
C40030-3.0m-2e 102.25 102.60 1.003
C40030-3.0m-3e 102.65 102.80 1.001
Mean 1.009
Standard deviation 0.012

3.2 Deformation Behaviour

Fig. 4 shows a comparison of failure modes between FE modelling and test of C10030-
3.0m-3e. Flexural-torsional buckling modes were observed in both test and FE re-
sponse. Fig. 5(a) revealed a good agreement between modelling and experiment in
terms of load versus lateral deformation. Good agreement was obtained for global twist,
as shown in Fig. 5(b).
Finite Element Simulation of Member Buckling … 267

Fig. 4. Failure mode: C10030-3.0m-3e in the first configuration

(a) Axial load vs. lateral-deformation (b) Axial load vs. global twist
Fig. 5. Deformation behaviours: C10030-3.0m-3e

Fig. 6. Failure mode: C25025-3.0m-1 in the second configuration

(a) Axial load vs. lateral-deformation (b) Axial load vs. local deformations
Fig. 7. Deformation behaviours: C25025-3.0m-1
268 N. H. Pham et al.

Local-global interaction buckling mode was seen in the modelling and test behaviour
of C25025-3.0m-1, as illustrated in Fig. 6. Excellent agreements are also obtained for
lateral deformation and local deformations at mid-span (see Fig. 7(a) and (b)). It is
concluded that FE models accurately simulated the deformations and failure patterns.

4 Conclusion
This paper presents a numerical investigation of cold-rolled aluminium alloy channel
columns using the software ABAQUS V.6.14 [4] to capture the global and local-global
interaction buckling modes. The models were validated against the tests performed in
a separate experimental program [5]. The reliable FE models will be subsequently used
to develop parametric studies to extend larger ultimate strength data with a wider range
of slenderness for further research on the development of design guidelines for these
types of cold-rolled aluminium channel columns.
Acknowledgement
Funding provided by the Australian Research Council Linkage Research Grant
LP140100563 between BlueScope Lysaght and the University of Sydney has been used
to perform this research. The authors would like to thank Permalite Aluminum Building
Solutions Pty Ltd for the supply of the test specimens and financial support for the
project. The first author is sponsored by the scholarship provided by Australian Awards
Scholarships (AAS) scheme from Australian Government.

References
1. Blue Scope Lysaghts: Permalite – Aluminium Roll-formed Purlin Solutions. Permalite Al-
uminium Building Solutions Pty Ltd, Eagle Farm Qld 4009, Australia (2015).
2. Zhu, J.H. and Young, B.: Aluminum alloy tubular columns- Part II: Parametric study and
design using direct strength method. Journal of Thin-Walled Structures 44(9), 969-985
(2006).
3. Zhu, J.H. and Young, B.: Numerical investigation and design of aluminum alloy circular
hollow section columns. Journal of Thin-Walled Structures 46(12), 1437-1449 (2008).
4. ABAQUS/Standard: ABAQUS/CAE User’s Manual. Dassault Systemes Simulia Corp,
Providence, RI, USA (2014).
5. Pham, N.H., Pham, C.H. and Rasmussen, K.J.R.: Experimental investigation of the member
buckling of cold-rolled aluminium alloy 5052 Channel columns. In: 9th International Con-
ference on Advances in Steel Structures, pp. 491-503, Hong Kong (2018).
6. Huynh, L.A.T., Pham C.H. and Rasmussen, K.J.R.: Mechanical properties of cold-rolled
aluminium alloy 5052 channel sections. In: 8th International Conference Steel and Alumin-
ium Structures, Hong Kong (2016).
7. Huynh, L.A.T., Pham, C.H. and Rasmussen, K.J.R.: Numerical Simulations of Cold-Rolled
Aluminium Alloy 5052 Channel Sections in Stub Column Tests. In:4th Congrès International
Géotechnique - Ouvrages –Structures, pp. 202-212, Ho Chi Minh city (2017).
8. Pham, N.H., Pham, C.H. and Rasmussen, K.J.R.: Incorporation of measured geometric im-
perfections into finite element models for cold-rolled aluminium sections. In: 4th Congres
International de Geotechnique – Ouvrages – Structures, pp. 161-171, Ho Chi Minh city
(2017).
 Lateral Buckling Tests of Cold-rolled
Aluminium Alloy 5052 Zee Beams

Ngoc Hieu PHAM1, Cao Hung PHAM1 and Kim J.R. RASMUSSEN1
1
The University of Sydney, New South Wales, Australia
ngochieu.pham@sydney.edu.au;caohung.pham@sydney.edu.au;
kim.rasmussen@sydney.edu.au

Abstract. This paper presents a series of four-point bending tests of cold-rolled

aluminium alloy 5052 Zee beams subjected to lateral buckling. Six Zee beams
with three different lengths were selected for testing. A dual-actuator loading
system was specifically designed to keep the load always in vertical direction and
through the shear centre of the cross sections during testing. Lateral buckling
failure modes were observed in long beams, while local-lateral interaction buck-
ling was seen in short ones. The test results in this study will be used to calibrate
the new proposed design guidelines for cold-rolled sections subjected to lateral
buckling in an ongoing research project at the University of Sydney.

Keywords: Cold-rolled Aluminium Alloy, Lateral Buckling, Zee Beams.

1 Introduction

The use of aluminium alloys in structural applications has been significantly increased
in recent years due to their advantageous properties [1] and the ability to resist the cor-
rosion of harsh environments. Extrusion has been the conventional method to produce
aluminium alloy sections. Until recently, a new method of roll-forming has been intro-
duced by BlueScope Permalite in Australia [2] to successfully roll-form aluminium
Channel and Zee sections. The latter method is found to be more cost effective and
faster than the former one. This new method has attracted the attention in the worldwide
due to its advantages.
Recent experiments to investigate the buckling modes of extruded square and rec-
tangular hollow aluminium beams have been conducted by Su et al. [3], while experi-
mental data on lateral buckling of aluminium Zee beams remain limited. This paper
presents an experimental program on cold-rolled aluminum alloy Zee-beams subjected
to lateral buckling. Experimental results will be used for the development of new pro-
posed design guidelines for cold-rolled aluminium sections subjected to lateral buck-
ling in an ongoing research project at the University of Sydney.

© Springer Nature Singapore Pte Ltd. 2020 269

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_40
270 N. H. Pham et al.

2 Material properties

Flat and corner coupons were conducted to determine the mechanical properties, and
were reported in Huynh et al [4]. Material properties for both flat and corner parts are
summarized in Table 1.

Table 1. Tensile material properties

Coupon Position ı0.2 (MPa) ıu (MPa) E0 (GPa) İf n

C10030 flat 224 277 69.3 5.6 12.4
C25025 flat 218 267 69.7 9.0 12.1
C40030 flat 222 271 69.6 10.0 11.9
C10030 corner 248 290 70.4 5.2 12.5
C25025 corner 243 291 70.5 6.4 12.6
C40030 corner 240 279 70.5 5.3 12.5
Note: ı0.2 is the 0.2% proof strength; ıu is the ultimate tensile strength; E0 is the Young’s modu-
lus; İf is the uniform elongation corresponding the total elongation after fracture; and ‘n’ is the
Ramberg-Osgood index.

3 Specimen Dimensions and Geometric Imperfections

Cross-section geometries were chosen from the “Roll-formed aluminium Purlins Solu-
tions” Catalogue [2]. The selection of sections for this investigation was based on the
buckling analyses using THIN-WALL-2 software [5] and was reported in Pham et al.
[6]. Section Z25025 was selected for experimental program with nominal dimensions
given in Table 2.

Table 2. Nominal cross-section dimensions

Section Z25025
t(mm) 2.5
D(mm) 255
E (mm) 81
F(mm) 73
L (mm) 25.5

In order to obtain lateral buckling modes, specimen lengths were selected using elas-
tic global buckling analyses from AS/NZS 4600:2005 [7], and presented in Pham et al
[8]. Test specimens are given in Table 3. A typical specimen Z25025-3m-1 is defined
as follows: (i) “Z25025” indicates a Zee section of 255mm depth and 2.5mm thickness,
(ii) “3.0m” is the specimen length, and (iii) “1” indicates the test number 1.
Lateral Buckling Tests of Cold-rolled Aluminium … 271

Table 3. Test specimens

Cross-section Specimen labels

Z25025-3.0m-1; Z25025-3.0m-2
Z25025 Z25025-4.0m-1; Z25025-4.0m-2
Z25025-5.0m-1; Z25025-5.0m-2

Geometric imperfections were measured using laser scanning method with nine laser
points around the cross-section (see Fig. 1). The laser scanner devices ran along the
specimen length with a constant speed to record the distance to surface specimens. The
details of laser scanner method were presented in Pham et al [9].

Fig. 1. Laser scanner method for measuring geometric imperfections

4 Test Set-Up

Cold-rolled aluminium alloy 5052 Zee beams were tested in four-point bending con-
figuration. The basic design of test rig was developed by Niu and Rasmussen [10] for
Channel sections. The mechanism of the test rig was presented in Pham et al [8]. The
test rig for Channel sections was then modified to convert into the test rig for Zee sec-
tions (see Fig. 2). In Z25025 sections, the inclination of principal axis and rectangular
is 15.1o. The web was therefore rotated 15.1o by using a pair of right triangle hollow
blocks at end sections as seen in Fig. 3. The arms of each loading frame attached to the
web were modified by an inclination 15.1o (see Fig. 2).

Fig. 2. Test rig for Zee sections

272 N. H. Pham et al.

Fig. 3. Connection between connecting plates and web of Z25025 section

Transducers were attached at loading points, two end sections and mid-span to meas-
ure the local and global displacements (Fig. 2). Transducers at the ends of specimen to
monitor end-shortening. Transducers at the aluminium frame at mid-span to measure
local deformations.

5 Specimen behaviours and test results

Lateral buckling modes were observed in long beam tests. There are no local or distor-
tional deformations prior to peak load. The lateral buckling modes were captured as
shown in Fig. 4. Fig. 5 shows the lateral deformation and global twist at load point of
Z25025-4.0m-2.
Twist rotations occurred in two directions including positive (clock wise) and nega-
tive directions (anti-clock wise) (see Fig. 6). The strength in positive direction was also
higher than that in negative direction as discussed in Pham et al [8]. Failure process in
the positive direction occurred gradually with load rate of shedding, while that process
in the negative direction was almost snap-through soon after the peak load (Fig. 6). The
ultimate capacity of all specimens are given in Table 4.

Table 4. Ultimate strength of all tests

Specimens Peak load (kN)
Z25025-3.0m-1 11.926*
Z25025-3.0m-2 11.598*
Z25025-4.0m-1 6.444
Z25025-4.0m-2 6.829*
Z25025-5.0m-1 4.385
Z25025-5.0m-2 4.287
(*): The tests failed in positive direction (clockwise)
Lateral Buckling Tests of Cold-rolled Aluminium … 273

Fig. 4. Lateral buckling mode of specimen Z25025-4.0m-2

Fig. 5. The deformation behaviours of specimen Z25025-4.0m-2

Fig. 6. Effect of direction of twist rotation on strength

274 N. H. Pham et al.

6 Conclusion

The paper presents a series of bending test to investigate the ultimate strengths of cold-
rolled aluminium alloy 5052 Zee beams subjected to lateral buckling. Test rig for Chan-
nel sections presented in Pham et al [8] was modified to convert into the test rig for Zee
sections. Lateral buckling modes were observed in long beam tests. The test results will
be used for the development of new proposed design guidelines for cold-rolled alumin-
ium alloy sections subject to lateral buckling. Research is ongoing at the University of
Sydney.

Acknowledgement

Funding provided by the Australian Research Council Linkage Research Grant

LP140100563 between BlueScope Lysaght and the University of Sydney has been used
to perform this research. The authors would like to thank Permalite Aluminum Building
Solutions Pty Ltd for the supply of the test specimens and financial support for the
project. The first author is sponsored by the scholarship provided by Australian Awards
Scholarships (AAS) scheme from Australian Government.

References
1. Szumigala, M. and Polus, L.: Application of Aluminium and Concrete Composite Struc-
tures. Precedia Engineering 108(61), 544-549 (2015).
2. Blue Scope Lysaghts: Permalite – Aluminium Roll-formed Purlin Solutions. Permalite Al-
uminium Building Solutions Pty Ltd, Eagle Farm Qld 4009, Australia (2015).
3. Su, M.N., Young, B. and Gardner, L.: Deformation-based design of aluminium alloy beams.
Engineering Structures 80, 339-349 (2014).
4. Huynh, L.A.T., Pham C.H. and Rasmussen, K.J.R.: Mechanical properties of cold-rolled
aluminium alloy 5052 channel sections. In: 8th International Conference Steel and Alumin-
ium Structures, Hong Kong (2016).
5. Nguyen, V.V., Hancock, G.J. and Pham, C.H.: Development of the Thin-Wall-2 Program
for buckling analysis of thin-walled sections under generalized loading. In: 8th International
Conference on Advances in Steel Structures, Lisbon (2015).
6. Pham, N.H., Pham, C.H. and Rasmussen, K.J.R.: Experimental investigation of the member
buckling of cold-rolled aluminium alloy 5052 Channel columns. In: 9th International Con-
ference on Advances in Steel Structures, pp. 491-503, Hong Kong (2018).
7. AS/NZS 4600:2005: Cold-formed Steel Structures. Standards Australia, Sydney (2005).
8. Pham, N.H., Pham, C.H. and Rasmussen, K.J.R.: Member buckling tests on cold-rolled al-
uminium alloy Channel beams. In: 9th International Conference on Steel and Aluminium
Structures, UK (2019).
9. Pham, N.H., Pham, C.H. and Rasmussen, K.J.R.: Incorporation of measured geometric im-
perfections into finite element models for cold-rolled aluminium sections. In: 4th Congres
International de Geotechnique – Ouvrages – Structures, pp. 161-171, Ho Chi Minh city
(2017).
10. Niu, S.: Interaction buckling of cold-formed stainless steel beams. PhD thesis, the University
of Sydney (2014).
 (ODVWLF%XFNOLQJ6ROXWLRQIRU3HUIRUDWHG7KLQZDOOHG
&KDQQHO6HFWLRQVLQ6KHDUZLWKDQ$VSHFW5DWLRRI

'X\.KDQK3KDP&DR+XQJ3KDPDQG*UHJRU\-+DQFRFN
6FKRRORI&LYLO(QJLQHHULQJ7KH8QLYHUVLW\RI6\GQH\6\GQH\16:$XVWUDOLD
duykhanh.pham@sydney.edu.au; caohung.pham@sydney.edu.au;
gregory.hancock@sydney.edu.au

$EVWUDFW7KLQZDOOHGFKDQQHOPHPEHUVVXEMHFWHGWRVKHDUDUHFRPPRQO\SHU
IRUDWHGZLWKZHERSHQLQJVWRSURYLGHDFFHVVIRUEXLOGLQJVHUYLFHV\VWHPV7KH
FKDQJHV LQ WKH VWUHVV GLVWULEXWLRQ GXH WR WKH SUHVHQFH RI KROHV FDQ FDXVH WKH
FKDQJHVLQWKHFULWLFDOEXFNOLQJORDGDQGWKHRYHUDOOVWUHQJWKRIWKHSHUIRUDWHG
PHPEHUVLQVKHDU5HFHQWUHVHDUFKE\6+3KDPKDVSURYLGHGEXFNOLQJVROX
WLRQVEDVHGRQ)LQLWH(OHPHQW0HWKRGIRUGHWHUPLQLQJWKHVKHDUEXFNOLQJORDGV
RIFKDQQHOVHFWLRQVZLWKZHEKROHVZLWKVKHDUDVSHFWUDWLRXSWR+RZHYHU
WKHUHVHDUFKRQO\IRFXVHGRQVTXDUHDQGFLUFXODUKROHV:LWKWKHVDPHPHWKRG
RORJ\WKLVSDSHUSURYLGHVDEXFNOLQJVROXWLRQIRUSHUIRUDWHGFKDQQHOPHPEHUV
ZLWKHORQJDWHGZHEKROHVLQVKHDU)(PRGHOVZHUHFRQVWUXFWHGWRJHQHUDWHWKH
HODVWLFEXFNOLQJORDGVZLWKDYHU\ZLGHUDQJHRIKROHGLPHQVLRQV7KHDVVXPS
WLRQVDERXWWKHVWUHVVGLVWULEXWLRQIURPWKHSUHYLRXVVWXGLHVZHUHDOVRXWLOL]HGLQ
WKLVVWXG\$GLPHQVLRQDOWUDQVIRUPDWLRQLVDOVRSURSRVHGLQWKLVSDSHU

.H\ZRUGV7KLQZDOOVWUXFWXUHVFROGIRUPHGVWHHOVKHDUEXFNOLQJZHERSHQ
LQJVILQLWHHOHPHQWPHWKRG

 ,QWURGXFWLRQ

+LJKVWUHQJWKFROGIRUPHGVWHHOPHPEHUVLQVKHDUDUHFRPPRQO\PDQXIDFWXUHGZLWK
SHUIRUDWLRQVWRSURYLGHDFFHVVIRUEXLOGLQJV\VWHPVLQSUDFWLFH7KHVWUHVVGLVWULEXWLRQ
DQGWKHIDLOXUHPRGHVRIORDGHGPHPEHUVFDQEHFKDQJHGE\WKHLQWURGXFHRIDQRSHQ
LQJ7KLVPLJKWUHVXOWLQDUHGXFWLRQRIERWKWKHVKHDUEXFNOLQJFDSDFLW\DQGXOWLPDWH
VKHDUVWUHQJWKZKLFKLVLQIOXHQFHGE\WKHVKDSHDQGVL]HRIRSHQLQJV,QDGGLWLRQWKH
HODVWLFVKHDUEXFNOLQJORDGLVDUHTXLUHGLQSXWRIWKH'LUHFW6WUHQJWK0HWKRG'60
RIGHVLJQIRUPHPEHUVLQVKHDU7KHUHIRUHWKHHIIHFWRIFXWRXWVRQWKHVKHDUEXFNOLQJ
FKDUDFWHULVWLF LV DFULWLFDODVSHFWLQ WKHLQYHVWLJDWLRQ RIWKH EHKDYLRXU RI WKLQZDOOHG
VHFWLRQVVXEMHFWHGWRVKHDU
)RU HODVWLF EXFNOLQJ VROXWLRQV RI WKLQZDOOHG FKDQQHO PHPEHUV WKH 6SOLQH )LQLWH
6WULS0HWKRG6)60 GHYHORSHG E\/DX +DQFRFN >@ KDV EHHQ XVHG E\ 3KDP 
+DQFRFN>@7KHQHZWKHRU\RIWKH6HPL$QDO\WLFDO)LQLWH6WULS0HWKRG6$)60
XVLQJPXOWLSOHVHULHVWHUPVKDVEHHQGHYHORSHGLQ+DQFRFN 3KDP>@WRVWXG\WKH
HODVWLFEXFNOLQJFKDUDFWHULVWLFVRIFKDQQHOVHFWLRQVZLWKZHEKROHVVXEMHFWHGWRVKHDU
&+3KDP>@XVHGWKH6)60HQFRGHGLQWKHLVRSDUDPHWULFVSOLQHILQLWHVWULSPHWKRG
,6)60GHYHORSHGE\(FFKHU>@WRHVWDEOLVKVROXWLRQVIRUVKHDUEXFNOLQJORDGVIRU

© Springer Nature Singapore Pte Ltd. 2020 275

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_41
276 D. K. Pham et al.

FKDQQHOVHFWLRQVZLWKFHQWUDOZHERSHQLQJV7KH6)60ZDVDOVRXVHGE\&+3KDP
>@WRSURYLGHHODVWLFEXFNOLQJVROXWLRQVWRSHUIRUDWHGFKDQQHOVZLWKVTXDUHDQGFLUFX
ODU ZHE KROHV LQ VKHDU DQG WKLV PHWKRG ZDV DOVR FRQVLVWHQW ZLWK WKH )LQLWH (OHPHQW
0HWKRG)(0XVLQJ$EDTXV6WDQGDUG7KH)(0ZDVDOVRHPSOR\HGE\6+3KDP
>@ WR JHQHUDWH VKHDU EXFNOLQJ ORDGV DQG VKHDU EXFNOLQJ PRGHV RI FKDQQHO VHFWLRQV
ZLWKFHQWUDOVTXDUHDQGFLUFXODUKROHVLHARh WRWDOKROHOHQJWKWRWDOKROHGHSWK 
ZLWKARVSDQOHQJWKZHEGHSWKXSWR7KH)(UHVXOWVSURGXFHGIURPWKH
HODVWLFVKHDUEXFNOLQJDQDO\VHVE\6+3KDP>@ZDVDOVREHQFKPDUNHGDJDLQVW WKH
VKHDUEXFNOLQJORDGVXVLQJWKH6$6)0$EXFNOLQJVROXWLRQZLWKWKHVDPHPHWKRGRO
RJ\ )(0 DSSURDFK WR GHWHUPLQH WKH VKHDU HODVWLF EXFNOLQJ ORDGV RI SHUIRUDWHG
PHPEHUV ZLWK GLIIHUHQW KROH VKDSHV LQFOXGLQJ UHFWDQJXODU DQG VORWWHG ZHE RSHQLQJV
DUHXVHGLQWKLVSDSHUDVGHWDLOHGEHORZ

 6ROXWLRQWRHODVWLFEXFNOLQJDQDO\VLVRIFKDQQHOVHFWLRQVLQ
VKHDUZLWKZHEKROHV

 6KHDUHODVWLFEXFNOLQJDQDO\VLVXVLQJILQLWHHOHPHQWPRGHO

 
&DVH% &DVH&


%&V
)LJ%RXQGDU\FRQGLWLRQV%&VDQGVWUHVVGLVWULEXWLRQVLQ&DVH%DQG&

7KH)LQLWH(OHPHQW0HWKRG)(0PRGHOOLQJXVLQJ$EDTXVLVXVHGIRUWKHVKHDUHODV
WLFEXFNOLQJDQDO\VHVRIFKDQQHOVHFWLRQVZLWKZHEKROHV7KHQRGHGRXEO\FXUYHG
VKHOO HOHPHQW 65 LV XWLOL]HG WR PRGHO WKH ZKROH VHFWLRQ ZLWK D PHVK VL]H RI [
PPWRLQWURGXFHDQDGHTXDWHPHVKUHILQHPHQW7KHDUHDVVXUURXQGLQJWKHZHEKROHV
DUHPHVKHGZLWKWKHVZHHSPHVK3RLVVRQ¶VUDWLRμRIDQGWKH<RXQJ¶VPRGXOXV
ERI03DDUHFKRVHQDVLQWKH$61=6>@)RUVLPXODWLRQRIWKH
VLPSO\VXSSRUWERXQGDU\FRQGLWLRQVWZRHQGVHFWLRQVDUHUHVWUDLQHGLQWKH[DQG\
GLUHFWLRQV WR SUHYHQW DQ\ GHIRUPDWLRQV LQ D FURVVVHFWLRQDO SODQH ,Q RUGHU WR DYRLG
ULJLGERG\PRWLRQDORQJLWXGLQDOUHVWUDLQWLVDWWDFKHGWRDPLGGOHSRLQWRIWKHZHEDW
RQHHQGVHFWLRQ7KHPRPHQWJUDGLHQWVMJHQHUDWLQJDXQLIRUPVKHDUIRUFHVLQ
Elastic Buckling Solution for Perforated Thin-walled Channel … 277

)( PRGHOV DUH VLPXODWHG E\ WKH GLVWULEXWHG OLQH ORDGV LQ $EDTXV 7KH VKHDU HODVWLF
EXFNOLQJORDGV Vcr DQG WKH VKHDU EXFNOLQJ PRGHV DUH JHQHUDWHG WKURXJK WKH SURFH
GXUH%8&./(LQ$EDTXV6WDQGDUG
7ZRFDVHVRIVWUHVVGLVWULEXWLRQLQFOXGLQJ&DVH%DQG&DVH&PHQWLRQHGLQ&+
3KDP>@DQG6+ 3KDP>@DUH VWXGLHG WRREWDLQSUDFWLFDO YDOXHV RI Vcr $SDLURI
EHQGLQJPRPHQWV(M/2 = V.L/2LVDSSOLHGDWERWKHQGVHFWLRQVLQ&DVH&LQWKHVDPH
GLUHFWLRQZKHUHDVWKHFKDQQHOLVVXEMHFWHGWRRQO\RQHEHQGLQJPRPHQWM = V.LDW
RQHHQGLQ&DVH%WREDODQFHZLWKWKHORQJLWXGLQDOVKHDUVWUHVVHVUHVXOWLQJIURPWZR
FRXSOLQJVKHDUIRUFHV7KHERXQGDU\FRQGLWLRQVDQGVWUHVVGLVWULEXWLRQVRISHUIRUDWHG
FKDQQHOVHFWLRQVZLWKZHEKROHVDUHVKRZQLQ)LJ

 5HVXOWVRIVKHDUHODVWLFEXFNOLQJDQDO\VLV
$ VWDQGDUG SODLQ &FKDQQHO ZLWK D PHPEHU OHQJWK RI  PP D ZHE GHSWK RI 
PPDIODQJHZLGWKRIPPDQGDOLSVL]HRIPP ZDVVHOHFWHG DV DSUHIHUUHG
VHFWLRQ WKURXJKRXW DOO WKH DQDO\VHV 7KH WRWDO OHQJWK DQG WRWDO GHSWK RI WKHZHEFXW
RXWVDUHUHIHUUHGWRDVLhDQGdhUHVSHFWLYHO\


  
&DVH% &DVH& &DVH% &DVH&
5HFWDQJXODU:HE+ROHVdh PPLh PP 5HFWDQJXODU:HE+ROHVdh PPLh PP

   
&DVH% &DVH& &DVH% &DVH&
6ORWWHG:HE+ROHVdh PPLh PP 6ORWWHG:HE+ROHVdh PPLh PP
)LJ6KHDUEXFNOLQJPRGHVKDSHV

)LJVKRZVWKHVKHDUEXFNOLQJPRGH VKDSHVRI FKDQQHOVHFWLRQVZLWKVPDOODQG
ODUJH FHQWUDO ZHE RSHQLQJV 7KH EXFNOLQJ PRGH VKDSHV IRU &DVH & DUH V\PPHWULFDO
DERXWWKHGLDJRQDOD[LVUHVXOWLQJIURPDSDLURIHTXDOPRPHQWVDFWLQJDWWZRHQGVHF
WLRQVLQWKHVDPHGLUHFWLRQ:KHQPRPHQWLVDSSOLHGDWRQHHQGLQ&DVH%WKHEXFN
OLQJSDWWHUQVFRQFHQWUDWHGDWRQHFRUQHURIWKHZHEKROHGXHWRWKHHIIHFWRIWKHEHQG
LQJDWRQHHQG)XUWKHULQWKHFDVHVRIODUJHRSHQLQJVWKHEXFNOLQJPRGHVKDSHVEH
FRPHPRUHORFDOL]HG
7KH VKHDU EXFNOLQJ FRHIILFLHQWV kv IRU WKH ZKROH VHFWLRQ ZHUH EDFN FDOFXODWHG
EDVHGRQWKHIROORZLQJHTXDWLRQ
278 D. K. Pham et al.



  P  ¨
§h· V

¸ cr 
kv  
©t¹ 

S EAw
ZKHUHhLVWKHGHSWKRIWKHIODWSRUWLRQRIWKHZHEtLVWKHWKLFNQHVVRIWKHZHEELV
<RXQJ¶VPRGXOXVμLV3RLVVRQ¶VUDWLRAwLVWKHFURVVVHFWLRQDODUHDRIZHEHOHPHQW
DQGfyLVWKHGHVLJQ\LHOGVWUHVV
7KHUHGXFWLRQLQWKHVKHDUEXFNOLQJFRHIILFLHQWVkvRIFKDQQHOVZLWKFHQWUDOZHE
KROHVUHVXOWLQJIURPWKHLQFUHDVHLQKROHVL]HDUHLOOXVWUDWHGLQ)LJ

 
$5K &DVH&
$5K &DVH%
$5K &DVH&
 
$5K &DVH%
$5K &DVH&
6KHDUEXFNOLQJFRHIILFLHQWkv

6KHDUEXFNOLQJFRHIILFLHQWkv
$5K &DVH%
 

 

 

 
$5K &DVH&
$5K &DVH%
$5K &DVH&
 
$5K &DVH%
$5K &DVH&
$5K &DVH%
 
               
5DWLRRIUHFWDQJXODUKROHGHSWKWRZHEGHSWKdh/h) 5DWLRRIVORWWHGKROHGHSWKWRZHEGHSWKdh/h)
)LJ6KHDUEXFNOLQJFRHIILFLHQWVRIFKDQQHOVZLWKYDULRXVKROHVL]HV

2YHUDOOWKHVKHDUEXFNOLQJFRHIILFLHQWVRIFKDQQHOVZLWKUHFWDQJXODUKROHVDUHORZ
HUWKDQWKRVHZLWKVORWWHGRSHQLQJVUHVXOWLQJIURPWKHODUJHUDUHDRIUHFWDQJXODUKROHV
FRPSDUHGWRWKDWRIVORWWHGRSHQLQJVRIWKHVDPHRYHUDOOGLPHQVLRQV7KHUHODWLRQVKLS
FXUYHIRU&DVH%KDYHWKHVDPHWUHQGRIUHGXFWLRQDVWKDWRI&DVH&7KHYDOXHVkvIRU
&DVH%DUHDOZD\VORZHUWKDQWKRVHIRU&DVH&GXHWRWKHFRPSUHVVLYHVWUHVVLQWKH
ERWWRPIODQJHDQGOLSFDXVHGE\EHQGLQJPRPHQWDWRQHHQG)XUWKHUWKHODUJHUWKH
ZHERSHQLQJDUHDWKHFORVHUDUHWKHWZRUHODWLRQVKLSFXUYHVRI&DVH%DQG&DVH&,W
FDQ EHH[SODLQHGWKDWVKHDU JRYHUQVWKH EXFNOHRIWKH ZKROHVHFWLRQDQGWKHUHLVQR
HIIHFWRIEHQGLQJDWDYHU\ORZVKHDUEXFNOLQJFRHIILFLHQWLQERWK&DVH%DQG&DVH&
7KLVZDVDOVRUHFRJQL]HGLQWKHSUHYLRXVVWXGLHVE\&+3KDP>@DQG6+3KDP>@

 'LPHQVLRQDOWUDQVIRUPDWLRQ

2QWKHEDVLVRIHODVWLFVKHDUEXFNOLQJDQDO\VHVRISHUIRUDWHGFKDQQHOVZLWKHORQJDWHG
ZHEKROHV ZLWK D ZLGHUDQJH RI KROHVL]HV IURP DVPDOOVL]H[ PP WR D YHU\
ODUJHDUHD[PPDGLPHQVLRQDOWUDQVIRUPDWLRQLVSURSRVHGLQWKLVSDSHUWR
WUDQVIRUP WKH GLPHQVLRQV RI VORWWHG RSHQLQJV LQWR HTXLYDOHQW VL]HV RI UHFWDQJXODU
KROHVDVIROORZV
 Ah  eq  Ah  D
Elastic Buckling Solution for Perforated Thin-walled Channel … 279

§ Lh ·
 d h  eq ¨  d   ¸ d h  E
© h ¹
Ah  eq
 Lh  eq  F
d h eq
ZKHUHAh LVWKHRULJLQDODUHDRIDVORWWHGKROH Ah  eq LVWKHDUHDRIWKHHTXLYDOHQWUHF
WDQJXODU KROH L h  e q LV WKH HTXLYDOHQW KROH OHQJWK DQG d h  eq LV WKH HTXLYDOHQW KROH
GHSWK




(TXLYDOHQWKROHOHQJWKPP

6KHDUEXFNOLQJFRHIILFLHQWkv
 

 

 
\ [ (
 \§[
5ð  
$5K 
 $5K (4
$5K 
 $5K (4
 $5K 
$5K (4
        
       
(TXLYDOHQWKROHGHSWKPP
5DWLRRIRULJLQDOKROHGHSWKWRZHEGHSWKdh/h

)LJ7UDQVIRUPHGKROHGHSWKVDQGOHQJWKV )LJ6KHDUEXFNOLQJFRHIILFLHQWVRIRULJLQDO
ZLWKARh  IRU&DVH& DQGWUDQVIRUPHGVORWWHGKROHVIRU&DVH&
7KHOLQHDUUHJUHVVLRQHTXDWLRQ\§[LH L h  eq § d h  eq LQ)LJGHPRQVWUDWHVWKDW
WKHHTXLYDOHQWKROHVKDSHVRIVORWWHGRSHQLQJVZLWKARh LHFLUFXODUKROHVDUH
VTXDUHV)XUWKHUWKHUHODWLRQ d h  eq dh LVREWDLQHGZKHQVXEVWLWXWLQJWKHUDWLRLh
/dh LQWR(T7KLVSURYHVWKDWWKHGLPHQVLRQDOWUDQVIRUPDWLRQLQWKLVSDSHU
JLYHQE\(TLQWKHFDVHRI ARh  LVFRQVLVWHQWZLWKWKH WUDQVIRUPDWLRQd 
DSURSRVHGE\6+3KDPHWDO>@DQGXVHGE\6+3KDP>@IRUFLUFXODU
DQGVTXDUHKROHVZLWKARRIDQGXSWRZKHUHdLVWKHVTXDUHVL]HDQGDLVWKH
FLUFOHGLDPHWHU$VFDQEHVHHQLQ)LJWKHVTXDUHGDVKHGOLQHVKDYLQJOHJHQGVVXI
IL[HGZLWKµ(4¶UHSUHVHQWLQJWKHYDOXHV kvRIFKDQQHOVZLWK HTXLYDOHQW RSHQLQJV DUH
DOPRVW LGHQWLFDO WR WKH FXUYHV IRU WKH RULJLQDO VORWWHG KROHV 7KLV SURYHV WKDW WKH GL
PHQVLRQDOWUDQVIRUPDWLRQSURSRVHG LQ WKLV VHFWLRQ DUH FDSDEOH RI EHLQJ XVHG IRU WKH
VKHDUEXFNOLQJDQDO\VHVLQWKHFDVHVRIVORWWHGZHERSHQLQJV

 &RQFOXVLRQ

7KH DLP RI WKLV SDSHU LV WR SURYLGH D VROXWLRQ WR HODVWLF VKHDU EXFNOLQJ DQDO\VHV RI
SHUIRUDWHGFKDQQHOVHFWLRQVZLWKHORQJDWHGZHERSHQLQJVZLWKDQDVSHFWUDWLRRI
7KHFULWLFDOEXFNOLQJORDGVZHUHGHWHUPLQHGEDVHGRQWKH)(PRGHOVZLWKWZRFDVHV
RI VWUHVV GLVWULEXWLRQ &DVH % DQG &DVH & PHQWLRQHG LQ WKH SUHYLRXV VWXGLHV 7KH
280 D. K. Pham et al.

UHVXOWVRIWKHVKHDUEXFNOLQJDQDO\VHVVcrZLWKDZLGHUDQJHRIKROHVL]HVLQWKLVSD

SHUFDQEHXVHGDVDGDWDEDVHIRUWKHVKHDUEXFNOLQJLQSXWLQDFRPSDQLRQSDSHULQWKLV
FRQIHUHQFH ³The Behavior of Cold-formed Channel Sections with Elongated Web
holes in Shear´7KHSURSRVHGGLPHQVLRQDOWUDQVIRUPDWLRQLQWKLVSDSHULVDOVR XWL
OL]HG LQ WKH FRPSDQLRQ SDSHU WR GHWHUPLQH WKH HTXLYDOHQW VL]HV RI WKH VORWWHG ZHE
RSHQLQJV

$FNQRZOHGJHPHQW )XQGLQJ SURYLGHG E\ WKH $XVWUDOLDQ 5HVHDUFK &RXQFLO 'LVFRYHU\ 3URMHFW
*UDQW'3KDVEHHQXVHGWRSHUIRUPWKLVSURMHFW7KHDXWKRUVDUHJUDWHIXOWR'U6RQJ
+RQJ3KDPIRUKLVHQWKXVLDVWLFVXSSRUW

5HIHUHQFHV

>@ 6&:/DXDQG*-+DQFRFNBuckling of thin flat-walled structures by a spline finite strip

method.7KLQZDOOHGVWUXFWXUHVS
>@ &+ 3KDP DQG *- +DQFRFN Elastic buckling of cold-formed channel sections in shear.
7KLQ:DOOHG6WUXFWXUHVS
>@ &+ 3KDP DQG *- +DQFRFN Shear buckling of thin-walled channel sections. -RXUQDORI
&RQVWUXFWLRQDO6WHHO5HVHDUFKS
>@ *- +DQFRFN DQG &+ 3KDP Shear buckling of channel sections with simply supported
ends using the semi-analytical finite strip method.7KLQ:DOOHG6WUXFWXUHVS

>@ &+3KDPBuckling studies of thin-walled C-sections with square holes in shear using the
spline finite strip methodLQ8th Int. Conf. Adv. Steel Struct., Lisbon, Portugal
>@ * (FFKHU Isoparametric spline finite strip analysis of perforated thin-walled steel
structures6FKRRORI&LYLO(QJLQHHULQJ)DFXOW\RI(QJLQHHULQJDQG,QIRUPDWLRQ7KH
8QLYHUVLW\RI6\GQH\
>@ &+3KDPShear buckling of plates and thin-walled channel sections with holes.-RXUQDO
RI&RQVWUXFWLRQDO6WHHO5HVHDUFKS
>@ 6+ 3KDP Design of Cold-Formed Steel Beams with Holes and Transverse Stiffeners in
Shear3K'7KHVLV6FKRRORI&LYLO(QJLQHHULQJ7KH8QLYHUVLW\RI6\GQH\
>@ $61=6AS 4600 Cold-formed Steel StructuresAS/NZS 4600:20186WDQGDUGV
$XVWUDOLD6WDQGDUGV$XVWUDOLD6WDQGDUGV1HZ=HDODQG
>@ 6+3KDP&+3KDPDQG*-+DQFRFNDirect strength method of design for channel
sections in shear with square and circular web holes. -RXUQDO RI 6WUXFWXUDO (QJLQHHULQJ
S


 7KH%HKDYLRXURI&ROGIRUPHG&KDQQHO6HFWLRQVZLWK
(ORQJDWHG:HE+ROHVLQ6KHDU

'X\.KDQK3KDP&DR+XQJ3KDP6RQJ+RQJ3KDPDQG*UHJRU\-+DQFRFN

6FKRRORI&LYLO(QJLQHHULQJ7KH8QLYHUVLW\RI6\GQH\6\GQH\16:$XVWUDOLD
duykhanh.pham@sydney.edu.au; caohung.pham@sydney.edu.au;
songhong.pham@sydney.edu.au; gregory.hancock@sydney.edu.au

$EVWUDFW&ROGIRUPHG&VHFWLRQVSHUIRUDWHGZLWKFLUFXODUDQGVTXDUHZHERSHQ
LQJVLQVKHDUKDYHEHHQVWXGLHGWKRURXJKO\+RZHYHUWKHZHEKROHVDUHOLNHO\WR
EHHORQJDWHGDORQJWKHVSDQOHQJWKGXHWROLPLWHGZHEGHSWKWRDOORZPRUHDFFHVV
IRUEXLOGLQJVHUYLFHVLQSUDFWLFH7KLVSDSHUVXPPDUL]HVDQH[SHULPHQWDOVWXG\
RQFKDQQHOVLQVKHDUZLWKHORQJDWHGZHEKROHV7KHVWXG\XVHVDGXDODFWXDWRU
WHVWULJZLWKWKHDLPRIPLQLPL]LQJWKHHIIHFWVRIEHQGLQJPRPHQWVDWWKHWZR
HQGVRIWKHVKHDUVSDQDQGREWDLQLQJWKHSXUHVKHDUFDSDFLW\ZLWKDQDVSHFWUDWLR
XSWR$FRPSDULVRQRI'LUHFW6WUHQJWK0HWKRGGHVLJQORDGVIRUVKHDUZLWK
WKHWHVWUHVXOWVLVFRQGXFWHG7KHVKHDU\LHOGORDGVEDVHGRQWKHSUHYLRXVSUR
SRVDOVIRUWKH'60LQVKHDUIRUSHUIRUDWHGFKDQQHOVZLWKVTXDUHDQGFLUFXODUZHE
KROHVDUHXVHGLQWKLVSDSHUWRDVVHVVWKHDSSOLFDELOLW\RIWKRVHSURSRVDOVWRWKH
FDVHVRIHORQJDWHGZHERSHQLQJV

.H\ZRUGV&ROGIRUPHGVWHHOVKHDUVWUHQJWKHORQJDWHGZHERSHQLQJVGLUHFW
VWUHQJWKPHWKRG

 ,QWURGXFWLRQ
&ROGIRUPHGVWHHOPHPEHUVDUHFRPPRQO\SHUIRUDWHGWRDOORZDFFHVVIRUEXLOGLQJVHU
YLFHVVXFKDVSOXPELQJKHDWLQJDQGHOHFWULFDOV\VWHPVDQGWRLQFUHDVHWKHFHLOLQJFOHDU
DQFHKHLJKW+HQFHWKLVKDVUHVXOWHGLQDFRQVLGHUDEOHQXPEHURIVWXGLHVFRQGXFWHGWR
LQYHVWLJDWHWKHHIIHFWVRIZHEKROHVRQFROGIRUPHGVWHHOPHPEHUVPDLQO\ZLWKVTXDUH
DQGFLUFXODURSHQLQJV%HFDXVHRIWKHGHPDQGIRUPRUHVHUYLFHV\VWHPVLQSUDFWLFHWKH
ZHEKROHVDUHOLNHO\WREHHQODUJHGDORQJWKHVSDQOHQJWKRIPHPEHUVGXHWRWKHOLPLWHG
ZHEGHSWK7KHUHIRUHWKHHIIHFWVRIWKHVHKROHVKDSHVRQWKHVWUXFWXUDOEHKDYLRXUDQG
WKHVWUHQJWKFDSDFLW\RISHUIRUDWHGPHPEHUVVKRXOGDOVREHVWXGLHG
)RUSHUIRUDWHGPHPEHUVVXEMHFWHGWRVKHDUPDQ\H[SHULPHQWDOLQYHVWLJDWLRQVZHUH
SHUIRUPHG E\&+ 3KDPHW DO >@ DQG .HHUWKDQ 0DKHQGUDQ >@+RZHYHUWKHVH
VWXGLHV RQO\ IRFXVHG RQ VTXDUH DQG FLUFXODU ZHE KROHV $ 'LUHFW 6WUHQJWK 0HWKRG
'60GHVLJQUXOHIRUSHUIRUDWHGPHPEHUVZLWKVTXDUHDQGFLUFXODUZHEKROHVLQVKHDU
KDVEHHQGHYHORSHGE\6+3KDP>@ZLWKDVSHFWUDWLRVAR VKHDUVSDQZHEGHSWK
XSWR2QWKHEDVLVRIWKHGHYHORSPHQWRIWKH'LUHFW6WUHQJWK0HWKRG'60DQ
H[SHULPHQWDOSURJUDPFRPSULVLQJWKLUW\VKHDUWHVWVRQFKDQQHOVHFWLRQVZLWKUHFWDQJX
ODUDQGVORWWHGZHEKROHVZDVFRQGXFWHGDWWKH8QLYHUVLW\RI6\GQH\WRREVHUYHWKH

© Springer Nature Singapore Pte Ltd. 2020 281

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_42
282 D. K. Pham et al.

LQIOXHQFHRIHORQJDWHGZHERSHQLQJVRQFROGIRUPHGFKDQQHOVHFWLRQVLQVKHDUDQGWR
HYDOXDWHWKHDSSOLFDELOLW\RIWKHSUHYLRXVSURSRVDOVWRSHUIRUDWHGPHPEHUVZLWKHORQ
JDWHGZHEKROHV
7KH'LUHFW6WUHQJWK0HWKRGRIGHVLJQIRUFKDQQHOVHFWLRQVLQVKHDULVDQDSSURDFK
UHTXLULQJWZRLQSXWVLQFOXGLQJWKHVKHDUHODVWLFEXFNOLQJORDGVcrDQGWKHVKHDU\LHOG
ORDGVy7KHYDOXHVRIQRPLQDOVKHDUVWUHQJWKVnRIFKDQQHOVZLWKRXWZHEKROHVDQG
ZLWKRXWZHEVWLIIHQHUVH[FOXGLQJDQGLQFOXGLQJ7HQVLRQ)LHOG$FWLRQ7)$DUHVSHFL
ILHGLQ$61=6>@DVIROORZV
:LWKRXW7)$ )RU Ov ” Vn Vy  D
)RU Ov ” Vn  Vcr V y  E
)RU Ov ! Vn Vcr  F
   
:LWK7)$ )RU Ov ” Vn Vy  D
 
ª § Vcr · º § Vcr ·
)RU Ov ! Vn «    ¨ V ¸ » ¨ V ¸ Vy  E
«¬ © cr ¹ »¼ © cr ¹
7KHQRQGLPHQVLRQDOVOHQGHUQHVVȜvDQGWKHVKHDUHODVWLFEXFNOLQJORDGVcrDUH
GHWHUPLQHGUHVSHFWLYHO\E\
 Ov V y  Vcr  

k vS EAw
 Vcr 
 
  P   h  t 


ZKHUHkvLVWKHVKHDUEXFNOLQJFRHIILFLHQWIRUWKHZKROHVHFWLRQhLVWKHGHSWKRIWKHIODW
SRUWLRQRIWKHZHEtLVWKHWKLFNQHVVRIWKHZHEELV<RXQJ¶VPRGXOXVμLV3RLVVRQ¶V
UDWLRAwLVWKHFURVVVHFWLRQDODUHDRIZHEHOHPHQWDQGfyLVWKHGHVLJQ\LHOGVWUHVV
 7KHVKHDU\LHOGORDGVyRIXQSHUIRUDWHGFKDQQHOVDFFRUGLQJWR$61=6
>@LVGHWHUPLQHGE\
 Vy  Aw f y  
7KHVKHDU\LHOGORDGVy,netRISHUIRUDWHGPHPEHUVZLWKZHERSHQLQJVEDVHGRQWKH
QHWZHEDUHDPRGHOSURSRVHGE\8QDELD%>@LVIRUPXODWHGE\WKHIROORZLQJH[SUHV
VLRQ
 V y  net  h  d h  f y t  
7KHVKHDU\LHOGORDGVyhRISHUIRUDWHGPHPEHUVZLWKVTXDUHDQGFLUFXODUZHEKROHV
EDVHGRQWKH9LHUHQGHHOPRGHOSURSRVHGE\6+3KDP>@IRUVKHDUDVSHFWUDWLRVXSWR
LVJLYHQDVIROORZV
d
:KHQ   h d   Vyh Vy  D
h
d § d h   · V  V  
:KHQ   h    Vyh Vy   ¨ ¸ y vrd  E
h © h ¹
The Behaviour of Cold-formed Channel Sections … 283

dh
:KHQ  d  V yh Vvrd  F
h
ZKHUHVvrdLVWKHVKHDUORDGRISHUIRUDWHGVHFWLRQVZLWKODUJHZHERSHQLQJVEDVHGRQ
WKH9LHUHQGHHOPHFKDQLVPVvrd,0.6LVWKHYDOXHRIVvrdWKDWLVFRPSXWHGZLWKWKHUDWLR
dh /h dhLVWKHWRWDOGHSWKRIZHEKROH
 6KHDU7HVWVRI&ROGIRUPHG&KDQQHOVZLWK:HE+ROHVXVLQJ
'XDO$FWXDWRU7HVW5LJ
 7HVW3URJUDP7HVW5LJ'HVLJQDQG7HVW3URFHGXUH
$VWDQGDUGSODLQ&FKDQQHOZLWKDZHEGHSWKRIPPDIODQJHZLGWKRIPPDQG
DOLSVL]HRIPPZDVVHOHFWHGDVDSUHIHUUHGVHFWLRQWKURXJKRXWDOOWKHWHVWV7KH
GHSWKVRIKROHVZHUHGHVLJQHGZLWKLQDUDQJHRIWRPP%RWKUHFWDQJXODU5(&
DQGVORWWHG6/7KROHVZHUHVWXGLHGDVVKRZQLQ)LJ7KHFKRVHQDVSHFWUDWLRVRI
KROHOHQJWKLhWRKROHGHSWKdhZHUHDQGARh

 
)LJ'LPHQVLRQVRI&FKDQQHOVSHFLPHQVZLWKZHERSHQLQJVLQPP

$GXDODFWXDWRUWHVWULJZDVGHVLJQHGDQGGHYHORSHGE\6+3KDPHWDO>@IRUWKH
SXUSRVH RI PLQLPL]LQJ WKH HIIHFWV RI EHQGLQJ PRPHQWV WR REWDLQ WKH SUHGRPLQDQWO\
VKHDUEHKDYLRXUDQGWKHVKHDUVWUHQJWKFORVHWRSXUHVKHDUFDSDFLW\ZLWKARXSWR
7KLVWHVWULJZDVDOVRXVHGWKURXJKRXWWKHVKHDUWHVWVRQFROGIRUPHGFKDQQHOVZLWKZHE
RSHQLQJVLQWKLVVWXG\$QDFWXDOWHVWVHWXSDQGVFKHPDWLFLQWHUQDOIRUFHGLDJUDPVLQ
WKHWHVWVDUHVKRZQLQ)LJVDQG
2877(5$&78$725 ,11(5$&78$725

6KHDU6SDQ

%HQGLQJ0RPHQW 0&
0$

9$ 9&
6KHDU)RUFH
 

)LJ$FWXDOVKHDUWHVWVHWXS )LJ)RUFHGLDJUDPVLQWKHVKHDUWHVWV

,QRUGHUWRDSSURDFKSUHGRPLQDQWO\VKHDUEHKDYLRXUDQGFDSWXUHWKHSXUHVKHDUFD
SDFLW\WKHORDGVZHUHWUDQVIHUUHGIURPWZRLQGHSHQGHQWDFWXDWRUVWRDSDLURIFDQWLOHYHU
FROGIRUPHGVWHHOFKDQQHOVYLDORDGLQJSODWHVZLWKSURSHUPRYHPHQWUDWHVWRREWDLQD
SDLURIHTXDODQGRSSRVLWHVLJQEHQGLQJPRPHQWVDWWZRHQGVRIWKHVKHDUVSDQ)RXU
SDLUVRIWXUQEXFNOHVFRQQHFWLQJWKHULJLGIUDPHZLWKURWDWLQJIUHHSODWHVRYHUDQGXQGHU
284 D. K. Pham et al.

WKHORDGLQJSODWHVWRSUHYHQWODWHUDOGLVSODFHPHQWVDQGWRPDLQWDLQWKHYHUWLFDOGLVSODFH
PHQWRIWKHEHDPVGXULQJWKHWHVWV7KHGLVWRUWLRQRIWKHIODQJHVLVSUHYHQWHGE\SDLUV
RIVWUDSVVFUHZHGWRERWKWRSDQGERWWRPIODQJHV
)RUPLQLPL]LQJWKHHIIHFWVRIEHQGLQJDPRPHQWUDWLRMC/MARIZDVPDLQ
WDLQHGGXULQJWKHWHVWV2QFHWKHUDWLRMC/MAGHYLDWHGIURPWKHUHTXLUHGYDOXH
E\ WKH PRYHPHQW UDWHVPPPLQRIWKHWZR DFWXDWRUV ZHUHPRGLILHG SURSHUO\
7KLVZHUHUHSHDWHGFRQWLQXRXVO\GXULQJWKHWHVWVXQWLOWKHPHPEHUUHDFKHGWKHXOWLPDWH
VKHDUVWUHQJWKVn,test
 ([SHULPHQW5HVXOWV
7KHIDLOXUHPRGHVKDSHVRIWKHWHVWVRQFKDQQHOEHDPVLQVKHDUZLWKUHFWDQJXODUDQG
VORWWHG KROHV DUH VKRZQ LQ )LJ  6KHDU IDLOXUH EDQGV ZLWK ARh   DQG  ZHUH
V\PPHWULFDODURXQGWKHGLDJRQDOD[LVDFURVVWKHVKHDUVSDQ7KLVGHPRQVWUDWHVWKDWWKH
PRPHQWUDWLRMC/MARIZDVPDLQWDLQHGGXULQJWHVWLQJ7KHVKHDUIDLOXUHEDQGV
VSUHDGLQJIURPWZRILOOHWHGFRUQHUVRIKROHVEHFRPHPRUHORFDOL]HGZKHQWKHKROHVDUH
HORQJDWHGDORQJERWKWKHOHQJWKRIWKHVKHDUVSDQDQGWKHZHEGHSWK7KHVHIDLOXUHEDQGV
ZHUHIRUPHGIURPWKHLQWHUVHFWLRQVRIWKHILOOHWHGFXUYHVDQGWKHKROHHGJHVDORQJWKH
VKHDUVSDQWRWKHWZRIODQJHVLQVWHDGRIWKHWZRFRUQHUVRIWKHVKHDUVSDQDVZLWKVPDOOHU
RSHQLQJVDVVKRZQLQ)LJFGZKHUHWKHIDLOXUHVKHDUEDQGVRFFXUUHGDURXQGWKH
DUHDVFORVHWRWKHKROHFRUQHUVLQWKHFDVHVRIVXEVWDQWLDOO\ODUJHRSHQLQJDUHDdh 
PPARh 

 
D&5(&[ E&6/7[

 
F&5(&[ G&6/7[
)LJ)DLOXUHPRGHVLQWKHVKHDUWHVWVZLWKHORQJDWHGZHERSHQLQJV

 (ODVWLF%XFNOLQJ$QDO\VLVRI3HUIRUDWHG&KDQQHOVZLWK:HE
2SHQLQJVLQ6KHDU
7KH)LQLWH(OHPHQW0HWKRG)(0XVLQJ$EDTXVLVXWLOL]HGIRUWKHGHWHUPLQDWLRQRI
WKHHODVWLFEXFNOLQJORDGVLQVKHDUVcr,QRUGHUWRGHWHUPLQHSUDFWLFDOYDOXHVRIWKH
VKHDUHODVWLFEXFNOLQJORDGZKHQWKHUDWLREHWZHHQWKHEHQGLQJPRPHQWVDWWKHWZRHQG
VHFWLRQVDVVXPHGWREH&DVH&LQ6+3KDP>@LVXVHGZKHUHDSDLURIEHQGLQJ
PRPHQWVZLWKDKDOIYDOXHM/2 = V. L/2ZKHUHLLVWKHVKHDUVSDQOHQJWKLVDSSOLHG
DWERWKHQGVHFWLRQVLQWKHVDPHGLUHFWLRQWREDODQFHZLWKWKHORQJLWXGLQDOVKHDUVWUHVVHV
UHVXOWLQJIURPWKHVKHDUIORZGLVWULEXWLRQIURPWKHWZRFRXSOLQJVKHDUIRUFHV$W\SLFDO
The Behaviour of Cold-formed Channel Sections … 285

FRQILJXUDWLRQIRUWKHHODVWLFEXFNOLQJDQDO\VLVRISHUIRUDWHGFKDQQHOVZLWKHORQJDWHG
ZHERSHQLQJVLVVKRZQLQ)LJ7KHGHWDLOHG)(PRGHOVDUHSUHVHQWHGLQDFRPSDQLRQ
SDSHU³Elastic Buckling Solution to Perforated Thin-walled Channel Sections in Shear
with an Aspect Ratio of 2.0´LQWKLVFRQIHUHQFH
$FFRUGLQJWRDQHZSURSRVDOLQWKHFRPSDQLRQSDSHUWKHYDOXHVRIVKHDUEXFNOLQJ
FRHIILFLHQWVkvRIWKHFKDQQHOPHPEHUVZLWKFHQWUDOVORWWHGZHEKROHVZLWKDQDVSHFW
UDWLRARRIFDQEHGHWHUPLQHGEDVHGRQDGLPHQVLRQDOWUDQVIRUPDWLRQDVIROORZV
 Ah  eq  Ah  D
§ Lh ·
 d h  eq ¨  d   ¸ d h  E
© h ¹
Ah  eq
 Lh  eq  F
d h  eq
ZKHUHAhLVWKHRULJLQDODUHDRIWKHVORWWHGKROH Ah  eq LVWKHDUHDRIWKHHTXLYDOHQWUHF
WDQJXODUKROH L h  eq LVWKHHTXLYDOHQWKROHOHQJWKDQG d h  eq LVWKHHTXLYDOHQWKROHGHSWK

 
)LJ%RXQGDU\FRQGLWLRQVDQGVWUHVVGLVWULEXWLRQVXQGHUFRQVLGHUDWLRQ

 $SSOLFDELOLW\RI3UHYLRXV'603URSRVDOVWR3HUIRUDWHG
&KDQQHOVLQ6KHDUZLWK(ORQJDWHG:HE+ROHV
$OOUHVXOWLQJSRLQWVIURPWKHVKHDUWHVWVDUHSORWWHGLQ)LJDJDLQVWWKHFXUUHQW'60
VKHDUFXUYHVZLWKDQGZLWKRXW7)$ZKHUHWKHVKHDU\LHOGORDGVZHUHGHWHUPLQHGXVLQJ
Vy,netDQG VyhJLYHQE\(TVDQGUHVSHFWLYHO\7KHKROHWUDQVIRUPDWLRQEDVHGRQ
(TZDVDSSOLHGWRGHWHUPLQHERWKWKHVKHDUHODVWLFEXFNOLQJORDGVVcrDQGWKHVKHDU
\LHOGORDGVVy,netVyh)RUWKH'60LQVKHDUZLWKVy,netDOOVKHDUWHVWSRLQWVZLWKVPDOO
ZHERSHQLQJdh PPOLHFORVHWRWKH'60VKHDUFXUYHZLWK7)$:KLOHWKHUHVXOWV
ZLWKUHVSHFWWRVTXDUHDQGFLUFXODUKROHVLQWKHJURXSRIPPKROHGHSWKDUHVWLOOYHU\
FORVHWRWKHFXUYHZLWK7)$WKHUHVXOWLQJSRLQWVIRUH[WHQGHGKROHOHQJWKVLh 
DQGPPVWDUWGHYLDWLQJIURPWKH'60FXUYHZLWK7)$,QWKHFDVHRIYHU\ODUJH
KROHGHSWKdh PPWKH'60VKHDUFXUYHZLWKRXW7)$FDQEHUHFRPPHQGHG
EHFDXVHWKHFKDQQHOVZLWKVXEVWDQWLDOZHERSHQLQJVGRQRWPRELOL]HWKH7)$)RUWKH
'60LQVKHDUZLWKVyhWKHUHVXOWLQJSRLQWVDUHVKLIWHGXSYHU\VLJQLILFDQWO\FORVHWRWKH
'60VKHDUFXUYHZLWK7)$7KLVFXUYHSUHGLFWVZHOOWKHVKHDUVWUHQJWKRIWKHPHPEHUV
ZLWKDQRSHQLQJVL]HXSWR[PP+RZHYHUWKHXVHRIVyhLVRYHUO\FRQVHUYDWLYH
IRUSHUIRUDWHGFKDQQHOVZLWKYHU\ODUJHKROHV[DQG[PP
 286 D. K. Pham et al.

 
'606KHDU&XUYHZLWK7)$ '606KHDU&XUYHZLWK7)$
'606KHDU&XUYHZLWKRXW7)$ '606KHDU&XUYHZLWKRXW7)$
&5(&GKK GK PP &5(&GKK GK PP
 &5(&GKK GK PP  &5(&GKK GK PP
&5(&GKK GK PP &5(&GKK GK PP
&6/7GKK GK PP &6/7GKK GK PP
&6/7GKK GK PP &6/7GKK GK PP
 &6/7GKK GK PP  &6/7GKK GK PP
Vn,test Vy,net

Vn,test Vyh
ARh  ARh 
 
ARh 
 ARh 


 

 
                   

O v
V y  net  Vcr   O V yh  Vcr  v

)LJ6KHDUWHVWUHVXOWVEDVHGRQVy,netDQGVyhYHUVXV'60VKHDUFXUYHV

 &RQFOXVLRQ
7KHWHVWUHVXOWVLQWKLVVWXG\DUHXVHGWRHYDOXDWHWKHDSSOLFDELOLW\RIWKHSUHYLRXVSUR
SRVDOVIRUWKH'60GHVLJQIRUFKDQQHOVZLWKUHFWDQJXODUDQGVORWWHGZHEKROHV2QWKH
EDVLVRIWKHFRPSDULVRQSUHVHQWHGDERYHWKHXVHRI VyhFDQEHUHFRPPHQGHGXSWRD
FHUWDLQOLPLWRIKROHGLPHQVLRQ[PP0RUHDFFXUDWHSUDFWLFDOPRGHOVRIVKHDU
\LHOGORDGVDUHEHLQJLQYHVWLJDWHGWRSURSRVHDEHWWHUGHVLJQIRUSHUIRUDWHGVHFWLRQVLQ
VKHDUZLWKDPRUHH[WHQVLYHUDQJHRIHORQJDWHGKROHVL]HV

$FNQRZOHGJHPHQW )XQGLQJ SURYLGHG E\ WKH $XVWUDOLDQ 5HVHDUFK &RXQFLO 'LVFRYHU\ 3URMHFW
*UDQW'3KDVEHHQXVHGWRSHUIRUPWKLVSURMHFW7KHFROGIRUPHGFKDQQHOVSHFLPHQV
ZHUHSURYLGHGE\%OXH6FRSH6WHHO7KHDXWKRUVJUDWHIXOO\DFNQRZOHGJHWKHVXSSRUW

5HIHUHQFHV
>@ &+3KDP$3HORVL7(DUOVDQG*-+DQFRFNExperimental investigation of cold-
formed C-sections with central square holes in shearLQ23rd International Specialty
Conference on Cold-formed Steel Structures%DOWLPRUH86$
>@ 3 .HHUWKDQ DQG 0 0DKHQGUDQ Experimental studies of the shear behaviour and
strength of lipped channel beams with web openings.7KLQ:DOOHG6WUXFWXUHV
S
>@ 6+3KDPDesign of Cold-Formed Steel Beams with Holes and Transverse Stiffeners
in Shear3K'7KHVLV6FKRRORI&LYLO(QJLQHHULQJ7KH8QLYHUVLW\RI6\GQH\

>@ $61=6AS 4600 Cold-formed Steel Structures6WDQGDUGV$XVWUDOLD
6WDQGDUGV$XVWUDOLD6WDQGDUGV1HZ=HDODQG
>@ 8QDELD%Numerical Simulation of Plain Cold-formed Steel C-sections with Square
Holes subjected to Shear LQ School of Civil Engineering  WKH 8QLYHUVLW\ RI
6\GQH\
>@ 6+3KDP&+3KDP&$5RJHUVDQG*-+DQFRFNExperimental Studies of Cold-
formed Beams under Uniformed Shear Forces with Minimal Bending MomentsLQThe
8th International Conference in Thin-walled Structures, ICTW2018  /LVERQ
3RUWXJDO

 Experimental Investigation of Cold-Rolled Aluminium
Alloy 5052 Columns Subjected to Distortional Buckling

Le Anh Thi HUYNH1, Cao Hung PHAM1 and Kim J.R. RASMUSSEN1
1
School of Civil Engineering, The University of Sydney, NSW 2006, Australia
leanhthi.huynh@sydney.edu.au; caohung.pham@sydney.edu.au;
kim.rasmussen@sydney.edu.au

Abstract. The behaviour and strength of cold-formed columns are mainly in-
fluenced by local, distortional and global buckling or coupled instability phe-
nomena. Distortional buckling may govern the column strengths of open cold-
formed sections with intermediate lengths. The objective of this paper is to in-
vestigate the distortional mode and ultimate capacity of cold-rolled aluminium
channel sections in compression. A total of nine columns with three different
cross-sectional geometries were performed at the University of Sydney. These
commercially available cross-sections were fabricated by using the cold-rolling
process instead of extrusion. The geometric imperfections of channel specimens
were measured using a specially designed measuring rig. Tensile coupon tests
were also conducted from the flat portions and the corner regions of the cross-
sections to determine the material properties. The specimens were tested in axi-
al compression between two fixed ends. The ultimate loads and observed failure
modes in the column tests are reported.

Keywords: Aluminium alloys 5052, Cold-rolled channel sections, Section ca-

pacity, Distortional buckling.

1 Introduction

The structural use of aluminium alloys has been enlarged in the past few decades due
to the advantages of aluminium alloys such as lightweight, superior corrosion re-
sistance, and recyclability. Most of aluminium structural members are manufactured
by extruded method. Recently, BlueScope Permalite Australia [1] has successfully
fabricated cold-rolled aluminium coil into C- and Z- sections by using existing rollers
for cold-formed steel sections, as illustrated in Fig. 1. As seen in these thin-walled
sections, cold-rolled channels may undergo one of three fundamental modes of buck-
ling: local, distortional and overall or a combination of these modes. Most of
aluminium research in the literature is on the extruded section. There have been re-
cently a few researches on cold-rolled aluminium sectional capacities [2–5]. Howev-
er, distortional buckling modes have not been considered in previous research. This
paper presents a series of an experimental investigation on cold-rolled aluminium
channel columns undergoing distortional buckling.

© Springer Nature Singapore Pte Ltd. 2020 287

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_43
288 L. A. T. Huynh et al.

Fig. 1. Cold-rolled C- section [1]

2 Experimental Investigation

2.1 Specimen Dimensions and Geometric Imperfections

A total of nine columns were conducted in the J.W. Roderick Laboratory for Materi-
als and Structures at the University of Sydney. Permalite Ltd (BlueScope) [1] manu-
factured and supplied all cold-rolled cross-sections. Nine column specimens were
selected from three commercially available C-sections (C10030, C25025 and
C40030) cold-rolled from 5052-H36 aluminium alloy strips. Fig.2 and Table 1 show
the nominal dimensions. The lengths of each specimen were chosen to be twice the
half wavelength of distortional buckling from buckling analysis using THIN-WALL-2
program [6] as 700mm for C10030, 1600mm for C25025, and 2000mm for C40030,
respectively.
Thin-wall structural members are more susceptible to buckling which is dramati-
cally affected by geometric imperfections. Therefore, it is necessary to measure the
pattern and magnitude of the imperfections of the specimens before testing. The de-
tails of the measurement are reported in [3,4]. The imperfection data provide essential
input for further numerical simulations of the tests.
Each test specimen was labelled to express the type of channel section, thickness
and number of column test. For example, the label for a typical test “C25025_DC_1”
is defined as follows:
i. “C250” indicates a C-section with the depth of 250 mm (alternatively “C100”
and “C400” for sections with depths of 100 mm and 400 mm respectively).
ii. “25” is the nominal thickness 2.5 mm times 10 in mm (alternatively “30” for 3.0
mm).
iii. “DC” indicates a column test subjected to distortional buckling.
iv. “1” is the sample number of three (alternatively 2 and 3)
Experimental Investigation of Cold-Rolled Aluminium … 289

Table 1. Nominal dimension of specimens

Specimen Thickness Depth Width Lip Inner

t D B L radius r i
mm mm mm mm mm
C10030 3.0 105 60.5 16.0 5
C25025 2.5 255 76.0 25.5 5
C40030 3.0 400 125.5 30.0 5
Fig. 2. Gross section
geometry

2.2 Material Properties

Tensile coupon tests for three different C-sections at the flat and corner position were
reported in [2]. Tensile tests were carried out following the Australian Standard AS
1391:2007 [7]. Yield strength was determined by 0.2% offset method for the continu-
ous yielding materials. It is noted that yield stress at the corner is about 10% higher
than yield stress at the flat position.
The average material properties are summarized in Table 2 where the symbols are
defined as follows: (1) E 0 is the initial Young’s modulus; (2) V 0.2 is the static 0.2%
proof stress; (3) V u is the material ultimate tensile strength; (4) n is the exponent of
the standard Ramberg-Osgood expression, obtained using
Q OQİ 0.01 İ 0.2 )/ln(V 0.01 /V 0.2 ); (5) H u is the plastic strain at ultimate strength; (6) H f is
the plastic strain at fracture, measured over a gauge length of 50 mm.

Table 2: Material properties obtained from tensile coupon tests.

E0 V 0.2 Vu n Hu Hf
Coupon GPa MPa MPa % %
(1) (2) (3) (4) (5) (6)
C10030_Flange 69.7 220 267 12.7 4.7 6.0
C10030_Web 69.6 220 267 12.5 5.1 6.4
C10030_Corner 70.5 244 286 12.5 3.3 4.7
C25025_Lip 69.9 217 257 12.3 5.9 7.7
C25025_Flange 69.7 219 269 12.4 6.1 7.5
C25025_Web 69.8 221 271 12.5 6.9 7.9
C25025_Corner 70.5 240 286 12.8 5.8 6.8
C40030_Lip 69.3 219 268 12.4 4.5 6.0
C40030_Flange 69.4 219 270 11.8 5.6 7.6
C40030_Web 69.5 223 272 12.3 6.4 8.2
C40030_Corner 70.4 244 284 12.5 4.2 6.1
290 L. A. T. Huynh et al.

2.3 Test set-up and Instrumentation

The columns of this experimental series were concentrically tested in a horizontal
position in compression between fixed ends. The specimens were cut and milled be-
fore testing to ensure that the load is evenly distributed around the cross-sections. Fig.
3 shows the test rig where two reaction I-beams were clamped into a rail system on
the strong floor to prevent the reaction I-beams from pulling out of the ground during
the test. The column specimen was attached between two fixed hinges which were
bolted on top of two reaction I-beams. The right fixed hinge was bolted to the ram of
a 500kN capacity hydraulic jack connected to MTS controls to allow stroke-
controlled operation. A constant displacement rate of 0.5mm/min was maintained
during the test. Two ends of specimens were poured pattern stone to avoid warping
and localised failure at two cross-section ends. The specimen was laterally retrained
by a pair of turnbuckles at mid-span to prevent lateral buckling of the column mem-
ber, and thus to allow distortional buckling. The displacements were measured with
13 transducers as shown in Fig. 3(b). The four LVDTs (A1, A2, B1 and B2) measured
the axial shortening of the column. The remaining transducers were attached to an
aluminium frame held at two corners of the web and flange to locally measure the
movement of the cross-section at mid-span as seen in Fig. 3.

(b) Longitudinal view test set up and Instrumentations

Fig. 3. Medium length column tests set-up
Experimental Investigation of Cold-Rolled Aluminium … 291

3 Experimental Results
Fig. 4 shows load vs end shortening curves of all C25025 specimens. The post-peak failure
mode of typical C25025_DC is shown in Fig. 5. All specimens failed by inelastic distortional
buckling or a combination of material yielding and inelastic distortional buckling. The results
of column tests are shown in Table 3. They include: (1) the area of each cross-sections A, (2)
the length of specimen L, (3) the elastic distortional buckling stress f od calculated based on
THIN-WALL-2 program [6], (4) the distortional buckling load N o d , (5) the ultimate test load
P EXP , (6) flange movement. It is noted that three repeated tests have been conducted, and the
differences between three test results were approximately 5%, which indicate the reliability of
the column tests. A detailed discussion was reported in [8].
120

100

80
Load (kN)

60

40
C25025_DC_1
20 C25025_DC_2
C25025_DC_3
0
0 2 4 6 8 10
End shortning (mm)

Fig. 4. Load-end shortening curves of Fig. 5 Post peak failure mode of typical
specimen C25025_DC C25025_DC_1

Table 3. Column tests result subjected to distortional buckling

Length Flange-lip
A f od N od P EXP
Specimen (L) movement
mm2 m MPa kN kN
(1) (2) (3) (4) (5) (6)
C10030_DC1 707 700 217.8 153.9 114.3 In/outward
C10030_DC2 706 700 217.9 153.8 121.2 In/outward
C10030_DC3 703 700 217.5 152.9 115.3 In/outward

C25025_DC1 1093 1600 63.8 69.7 100.6 In/outward

C25025_DC2 1092 1600 64.1 70.0 100.8 Inward
C25025_DC3 1092 1600 64.0 69.9 98.0 In/outward
C40030_DC1 2049 2000 36.2 74.2 170.1 Inward
C40030_DC2 2050 2000 36.4 74.6 160.6 Inward
C40030_DC3 2025 2000 36.3 73.5 169.4 Inward
292 L. A. T. Huynh et al.

4 Conclusion

An experimental program of nine cold-rolled aluminium alloy 5052 channel columns

with intermediate lengths has been recently performed at the University of Sydney to
study the behaviour and determine the ultimate capacities of cold-rolled aluminium
channel columns subjected to distortional buckling. Three commercially available
cross-sections supplied by BlueScope Permalite were chosen for this study. The ob-
served failure mode is inelastic distortional buckling. These test results will be used
along with the finite element models to calibrate new proposed design guidelines to
determine the distortional buckling strength for these types of cold-rolled channel
sections. Research is ongoing at the University of Sydney.

Acknowledgement
Australian Research Council Linkage Research Grant LP140100563 between the
University of Sydney and BlueScope Lysaght funded this study. The authors would
like to thank Permalite Aluminum Building Solutions Pty Ltd for the supply of the
test specimens and financial support for the project.

References
1. Permalite, http://permalite.com.au/, Permalite Alum. Build. Solut. Pty Ltd. (2010).
2. L.A.T. Huynh, C.H. Pham, K.J.R. Rasmussen, Mechanical properties of cold-rolled alu-
minium alloy 5052 channel sections, Proceeding of the 8th International Conference Steel
and Aluminium Structures, Hong Kong, China (2016).
3. L.A.T. Huynh, C.H. Pham, K.J.R. Rasmussen, Stub column tests and finite element model-
ling of cold-rolled aluminium channel section, Proceeding of the 8th International Confer-
ence Steel and Aluminium Structures, Hong Kong, China (2016).
4. N.H. Pham, C.H. Pham, K.J.R. Rasmussen, Incorporation of measured geometric imper-
fections into finite element models for cold-rolled aluminium sections, Proceeding of the
4th Congrès International Géotechnique - Ouvrages –Structures, Hochiminh city, (2017).
5. L.A.T. Huynh, C.H. Pham, K.J.R. Rasmussen, Numerical Simulations of Cold-Rolled Al-
uminium Alloy 5052 Channel Sections in Stub Column Tests, Proceeding of the 4th Congrès
International Géotechnique - Ouvrages –Structures, Hochiminh city, (2017).
6. V.V. Nguyen, G.J. Hancock, C.H. Pham, Development of the Thin-Wall-2 Program for
Buckling Analysis of Thin-Walled Sections Under Generalised Loading, Proceeding of the
8th International Conference in Advance Steel Structure, (2015).
7. AS 1391:2007, Metallic materials - Tensile testing at ambient temperature (2007).
8. L.A.T. Huynh, Strength and Behaviour of Cold-rolled Aluminium Sections, Thesis, Uni-
versity of Sydney (2019).
 Numerical Simulation of Cold-Rolled Aluminium Alloy
5052 Columns Subjected to Distortional Buckling

Le Anh Thi HUYNH1, Cao Hung PHAM1 and Kim J.R. RASMUSSEN1
1 School of Civil Engineering, The University of Sydney, NSW 2006, Australia
leanhthi.huynh@sydney.edu.au; caohung.pham@sydney.edu.au;
kim.rasmussen@sydney.edu.au

Abstract. This paper describes Finite Element (FE) analyses of cold-rolled al-
uminium alloy 5052-H26 columns subjected to distortional buckling using the
program ABAQUS. The results of distortional buckling tests in an experimental
program performed at the University of Sydney are fully described in a com-
panion paper. The simulation results are compared and validated against the
distortional buckling tests. In detailed FE models, effects of such input parame-
ters as mechanical properties of flat and corner areas and initial geometric im-
perfections are considered. The FE ultimate strengths are in good agreement
with the experimental results. Therefore, the FE analysis can be used to predict
the ultimate loads of cold-rolled aluminium alloy members including the post-
buckling behaviour of thin-walled aluminium sections in compression subjected
to distortional buckling. It is indicated that the reliable FE models in this study
can be used to extend data range for future calibration of distortional buckling
strength curve for design of cold-rolled aluminium columns.

Keywords: Aluminium alloys 5052, Cold-rolled channel sections, Distortional

buckling, Finite Element Model.

1 Introduction

Aluminium is commonly used in automotive and aerospace industries as a lightweight

alternative to other metals such as steel, cast iron etc. In construction applications,
aluminium alloy material appeared in roofing, flashing and wall panels. Traditionally,
aluminium alloy sections used in structural applications are fabricated by extrusion.
Past researches on aluminium structural members were focused on the ultimate
strengths of extruded sections for both compact and slender elements [1–4]. Recently
in Australia, aluminium structural solutions, commonly known as PERMALITE [5],
have successfully produced aluminium structural sections using the existing rolling
system for cold-formed steel sections. The aluminium alloy 5052-H36 is chosen for
these products due to its excellent corrosion resistance. To date, design guidelines for
these types of cold-rolled aluminium structural members are limited.
The experimental programs [6] were performed at the University of Sydney to in-
vestigate the behaviour and ultimate strength of cold-rolled aluminium column sub-
jected to distortional buckling. This paper describes the finite element (FE) modelling
and analysis of the experimental aluminium channels in column tests subjected to

© Springer Nature Singapore Pte Ltd. 2020 293

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_44
294 L. A. T. Huynh et al.

distortional buckling. The investigated models have been included the effects of such
input parameters as initial imperfection, mechanical properties at flat and corner are-
as.

2 Finite Element Simulation

2.1 General

A detail FE model has been developed to study the structural behaviour of cold-rolled
aluminium alloy 5052 channel section in the stub column test. The commercially
available software package ABAQUS/Standard, Version 6.14-1[7]. In the first step,
the initial imperfection and measured residual stresses were incorporated into a per-
fect FE model. The second step was a nonlinear analysis using the modified Riks
method [8,9]

2.2 Material Properties

The accuracy of material modelling is one of the most important requirements for the
accuracy of finite element simulations. In this study, measurements of the stress-strain
curves obtained from the coupon tests of the aluminium in each specimen were incor-
porated in the analyses [10]. Fig. 1 shows the typical stress-strain curves of C25025
coupons and the means values of coupon tests of all tested sections are tabulated in
Table 1. The material non-linearity was defined in the FE models by specifying sets of
values of true stress and plastic strain to define a piecewise linear response. The rela-
tionship between the true stress Vtrue and the engineering stress V, as well as the true
plastic strain Htrue and the engineering strain H were calculated using equations (1) and
(2) as follows:
Vtrue = V(1+H) (1)
Htrue = ln(1+H) - Vtrue/E (2)
where V and H are the measured engineering stress and strain respectively obtained
from the tensile coupon tests

Fig. 1 Typical stress-strain curves of C25025 coupons

Numerical Simulation of Cold-Rolled Aluminium … 295

Table 1. Material properties from tensile coupon tests

V0.2 Vu Eo
Coupon Position
MPa MPa GPa
C10030 flat 220 267 69.7
C25025 flat 218 266 69.9
C40030 flat 220 271 69.4
C10030 corner 239 285 70.5
C25025 corner 246 286 70.5
C40030 corner 240 284 70.4

2.3 Test Rig Configuration and Specimen Boundary Conditions

Fig. 2 shows the FE model of a typical C25025 experimental test. Fixed end boundary
conditions with a uniform displacement applied to one end was used in the model and
represented the actual experimental conditions. Contact between nodes of the cross-
section ends and the surfaces of the steel plates were modelled by means of a node-to-
surface contact pair, where the surface of the plate constituted the element-based mas-
ter surface and the edge of the channel was defined as a node-based slave surface. At
the middle of the specimen, two nodes close to the corner of flange-web were re-
strained out of plan displacement to simulate lateral restraints point in the actual test
rig, which was described in the companion paper [6].

Fig. 2 FE model of cold-rolled aluminium column tests subjected to distortional buckling

2.4 Initial Geometrical Imperfection

The effect of initial geometric imperfection is sensitive to the load carrying capacity
of cold-formed structures. Therefore, it is necessary to accurately measure the actual
initial imperfections of aluminium members prior to testing. Fig. 3 shows the rig set-
296 L. A. T. Huynh et al.

up for the imperfection measurement of each specimen. Nine laser devices travel
along high-precision tracks to record the distance from nine corresponding points to
the surface of the specimen. Each laser transducer recorded every 2mm along the
length of the specimen and a reading range around 10mm-25mm. The results of laser
readings were directly inputted into the node numbers of the perfect straight specimen
created by the finite element model. The detail of the process is described in [11].

Fig. 3 Imperfection measurement lines and rig

2.5 Element Type and Mesh

Two steel plates of testing rig at two ends were generated by using 3D-deformable
solid elements with standard mild steel properties. For cold-rolled aluminium mem-
bers, the 4-node shell element with reduced integration, type S4R, is an appropriate
choice because the material thickness of cold-formed members is thin compared to
the width. The mesh size was chosen approximately 10×10mm (length by width) for
all steel plates and aluminium member, a finer mesh was used at the rounded corners
of the aluminium cross-sections.

3 Verification of Finite Element Model

Fig. 4 compares the predicted deformations of specimen C10030_DC_1 with the ex-
perimentally recorded deformations. While Fig. 4a shows the end load versus end
shortening of the specimen, Fig. 4b compares the post-peak deformation of the
C10030_DC_1 test specimen and the FE model prediction. It is noted that the result
reveals an excellent agreement between the predicted and the experimental values.
Table 2 summaries the ultimate strength results of the 9 column tests on cold-rolled
aluminium channel sections. The average ratio of test strength to FE strength was
0.988, with a coefficient of variation of 0.015.
Numerical Simulation of Cold-Rolled Aluminium … 297

140
120
100
Load (kN)

80
60
40 C10030_DC_1_EXP
20 C10030_DC_1_FEM
0
0 2 4 6 8
End shortening (mm)
(a) Load vs end shortening (b) Observation failure mode shape
Fig. 4. Compare FE and test results of specimen C10030_DC_1

Table 2. Test and FE results of column tests subjected to distortional buckling

Specimen Length Test results Finite Element Model

PEXP Flange PFEM Flange PEXP
(mm)
(kN) movement (kN) movement PFEM
C10030_DC_1 700 114.3 In/outward 116.4 In/outward 0.982
C10030_DC_2 700 121.2 In/outward 121.8 In/outward 0.995
C10030_DC_3 700 115.3 In/outward 116.6 In/outward 0.989
C25025_DC_1 1600 100.6 In/outward 101.7 In/outward 0.989
C25025_DC_2 1600 100.8 Inward 101.3 Inward 0.995
C25025_DC_3 1600 98.0 In/outward 100.1 In/outward 0.979
C40030_DC_1 2000 170.1 Inward 169.8 Inward 1.002
C40030_DC_2 2000 169.4 Inward 168.2 Inward 1.007
C40030_DC_3 2000 160.6 Inward 167.3 Inward 0.960
Mean 0.988
CoV 0.015

4 Conclusion

The behaviour of cold-rolled aluminium columns subjected to distortional buckling

was simulated by using the FE program ABAQUS. The geometrical and material non-
linear FE analyses were used in the numerical investigation. The results of the FE
models were compared and calibrated against nine intermediate length column tests.
It is noted that the ABAQUS results were generally in good agreement with the ex-
perimental results. Therefore, these reliable FE models will be subsequently used for
further investigation to develop the design guidelines and optimise cold-rolled alu-
minium sections. Research is ongoing at the University of Sydney.
298 L. A. T. Huynh et al.

Acknowledgement
Funding provided by the Australian Research Council for Linkage Project Grant
LP140100563 between BlueScope Permalite and the University of Sydney has been
used to perform this research. The authors would like to thank Permalite Aluminium
Building Solutions Pty Ltd for the supply of the test specimens and financial support
for the project.

References

[1] M. Su, Behaviour and design of aluminium alloy structural elements, The University
of Hong Kong, 2014.
[2] Y. Zhao, X. Zhai, L. Sun, Test and design method for the buckling behaviors of 6082-
T6 aluminum alloy columns with box-type and L-type sections under eccentric
compression, Thin-Walled Structure, (2016).
[3] R. Feng, W. Zhu, H. Wan, A. Chen, Y. Chen, Tests of perforated aluminium alloy
SHSs and RHSs under axial compression, Thin Walled Structure, 130, 194–212,
(2018).
[4] J.H. Zhu, Z. qi Li, M.N. Su, B. Young, Behaviour of aluminium alloy plain and lipped
channel columns, Thin-Walled Structure, 135, 306–316, (2019).
[5] Permalite, http://permalite.com.au/, Permalite Alum. Build. Solut. Pty Ltd. (2010).
[6] L.A.T. Huynh, C.H. Pham, K.J.R. Rasmussen, Experimental Investigations of Cold-
Rolled Aluminium Alloy 5052 Columns Subjected to Distortional Buckling, Proceeding
of the 5th Congrès International Géotechnique - Ouvrages –Structures, Hanoi,
Vietnam, 2019.
[7] ABAQUS/Standard, ABAQUS/CAE User’s Manual, 2014.
[8] Riks EE, The Application of Newton’s Method to the Problem of Elastic Stability,
Journal of Applied Mechanics 39, 1060–6, (1972).
[9] Riks EE, An incremental approach to the solution of snapping and buckling problems,
International Journal of Solids Structures, 15, 529–51, (1979).
[10] L.A.T. Huynh, C.H. Pham, K.J.R. Rasmussen, Mechanical properties of cold-rolled
aluminium alloy 5052 channel sections, Proceeding of the 8th International Conference
Steel and Aluminium Structures, Hong Kong, China (2016).
[11] N.H. Pham, C.H. Pham, K.J.R. Rasmussen, Incorporation of measured geometric
imperfections into finite element models for cold-rolled aluminium sections, Proceed-
ing of the 4th Congrès International Géotechnique - Ouvrages –Structures, Hochiminh
city, (2017).
 Use of Kriging metamodels for seismic fragility analysis
of structures

Cong-Thuat Dang1, Thanh Tran2, Duy-My Nguyen1, My Pham1 and Thien-Phu Le3
1
University of Science and Technology - The University of Danang, Vietnam
2
Yokohama National University, Japan
3
/0((8QLYHUVLWpG¶eYU\9DO-G¶(VVRQQH(YU\FHGH[)UDQFH
dangcongthuat@dut.udn.vn

Abstract. In civil engineering, a seismic fragility curve is popularly used to predict failure
probability of structures under different earthquakes, and hence propose essential rehabilita-
tion strategies through risk assessment for future earthquakes. The curve shows the failure
probability as a function of seismic intensity, e.g., spectral acceleration at fundamental fre-
quencies of structures (Sa,T1), and can be obtained using one of three approaches: engineer-
ing judgment, empirical studies or numerical simulations. The paper focuses on constructing
seismic fragility curves using numerical simulations, where robust approaches of seismic re-
liability analysis are based on direct Monte Carlo simulation technique. The MCS based
method usually requires a relatively large number of simulations to obtain a sufficiently reli-
able estimate of the fragility. It therefore becomes computationally expensive and time con-
suming as generating the simulations using the actual model or called full model of the
structure. In this regard, this paper suggests using Kriging metamodel as a viable alternative
of the actual model to reduce computational costs in seismic fragility computation. The
Kriging metamodel is constructed based on the training samples of input and corresponding
output responses of the structure. The validation of this method is performed on two numer-
ical examples.

Keywords: Monte Carlo Simulation, Kriging Metamodeling, Fragility Curves

Introduction

Seismic probabilistic assessment (PRA) comprises seismic hazard, fragility curves

and dominant accident sequences, which lead to core damage of structures in nuclear
engineering field [1]. Fragility curves have been applied to diơ erent structural types,
e.g., buildings and bridges. They are useful for structural designs in earthquake-prone
areas, retrofit or disaster response planning. A seismic fragility curve expresses failure
probability of a structure or a mechanical system subjected to earthquakes as a func-
tion of a ground motion index.

Fragility P ¬ª D ! C IM º¼ (1)

where D and C are seismic demand and capacity of the structures; IM is an earthquake
intensity measure. There are three common approaches to build fragility curves such

© Springer Nature Singapore Pte Ltd. 2020 299

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_45
300 C.-T. Dang et al.

as scaled seismic intensity, and probabilistic seismic demand model/probabilistic

seismic capacity model, and maximum likelihood estimation (MLE) [2].
Fragility calculation methods based on Monte Carlo usually require a large amount
of numerical simulations to obtain reliable failure probability of the structures. Unfor-
tunately, these approaches are commonly not feasible in many real cases. The com-
plexity of civil structures such as bridges and buildings demand corresponding ap-
proximated numerical models which are costly to run with thousands of times. For
this reason, third-party approximated metamodels with sufficient accuracy to some
extent can be used alternatively. In recent decades, Kriging based metamodeling [3,4]
has emerged as an efficient statistical approach which involves random errors into the
process of constructing alternative input-output relations. In general, one advantage of
using Kriging metamodel or surrogate is accurate modeling capacity for nonlinear
responses.
In this study, Kriging metamodels are first developed to approximate dynamic re-
sponses of dynamic structures under seismic excitations. The metamodels are then
adopted to construct fragility curves of the structures. Particularly, the fragility curves
can be constructed without the log-normal assumption of distribution function be-
tween failure probability and ground motion index as most of traditional methods.
The efficiency of using the surrogates instead of actual responses is finally demon-
strated by two examples of linear and nonlinear oscillators.

Seismic fragility analysis using Kriging metamodels

2.1 Kriging metamodel

Kriging is a metamodeling technique which considers true model responses as real-

izations of a Gaussian process. Suppose y = f(x) is a function representing the rela-
tionship between input and output of a numerical model. A d-dimensional input vec-
tor x = (x1, x2«xd) of concern is experimentally designed into a sample Xs = (x1,
x2, ..., xn), and a corresponding output matrix is Ys = f (x1, x2«xn). The target is to
construct the function f(x) for any untried inputs; therefore, the pair of (Xs, Ys) can be
considered as a training data set D. A Gaussian process (GP) can interpolate an output
of a certain input in conjugation with the learning data via covariance functions, and
defined as follows:
k
f ( x) m( x )  e(x ) ¦ E h ( x )  e (x )
j 0
j j
(2)

with k basic functions. Without prior knowledge about the input-output relation-
ship, m(x) can be simplified as a one-degree polynomial with (d+1) basic functions
(also called trend), ȕ = [ȕ0ȕ1,«ȕd]T is the unknown regression vector, and h(x) = [1,
x1« xd] is the known corresponding function vector. The stochastic part e(x) is as-
sumed as a zero-mean Gaussian process characterized by covariance function.
The process of Kriging metamodeling can be briefly described within five steps.
The first step is determining all possible uncertainties originating from structures or
Use of Kriging metamodels for seismic fragility … 301

loadings. Secondly, samples of the identified uncertain parameters can be created by

Design-of-Experiments (DOE). Next step is computing corresponding responses from
the above samples of inputs. Then, the Kriging metamodels can be constructed by
functionalizing the relationship between input uncertain parameters and response
realizations. Ground motion related parameters, e.g., Sa,T1 or PGA; and structural
property parameters can be considered as the uncertain parameters. Finally, the
metamodels must be validated before using them as alternative models of actual mod-
els or structures.

2.2 Kriging metamodel based seismic fragility analysis

The seismic fragility analysis considering not only randomness of seismic excitations
but also the uncertainty of structural properties is expected to predict reliable failure
probability for civil structures. Two popular methods to develop the fragility curves as
Maximum Likelihood Estimation (MLE) and Kriging metamodel based MC simula-
tions are adopted in this paper. Herein, the authors focus on how to use the surrogates
to compute failure probability by the MCS method. Given an intensity level, N simu-
lations can be computed using the developed metamodels with considering the ran-
dom variability of the structural parameters. N corresponding outputs are then ob-
tained. For structures subjected to seismic loadings, maximum displacement is com-
monly chosen as a sensitive measure to damage. The responses are then compared to
limit states to attain exceeding failure probability. The fragility curve can be finally
derived from different failure probabilities under various intensities.

Applications

3.1 Synthetic ground motion histories

The stochastic method proposed by Boore [5] assumes that ground motion is dis-
tributed with random phase over a time duration related to earthquake size and propa-
gation distance. The ground motion is characterized by its spectrum - this is where the
physics of earthquake process and wave propagation are contained. The total spec-
trum of the motion at a site S(M0, R, f) is considered as a combination of earthquake
source (E), path (P), site (G) and type of motion (I).

S  M0 , R, f  E  M 0 , f  P(R, f )G( f )I ( f ) (3)

where M0 is the seismic moment that is related to the seismic magnitude M by

2 (4)
M log M 0  10
3

R is the distance from source to site and f is frequency. The source E(M0,f): is based
on the source spectral shape. The path P(R,f) accounts for the effects of geometrical
spreading, attenuation and the increase of duration with distance due to wave propa-
302 C.-T. Dang et al.

gation and scattering. The site G(f) accounts for the effects due to local site geology
and is separated to amplification A(f) and attenuation D(f): G(f) = D(f)A(f). The mo-
tion type I(f) for acceleration is defined as I(f ʌf)2.

3.2 Linear and non-linear oscillators.

Two linear and non-linear Bouc-Wen oscillators [6] were considered here.

x&t   2]Z0 x&t   Z 2 x t  a t 

& (5)

x& t   2]Z0 x& t   Z02 D x  t   1  D  Z  t   a  t  ,

&
(6)
nd 1 nd
with Z& t  C1 x& t   C2 x& t  Z  t  Z  t   C3 x& t  Z  t 

where Ȧ0 (rad/s) is the natural angular frequency; ȗ is the damping ratio; for Bouc-
Wen behavior: Ȧ(t) is the hysteretic displacement and Į, C1, C2, C3, nd are constants;
a(t) is ground acceleration and g is gravitational acceleration. Numerical parameter
values were inspired by Kafali et Grigoriu [6]: Ȧ0 = 5.97 rad/s, ȗ = 2%, C1 = 1, C2 =
C3 = 0.5 cm, Į = 0.1, nd = 1 for the Bouc-Wen oscillator. The failure of an oscillator
was verified by comparing displacement x(t) with a displacement limit x0 i.e., D =
Xmax = max|x(t)|. The displacement limit C = x0 = 7 cm was chosen in these examples.

3.3 Design-of-Experiment of uncertain parameters

Kriging metamodels are basically constructed from experimental designs of input
parameters and response outputs. The idea behind is the forward propagation of pa-
rameter uncertainties to the outputs through the Gaussian process. For this reason,
structural properties of two oscillators mentioned earlier should be considered as un-
certain parameters in the metamodeling. On the other hand, random earthquake inten-
sity is an inevitably decisive factor to structural failure. The intensity is hence consid-
ered as another input factor in the Kriging approximation. For the practical sake of
predicting the failure probability of structures under a certain earthquake, peak ground
acceleration (PGA) is commonly adopted to graph the relation between earthquake
intensity and exceeding failure probability. However, it is argued that spectral accel-
eration at the fundamental frequency of structures (Sa,T1) can be a more appropriate
measure for representing the intensity. The metamodeling process in this study will
deal with both uncertainties from structural properties and seismic loadings.
In this study, the authors chose two significant structural parameters including nat-
ural angular frequency, Ȧ; damping ratio, ȗ of linear and nonlinear oscillators; and the
spectral acceleration at the first frequency of the oscillators, Sa,T1 as uncertain parame-
ters. The determination of their variability range with upper bounds and lower bounds
is needed to create an experimentally design space. Table 1 shows bounds and nomi-
nal values of each parameter. These bounds are carefully chosen to enable covering
the entire design space of interest. On the other hand, they should not too wide to
keep physical meaning of each parameter as well as not to change the system order. A
Use of Kriging metamodels for seismic fragility … 303

multi-dimensional design space of 100 samples was created by Latin hypercube sam-
pling with the assumed uniform distributions for three parameters. Then, a training
data set including input samples and corresponding response samples is prepared for
constructing the metamodels, which will be addressed in the next section.

Table 1. Uncertain parameters and their design space

Parameters Symbol Lower bound Nominal value Upper bound

Omega (rad/s) Ȧ 5.37 5.97 6.57
Damping ratio ȗ 0.01 0.02 0.03
Spectral acceleration (g) Sa,T1 0.1 0.6 1.1

3.4 Kriging metamodeling

Kriging metamodeling for linear and nonlinear Bouc-Wen models used the training
samples of three inputs in Table 1 and corresponding outputs. Here the authors chose
maximum displacement of the SDOF structures as an output response measure. A
training data set of input and output samples with a 100™3 size matrix is used to
build the metamodels. Leave-one-out (LOO) cross-validation [7] is adopted to meas-
ure their errors. LOO error can measure insensitivity degree of the metamodels when
one of input points of DOE is set constant. A better metamodel can show smaller
LOO error which ideally equals zero. Small LOO errors, 3.21e-4 and 4.62e-4, were
found at the metamodels of linear and nonlinear oscillators, respectively. The validat-
ed ones were then used for Monte Carlo based fragility analysis to reduce computa-
tional costs.

3.5 Seismic fragility curves

Fragility curves obtained by MC simulations using Kriging metamodels and the
existing MLE are depicted in Fig. 1. A total 1e6 simulation runs were implemented
for the MC using the metamodels while only 5e3 simulations needed for the MLE
method. Obviously, the use of the alternative approximated models for the fragility
analysis can reduce significantly computational costs.

Fig. 1. Fragility curves of linear and nonlinear Bouc-Wen oscillators

304 C.-T. Dang et al.
The result of fragility analysis performed by MC simulations is close to the one by
MLE in both linear and nonlinear response cases under earthquakes with Sa,T1 smaller
than approximately 1g (Sa,T1 < 1g). However, the higher discrepancy between two
methods is found in the nonlinear case under huge earthquakes with Sa,T1 larger than
around 1.1g (Sa,T1 > 1.1g). This indicates that the precision of the metamodels should
be concerned under huge earthquakes. There are two possibilities which can explain
for this. Firstly, increasing nonlinearity in dynamic responses under very large earth-
quakes may cause the metamodels less accurate. The Gaussian process conditional on
a constant size of the training data, i.e. a data set of size 100, may not appropriate for
all earthquake levels. The second possibility is bad extrapolation of the Kriging
metamodels for Sa,T1 larger than 1.1g. It is concluded that verification and validation
of the surrogate models in large seismic intensities plays important role in structural
risk assessment through the fragility curves.

Conclusions

This study demonstrated effective use of the Kriging metamodel in the seismic fragili-
ty analysis of the structures based on Monte Carlo simulations. The process of fragili-
ty curve derivation considered uncertainties from both the structural properties and
input earthquake ground motions. The fragility analysis using the alternative approx-
imated models can reduce significantly computational costs. However, it is cautioned
that metamodeling process of the structures must be verified and validated carefully
for different earthquake intensities. For future work, the effect of nonlinearity degree
of dynamic responses in large earthquakes on metamodel accuracy can be investigat-
ed.

Acknowledgements

This work is supported by Vietnam Ministry of Education and Training under the
research project No. B2018.DNA.01.

References
1. B. Ellingwood. Validation studies of seismic PRAs. Nuclear Engineering and Design,
123, (1990), pp. 189±196.
2. T.-P. Le, C.-T. Dang, and P. Ray. A comparative study of construction methods for seis-
mic fragility curves using numerical simulations. Mechanics and Industry, 17, (6), (2016),
p. 12.
3. Cressie, Noel. "Spatial prediction and ordinary kriging." Mathematical geology 20.4
(1988): 405-421.
4. Sacks, J., Welch, W. J., Mitchell, T. J., & Wynn, H. P. Design and analysis of computer
experiments. Statistical science, (1989): 409-423.
5. D. M. Boore. Simulation of ground motion using the stochastic method. Pure and Applied
Geophysics, 160, (3), (2003), pp. 635±676.
6. C. Kafali and M. Grigoriu. Seismic fragility analysis: Application to simple linear and non- lin-
ear systems. Earthquake Engineering & Structural Dynamics, 36, (13), (2007), pp. 1885±1900.
7. 6WRQH0µ&URVV-Validatory Choice and Assessment of Statistical PrediFWLRQV¶Journal of
the Royal Statistical Society, 36(2), (1974), pp. 111±147.
 $SSOLFDWLRQRIZHLJKWHG/DWLQK\SHUFXEHVDPSOLQJLQ
VWRFKDVWLFPRGHOOLQJRIVKHDUVWUHQJWKRI5&EHDPV

1JRF/LQK75$1DQG.KXRQJ/(1*8<(1

7HFKQLVFKH8QLYHUVLWlW'DUPVWDGW)UDQ]LVND%UDXQ6WU'DUPVWDGW*HUPDQ\
tran@massivbau.tu-darmstadt.de

8QLYHUVLW\RI7UDQVSRUW7HFKQRORJ\7ULHX.KXF+DQRL9LHWQDP
khuongln@utt.edu.vn


$EVWUDFW6WRFKDVWLFPRGHOOLQJDQGSUREDELOLVWLFDQDO\VLVRIFRQFUHWHPHPEHUV
DUH YHU\ WLPHGHPDQGLQJ HVSHFLDOO\ IRU QRQOLQHDU ILQLWH HOHPHQW DQDO\VLV RI
FRQFUHWH VWUXFWXUHV 7KLV SDSHU SUHVHQWV DQ DSSOLFDWLRQ RI ZHLJKWHG /DWLQ
K\SHUFXEH VDPSOLQJ PHWKRG LQ VWRFKDVWLF PRGHOOLQJ RI VKHDU VWUHQJWK RI
UHLQIRUFHG FRQFUHWH EHDPV ZKLFK WDNHV LQWR DFFRXQW WKH VHQVLWLYLW\ IDFWRUV
UHJDUGLQJWKHEHKDYLRXURIWKHFRQVLGHUHGV\VWHP7KHUHVXOWVVKRZZLWKRQO\D
IHZVLPXODWLRQVWKHVWDWLVWLFDOYDOXHVRIWKHORDGEHDULQJFDSDFLWLHVRIFRQFUHWH
VWUXFWXUHVFDQEHGHWHUPLQHGZLWKKLJKDFFXUDF\LIZHLJKWHG/DWLQK\SHUFXEH
VDPSOLQJLVXVHG

.H\ZRUGVZHLJKWHG/DWLQK\SHUFXEHVDPSOLQJVWRFKDVWLFPRGHOOLQJ

 ,QWURGXFWLRQ

7KH VDIHW\ DVVHVVPHQW RI UHLQIRUFHG FRQFUHWH 5& VWUXFWXUHV UHJDUGLQJ WKHLU ORDG
EHDULQJ FDSDFLW\ LV RI LQFUHDVLQJ LQWHUHVW HVSHFLDOO\ IRU H[LVWLQJ VWUXFWXUHV )RU DV
VHVVPHQWRIVKHDUVWUHQJWKRI5&PHPEHUVfib0RGHO&RGH>@LQWURGXFHVWKH
OHYHORIDSSUR[LPDWLRQ,9UHJDUGLQJVFLHQWLILFDSSURDFKHVLQZKLFKFRPSOH[QXPHUL
FDOPRGHOVDUHDOORZHGWREHXVHG+RZHYHUQXPHULFDODQDO\VHVRIVXFKV\VWHPVZLWK
QRQOLQHDUEHKDYLRXU DUHYHU\WLPHGHPDQGLQJ HVSHFLDOO\IRUSUREDELOLVWLF PRGHOOLQJ
ZLWK/DWLQK\SHUFXEHVDPSOLQJ/+6VLQFHDODUJHQXPEHURIVLPXODWLRQVDUHRIWHQ
UHTXLUHG 7KH EDVLF LGHD RI D JRRG /+6 LV WKDW WKH SUREDELOLW\ VSDFH LV GLYLGHG LQWR
DSSUR[LPDWHO\HTXDOSUREDELOLW\VXEVSDFHVUHSUHVHQWHGE\VDPSOHSRLQWV7RVDYHWKH
FRPSXWDWLRQWLPHWKHQXPEHURIVLPXODWLRQVVKRXOGEHUHGXFHGPHDQLQJWKDWWKHVL]H
RIWKHDSSOLHG/+6LVUHGXFHG
,WVKRXOGEHQRWHGWKDWVXFK/+6VDUHLGHDOIRUV\VWHPVZKLFKKDYHWKHEDVLFYDUL
DEOHVZLWKHTXDOVHQVLWLYLW\IDFWRUV,QWKHDQDO\VHVRIUHDOVWUXFWXUHVLWLVRIWHQWKDW
RQO\VRPHYDULDEOHVGRPLQDWHWKHVFDWWHURIWKHV\VWHPEHKDYLRXU7KHUHIRUHWKHXVH
RIDQLGHDO/+6IRUWKHVHV\VWHPVPD\OHDGWRLQDFFXUDWHUHVXOWVHVSHFLDOO\ZKHQWKH
VL]HRI/+6LVVPDOO)RUH[DPSOHVFRQVLGHULQJDV\VWHPZLWKnSDUDPHWHUVLQZKLFK
RQO\ WZR SDUDPHWHUV GRPLQDWH WKH WRWDO VFDWWHU RI V\VWHP EHKDYLRXU D /+6 ZLWK 

© Springer Nature Singapore Pte Ltd. 2020 305

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_46
306 N. L. Tran and K. Le-Nguyen

VLPXODWLRQVVDPSOHVLVXVHGIRUWKHVWRFKDVWLFPRGHOOLQJ6XSSRVLQJWKDWWKHUH

VSRQVHIXQFWLRQRIWKHV\VWHPLVF = Xi + XjZKHUHXiDQGXjDUHWKHWZRGRPLQDWLQJ
EDVLFYDULDEOHVKDYLQJWKHVDPHQRUPDOGLVWULEXWLRQIXQFWLRQZLWKDPHDQYDOXHRI
DQG VWDQGDUG GHYLDWLRQ RI  WKH VWDQGDUG GHYLDWLRQ RI F LV WKXV ξʹ ൎ ͳǤͶͳͶ 8VLQJ
RSWLPDODQGDOPRVWRSWLPDO/+6VGHYHORSHGE\YDQ'DPHWDO>@ZKLFKDUHDYDLOD
EOHRQOLQHLQWKHZHEVLWHhttps://spacefillingdesign.nlWKHHPSLULFDOVWDQGDUGGHYLDWLRQ
sFRIWKHV\VWHPEHKDYLRXULVFDOFXODWHGDFFRUGLQJWRGLIIHUHQWQXPEHUQRIEDVLFYDUL
DEOHVDQGLVSUHVHQWHGLQ)LJ,WFDQEHVHHQWKDWWKRXJKRSWLPDO/+6VDUHXVHGWKH
FDOFXODWHGsFVWURQJO\VFDWWHUVHJE\V\VWHPZLWKEDVLFYDULDEOHVWKHFRHIILFLHQWRI
YDULDWLRQ RI WKH HPSLULFDO VWDQGDUG GHYLDWLRQ LV DSSUR[LPDWHO\  ,I WKH RSWLPDO
/+6IRURQO\WZRGRPLQDWLQJEDVLFYDULDEOHVLVXVHGWKHIDVWH[DFWYDOXHRIsFFDQEH
REWDLQHGsF ZLWKRXWVFDWWHU7RRYHUWKHVKRUWFRPLQJVRIWKHXVHRIDFRP
PRQ/+6ZHLJKWHG/+6WDNLQJLQWRDFFRXQWVHQVLWLYLW\IDFWRUVFDQEHXVHGZDVILUVW
LQWURGXFHGE\7UDQ>@ZLWKDPRUHGHWDLOHGIRUPXODWLRQSUHVHQWHGLQ>@


)LJ([DPSOHRIWKHLQDFFXUDWHVFDWWHUsFZKHQXVLQJRSWLPDO/+6V

7KLVSDSHUSUHVHQWVDQDSSOLFDWLRQRIZHLJKWHG/+6VLQVWRFKDVWLFPRGHOOLQJDQG
DQDO\VLV RI VKHDU FDSDFLW\ RI 5& PHPEHUV ZLWKRXW VKHDU UHLQIRUFHPHQW WR VKRZ WKH
HIILFLHQF\RIWKLVPHWKRG

 :HLJKWHG/DWLQK\SHUFXEHVDPSOLQJ

$ ZHLJKWHG /DWLQ K\SHUFXEH VDPSOLQJ LV D /+6 JHQHUDWHG ZLWK WKH XVH RI
ZHLJKWLQJ IDFWRUV IRU GLPHQVLRQV RI WKH SUREDELOLW\ VSDFH 7KHVH ZHLJKWLQJ IDFWRUV
UHODWHWRWKHVHQVLWLYLW\IDFWRUVRIEDVLFYDULDEOHV$VVXPLQJWKDWFLVDIXQFWLRQRIn
XQFRUUHODWHG EDVLF YDULDEOHV X DQG WKH VFDWWHU DURXQG WKH PHDQ YDOXH LV FRQVLGHUHG
WKHVHQVLWLYLW\IDFWRUFRUUHVSRQGLQJWRWKHEDVLFYDULDEOHkFDQEHFDOFXODWHGDFFRUGLQJ
WR(T7KHJUHDWHUWKHVHQVLWLYLW\IDFWRU DkWKHVWURQJHUWKHLQIOXHQFHRIWKHEDVLF
YDULDEOHXkRQWKHYDULDQFHRIF,WVKRXOGEHQRWHGWKDW(TGRHVQRWFRQVLGHUWKH
W\SHVRIWKHSUREDELOLW\GLVWULEXWLRQIXQFWLRQRIEDVLFYDULDEOHV6LQFHRQO\WKHVFDWWHU
DURXQGWKHPHDQYDOXHLVRILQWHUHVWKHUH(TLVDYHU\JRRGHVWLPDWLRQ
Application of weighted Latin hypercube sampling … 307


§ wF · 
¨ ¸ V Xk
Dk © wX k ¹  

n § wF ·

¦¨ ¸ V Xi
i  © wX i ¹

,QD/+6HDFKGLPHQVLRQRIWKHSUREDELOLW\VSDFHUHSUHVHQWVDEDVLFYDULDEOH7R
WDNH LQWR DFFRXQW WKH VHQVLWLYLW\ IDFWRUV FRUUHVSRQGLQJ WR WKH EDVLF YDULDEOHV DOO WKH
GLPHQVLRQVVKRXOGEHZHLJKWHGZLWKDSSURSULDWHIDFWRUV7KHVXEVSDFHRISUREDELOLW\
UHSUHVHQWHGE\RQHSRLQWVDPSOHFDQEHGHVFULEHGE\WKHGLVWDQFHVIURPWKLVSRLQWWR
WKHQHLJKERXUSRLQWV(DFKFRRUGLQDWHFKDQJHRIDSRLQWSUHVHQWVWKHGHJUHHRILQIOX
HQFHRIWKHFRUUHVSRQGLQJGLPHQVLRQRUWKHFRUUHVSRQGLQJEDVLFYDULDEOH7KHUHIRUH
WKHZHLJKWLQJIDFWRUIRUHDFKGLPHQVLRQkLQWKHFDOFXODWLRQRIWKHGLVWDQFHEHWZHHQ
WZRSRLQWVRID/+6LVWKHVDPHDVWKHVHQVLWLYLW\IDFWRUDkFDOFXODWHGE\(T7KLV
DOORZVXVWRFDOFXODWHWKHGLVWDQFHEHWZHHQWZRSRLQWVLQDZHLJKWHG/+6DVIROORZV
m 
 dij 
¦ xik  x jk
k 
 D k  

7R LPSURYH WKH TXDOLW\ RI DZHLJKWHG /+6 WKH maximinIXQFWLRQ)pGHILQHGE\

0RUULV 0LWFKHOO>@FDQEHXVHG$GHFLVLRQWRH[FKDQJHFRRUGLQDWHVZLOOEHPDGH
LIWKLVOHDGVWRDVPDOOHUYDOXHRI)pVHH(T
 p
ª m  m pº
 )p « ¦ ¦ d ij »  
¬ i  j j  ¼
+HUHpLVDSRVLWLYHLQWHJHUDQGmLVWKHQXPEHURISRLQWVLQWKHGHVLJQ,QWKLV
VWXG\ p LV FKRVHQ DV  WKH GLVWDQFH dij FDOFXODWHG DFFRUGLQJ WR (T  7KH
RSWLPLVDWLRQLVVWRSSHGLIWKHUHLVQRSRVVLEOHFRRUGLQDWHH[FKDQJH

 (VWLPDWLQJVHQVLWLYLW\IDFWRUV

,Q PRVW FDVHVRI DQDO\VLV VHQVLWLYLW\ IDFWRUV RIEDVLFYDULDEOHV DUHXQNQRZQ 7KHUH

IRUHWKH\QHHGWREHHVWLPDWHGRQWKHEDVLVRIVRPHDVVXPSWLRQVRUFOXHVUHJDUGLQJ
WKHPHFKDQLFDOEHKDYLRXURIWKHFRQVLGHUHGVWUXFWXUHV)RULQVWDQFHWKHEHQGLQJFD
SDFLW\RI5&PHPEHUVVWURQJO\GHSHQGVRQWKHFRPSUHVVLYHVWUHQJWKRIFRQFUHWHDQG
WKH WHQVLOH VWUHQJWK RI UHLQIRUFHPHQW +HQFH WKHVH PDWHULDO VWUHQJWKV DUH GHFLVLYH
SDUDPHWHUVFRPSDUHGWRRWKHUVVXFKDVJHRPHWULHVRUHODVWLFPRGXOXVRIFRQFUHWHDQG
UHLQIRUFHPHQW 7KLV PHDQV WKDW RQO\ WKH GRPLQDWLQJ YDULDEOHV QHHG WR EH WDNHQ LQWR
DFFRXQWZLWKKLJKDFFXUDF\6XSSRVLQJWKDWZHKDYHDOLQHDUIXQFWLRQRIXDVIROORZV
n
 F X  ¦ ai X i  
i 
308 N. L. Tran and K. Le-Nguyen

ZLWK Xi EHLQJ D QRUPDOO\ GLVWULEXWHG UDQGRP YDULDEOH ZLWK D PHDQ YDOXH RI  DQG D
VWDQGDUGGHYLDWLRQRIaiDUHFRQVWDQWVZLWKσ ܽ௜ଶ ൌ ͳ)URP(TLWFDQEHVHHQ
WKDWaiLVWKHVDPHDVWKHVHQVLWLYLW\IDFWRUFRUUHVSRQGLQJWRWKHEDVLFYDULDEOHXi7KH
IXQFWLRQFXKDVDPHDQYDOXHRIσ ܽ௜ DQGDVWDQGDUGGHYLDWLRQRI,WLVDVVXPHG
ଶ
WKDWܽ௜ାଵ ൌ ܿ ή ܽ௜ଶ ZLWKcLVDFRQVWDQW,IczLVJLYHQWKHILUVWFRQVWDQWa1LVGHWHU
PLQHGDFFRUGLQJWRWKHIROORZLQJHTXDWLRQ


 c  
a
  cn
7KHPHDQYDOXHRIFXLVFDOFXODWHGDVIROORZV
n

 Fm
 c  c 
˜
  
  cn   c

,W LV QRWHG WKDW a1 DV\PSWRWLFDOO\ UHDFKHV ξͳ െ ܿ ZKHQ n LQFUHDVHV WR LQILQLW\
:KHQ c   ZH KDYH ܽଵ ൌ ܽଶ ൌ ‫ ڮ‬ൌ ܽ௡ ൌ ͳȀξ݊ ,W FDQ EH VHHQ WKDW RQO\ D IHZ
EDVLFYDULDEOHVJRYHUQWKHYDULDQFHRIWKHIXQFWLRQ6XSSRVLQJWKDWqLVWKHPLQLPXP
WRWDOVHQVLWLYLW\GRPLQDWHGE\WKHILUVWkYDULDEOHVkFDQEHGHWHUPLQHGDVIROORZV

 k t LQW ¨

§ OQ   q  qc n  ·¸  
¨ OQ c ¸
© ¹
7KHIXQFWLRQFXFDQEHXVHGDVDWHVWPRGHOWRHVWLPDWHWKHXQNQRZQVHQVLWLYLW\
IDFWRUVDQGWRFKHFNWKHTXDOLW\RIDZHLJKWHG/+6,IWKHVHQVLWLYLW\IDFWRUVDUHQRW
NQRZQWKHODUJHVWVHQVLWLYLW\IDFWRUFDQEHDVVXPHGDQGWKHRWKHUVHQVLWLYLW\IDFWRUV
FDQEHHVWLPDWHGXVLQJDVFDOLQJIDFWRUc

 $SSOLFDWLRQWRVKHDUFDSDFLW\FDOFXODWLRQ

 6FDWWHURIVKHDUFDSDFLW\RI5&EHDPVGXHWRSDUDPHWHUXQFHUWDLQW\

6KHDUFDSDFLW\RIVLPSO\VXSSRUWHGD5&EHDPZLWKRXWVKHDUUHLQIRUFHPHQWXQGHU
FRQFHQWUDWHGORDGLVQXPHULFDOO\GHWHUPLQHG7KH5&EHDPKDVDVKHDUVSDQWRGHSWK
UDWLRa / d FRQFUHWHFRPSUHVVLYHVWUHQJWKfcm 03DZLWKPD[LPXPDJJUHJDWH
VL]H ag RI  PP 7KH PHDQ YDOXHV RI RWKHU FRQFUHWH SURSHUWLHV DUH HVWLPDWHG RQ WKH
EDVLV RI fcm DQG SUHVHQWHG WRJHWKHU ZLWK WKH HVWLPDWLRQ XQFHUWDLQWLHV LQ 7DEOH  7KH
DSSOLHGHVWLPDWLQJIRUPXODHDUHWDNHQIURP7UDQ *UDXEQHU>@
)URP>@LWLVNQRZQWKDWIRU5&PHPEHUVZLWKRXWVKHDUUHLQIRUFHPHQWWKHWHQVLOH
VWUHQJWKfctDQGWKHIUDFWXUHHQHUJ\GFRIFRQFUHWHDUHWKHWZRGHFLVLYHSDUDPHWHUVRI
VKHDUFDSDFLW\$VVXPLQJDGHFUHDVLQJVHQVLWLYLW\IDFWRUZLWKDFRQVWDQWVFDOLQJIDFWRU
ଶ
c KHUH ߙ௜ାଵ ൌ ܿ ή ߙ௜ଶ  DQG VXSSRVLQJ WKDW WKHVH WZR EDVLF YDULDEOHV GRPLQDWH DERXW
Application of weighted Latin hypercube sampling … 309

RIWKHWRWDOVKHDUVWUHQJWKVFDWWHUWKHVFDOLQJIDFWRUcFDQWKXVEHVHOHFWHGDV
IRUFUHDWLQJDZHLJKWHG/+6VHH(TIRUHVWLPDWLQJWKHILUVWVHQVLWLYLW\IDFWRU
7DEOH%DVLFYDULDEOHVIRUVKHDUFDOFXODWLRQ
nSim 3') 8QLW 0HDQ &R9
&RQFUHWHWHQVLOHVWUHQJWK fct /1 03D  
&RQFUHWHIUDFWXUHHQHUJ\ GF /1 1P  
(ODVWLFPRGXOXVRIFRQFUHWH Ec /1 03D  
(ODVWLFPRGXOXVRIUHLQIRUFHPHQW Es  /1 03D  
5HLQIRUFHPHQWUDWLRUs 1   
:LGWKRIEHDP b 1 PP  
'HSWKRIEHDPd 1 PP  

7RPRGHOFRQFUHWHEHKDYLRXUWKHQRQOLQHDU)(SURJUDP$7(1$ZLWKWKHIUDFWXUH
SODVWLFFRQVWLWXWLYHPRGHOCC3DNonLinCementitious2LVXVHG)RUDGHWDLOHGGHVFULS
WLRQRIWKHFRQFUHWHPRGHOLWLVUHIHUUHGWRýHUYHQNDHWDO>@'HWDLOVRQWKHVWRFKDV
WLF PRGHOOLQJ FDQ EH IRXQG LQ >@ %RWK DOPRVW RSWLPDO /+6V WDNHQ IURP ZHEVLWH
https://spacefillingdesign.nlDQGZHLJKWHG/+6VJHQHUDWHGDFFRUGLQJWRWKHDOJRULWKP
SUHVHQWHGLQ6HFWLRQDUHDSSOLHGIRUWKHVLPXODWLRQV7KHFDOFXODWHGUHVXOWVDUHSUH
VHQWHG LQ 7DEOH  ,W FDQ EH VHHQ WKDW ZHLJKWHG /+6V FRQVLVWHQWO\ VKRZ YHU\ KLJK
SUHGLFWLRQDFFXUDF\LQWHUPVRIPHDQYDOXHDQGYDULDWLRQFRHIILFLHQWHYHQZLWKIHZ
VLPXODWLRQV
7DEOH6LPXODWLRQUHVXOWV
RSW/+6 ZHLJKWHG/+6
nSim VmN1 CoVVSim Vm N1 CoV
    
    
    
    
1RWHWKHRUHWLFDOtrueYDOXHVREWDLQHGIURPVLPXODWLRQVVm N1CoV 
 6FDWWHURIVKHDUFDSDFLW\RI5&EHDPVGXHWRVSDWLDOYDULDELOLW\RI
FRQFUHWHSURSHUWLHV
7KHVKHDUFDSDFLW\RI5&PHPEHUVGHSHQGVVWURQJO\RQWKHVSDWLDOYDULDELOLW\RIWKH
PDWHULDO SURSHUWLHV &RQVLGHULQJ WKH VSDWLDO YDULDELOLW\ RI FRQFUHWH SURSHUWLHV IRU WKH
EHDPGHVFULEHGLQ6HFWLRQZLWKWKHVWDWLVWLFSURSHUWLHVGHVFULEHGLQ>@WKHFUDFN
GHYHORSPHQWIURPWKH)(QRQOLQHDUVLPXODWLRQDSSHDUVPRUHQDWXUDOWKDQIURPDGH
WHUPLQLVWLFDQDO\VLVVHH)LJ$IWHUWKHHLJHQYDOXHGHFRPSRVLWLRQRIWKHFRYDULDQFH
PDWUL[ IRU WKLV EHDP VHH WKH GHWDLOV RI WKLV PHWKRG LQ >@ WKH VFDOLQJ IDFWRU c LV
HVWLPDWHGDV,WLVQRWHGWKDWDPRQJUDQGRPYDULDEOHVLQWKLVFDVHWKHILUVW
YDULDEOHVGRPLQDWHRIWKHWRWDOGLVSHUVLRQVHHDOVR(TIRUGHWHUPLQLQJWKH
QXPEHURIGRPLQDWLQJYDULDEOHV7KHUHIRUHZHRQO\QHHGWRJHQHUDWHDZHLJKWHG/+6
IRUWKHVHUDQGRPYDULDEOHVDQGFDQXVHDFRPPRQ/+6IRUWKHRWKHUYDULDEOHV7KH
310 N. L. Tran and K. Le-Nguyen

UHVXOWVRIVWRFKDVWLFDQDO\VLVZLWKVLPXODWLRQVKRZDPHDQYDOXHRIVKHDUFDSDFLW\
RI  N1 DQG D FRHIILFLHQW RI YDULDWLRQ RI  'XH WR VSDWLDO YDULDELOLW\ RI
FRQFUHWH SURSHUWLHV WKH FDOFXODWHG PHDQ YDOXH RI VKHDU VWUHQJWK LV UHGXFHG WR
DSSUR[LPDWHO\WLPHVRIWKHPHDQYDOXHUHVXOWHGIURPWKHGHWHUPLQLVWLFDQDO\VLV


)LJ6KHDUIDLOXUHVRIWKH5&EHDPDVUHVXOWVRIVWRFKDVWLF)(VLPXODWLRQV

 &RQFOXVLRQV

7KHQHZO\GHYHORSHGZHLJKWHG/+6WDNHVLQWRDFFRXQWWKHVHQVLWLYLW\IDFWRUVRIEDVLF
YDULDEOHVDQGRIIHUVDYHU\JRRGDQGVWDEOHTXDOLW\RIVWRFKDVWLFVLPXODWLRQLQWHUPVRI
VWDWLVWLFDOUHVXOWVHYHQZLWKIHZVLPXODWLRQV7KHH[DPSOHRID5&EHDPXQGHUVKHDU
ORDGVKRZHGWKDWDQH[FHOOHQWUHVXOWFDQEHREWDLQHGZLWKRQO\VLPXODWLRQV

5HIHUHQFHV
 fib0RGHO&RGHfib0RGHO&RGHIRU&RQFUHWH6WUXFWXUHV
 +XVVODJH%5HQQHQ*YDQ'DP(GHQ+HUWRJ'6SDFHILOOLQJ/DWLQK\SHUFXEHGH
VLJQVIRUFRPSXWHUH[SHULPHQWVOptim Eng±
 7UDQ 1/ 6KHDU VWUHQJWK RI VOHQGHU UHLQIRUFHG FRQFUHWH PHPEHUV ZLWKRXW VKHDU UHLQ
IRUFHPHQW±$PHFKDQLFDOPRGHOHabilitation, 7HFKQLVFKH8QLYHUVLWlW'DUPVWDGW
 7UDQ 1/ *UDXEQHU &$ 5RPEDFK * $Q HIILFLHQW /DWLQ K\SHUFXEH VDPSOLQJ IRU
SUREDELOLVWLFQRQOLQHDUILQLWHHOHPHQWDQDO\VLVRIUHLQIRUFHGFRQFUHWHVWUXFWXUHVThe 7th In-
ternational Conference on Structural Engineering, Mechanics and Computation 6(0&
6HSWHPEHU&DSH7RZHU6RXWK$IULFD
 0RUULV0'DQG0LWFKHOO7-([SORUDWRU\'HVLJQVIRU&RPSXWHU([SHULPHQWVJ. Stat-
ist. Plann. Inference
 7UDQ1/*UDXEQHU&$8QFHUWDLQWLHVRIFRQFUHWHSDUDPHWHUVLQVKHDUFDSDFLW\FDOFXOD
WLRQ RI 5& PHPEHUV ZLWKRXW VKHDU UHLQIRUFHPHQW ,Q 6WUDXVV $ HW DO (GV ([WHQGHG
$EVWUDFWV RI WKH WK ,QWHUQDWLRQDO 3UREDELOLVWLF :RUNVKRS  LQ 9LHQQD  ±  6HS
WHPEHU, Beton- und Stahlbetonbau113, Issue S2
 ýHUYHQND 9 -HQGHOH / ýHUYHQND - $7(1$ 3URJUDP 'RFXPHQWDWLRQ ± 7KHRU\ DQG
,QSXW)LOH)RUPDWýHUYHQNDFRPSDQ\3UDJXH
 Predicting onset and orientation of localisation bands
using a cohesive-frictional model

Linh A. Le1, Giang D. Nguyen1, Ha H. Bui2

1School of Civil, Environmental and Mining Engineering, the University of Adelaide, Australia
2
Department of Civil Engineering, Monash University, Australia

g.nguyen@adelaide.edu.au

Abstract. In this study, a localisation band idealised as zero-thickness surface

and represented by a cohesive-frictional model is directly incorporated into the
structure of a new constitutive modelling approach and is activated once a stress-
based condition is met. The embedded cohesive-frictional model provides a nat-
ural way to detect both the onset and orientation of localisation bands. This is
different from existing continuum approaches based on the loss of positiveness
of the determinant of the acoustic tensor. The formulation in this approach also
creates a strong link between quantities describing the localisation band in the
model and experiments, facilitating the calibration of model parameters.

Keywords: Bifurcation analysis, Geomaterials, Localised failure, cohesive-fric-

tional models

1 Introduction
Experimental studies on geomaterials [1±3] have indicated the key role that localisation
band plays in governing the post-peak behaviour and amount of energy dissipated dur-
ing the course of loading. Therefore, it is very important for any constitutive model
developed for geomaterials to be able to predict the onset and orientation of the locali-
sation band. Over the past few decades, several studies have been focusing on bifurca-
tion criteria to identify the commencement of the localisation band in geomaterials.
Developed from pioneering work by Rudnicki and Rice [4], acoustic tensor is one of
the most popular tools used to predict the bifurcation point when localisation into a
planar localisation band begins. Following this approach, the localisation starts when
the minimum determinant of the acoustic tensor ceases to be positive; the angle associ-
ated with this minimum determinant is usually selected as localisation band orientation.
This criterion has been adopted and given reasonable results in several studies [5±7].
This study proposes a new approach for predicting the onset and orientation of lo-
calisation by integrating localisation failure mechanism in the structure of constitutive
models, following the double-scale approach [8,9]. The behaviour of the band is gov-
erned by a cohesive-frictional model [10,11] whose quantities can be calibrated from
experiments involving localisation, while its activation is obtained from a stress-based

© Springer Nature Singapore Pte Ltd. 2020 311

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_47
312 L. A. Le et al.

condition. The proposed approach provides a natural way to predict the onset and ori-
entation of the localisation band, while also allowing post-localisation analysis to be
carried out.

2 Model description
2.1 Embedding localisation band in continuum models

Since localised failure is the mechanism governing material behaviour, the localisation
band should be included as an intrinsic part of a model. The key idea here is to connect
the localisation band as well as the surrounding bulk material to the overall responses
of the material within a rigorous framework. The constitutive model is thus constructed
from a material volume ȳ having an inner localised band ȳ୧ and an outer bulk material
ȳ୭ as shown in Fig. 1. The localisation band is represented by its area Ȟ୧ and thickness
݄ in association with a normal vector defining the band orientation ‫ܖ‬. The volume-
averaged stress and strain vectors of the element are expressed by ો and ઽ while ો୭ and
ઽ୭ stand for the stress and strain vectors of the outer bulk material. The stress and strain
vectors of the inner localisation zone are denoted as ૗ and ૆.

Fig. 1. The conceptual model based on the failure mechanism at mesoscale (a) and yield sur-
face with different parameters sets (b)
The strain rate of the localisation band and the averaged strain rate of the RVE can
be expressed as [12]:
ଵ ଵ
૆ሶ ൌ ઽሶ ୭ ൅ ‫ܝܖ‬ሶ and ઽሶ ൌ ߟ૆ሶ ൅ ሺͳ െ ߟሻઽሶ ୭ ൌ ઽሶ ୭ ൅ ‫ܝܖ‬ሶ (1ab)
௛ ு
where ‫ܝ‬ሶ is the rate of relative displacements between two sides of the localisation band
in the global coordinate system; ߟ ൌ ȳ୧ Ȁȳ ൌ ݄Ȁ‫ ܪ‬is the volume fraction of the locali-
sation band with respect to the volume element and ‫ ܪ‬ൌ ȳȀȞ୧ represents the
characteristic length of the RVE.
The kinematic relationships, presented in Eq. (1a), help the model capture the for-
mation and development of localisation band naturally. Specifically, right before local-
isation initiates, the strain is homogenous in the whole RVE domain, including the po-
tential localisation band and the surrounding bulk material (i.e., ‫ ܝ‬ൌ ૙ ՜ ઽ ൌ ૆ ൌ ઽ୭ ).
Once localisation band associated with an orientation ‫ ܖ‬appears, the strains outside and
inside the band are different and contribute to the macro strain (Eq. 1b). This helps the
model obtain a smooth transition from a homogenous state to localised phase and keeps
track of the high deformation inside the localisation band during its development.
Predicting onset and orientation of localisation bands … 313

By using the Hill-Mandel condition [13] for the conservation of work, the traction
continuity at the surface of localisation band can be obtained as (see [10,11]):ો் ‫ ܖ‬ൌ
૗் ‫ ܖ‬ൌ ‫ ܜ‬where ‫ ܜ‬is the traction acting on the band. The traction-separation relationship
and its progressive development, in principle, can be written in a generalised rate form
as ‫ܜ‬ሶ ൌ ۹ ୲ ‫ܝ‬ሶ , with ۹ ୲ being the tangent stiffness of the band written in the global coor-
dinate system. In this study, the behaviour inside the localisation band is governed by
our newly developed cohesive-frictional model [10,11] which is based on damage me-
chanics and plasticity theory; its yield-failure surface written as:
‫ ݕ‬ൌ ‫ݐ‬௦ଶ െ ሾሺͳ െ ‫ܦ‬ሻߤ଴ଶ ൅ ‫ߤܦ‬ଶ ሿሾ‫ݐ‬௡ െ ሺͳ െ ‫ܦ‬ሻ݂௧ ሿଶ ൅ ݂݉௖ ሺͳ െ ‫ܦ‬ሻሾ‫ݐ‬௡ െ ሺͳ െ ‫ܦ‬ሻ݂௧ ሿ ൑ Ͳ (2)
‫ݐ‬௦
where and ‫ݐ‬௡ are shear and normal tractions acting on the localisation plane in the
local coordinate system; ݂௧ is the tensile strength of the material; ݂௖ is its compressive
strength; ‫ ܦ‬is damage variable (Ͳ ൑ ‫ ܦ‬൑ ͳ);݉ǡ ߤƒ†ߤ଴ are the parameters control-
ling the shape of the initial yield and final failure surfaces. The yield-failure surface is
illustrated in Fig. 1b with respect to different sets of parameters. From the traction con-
tinuity condition, the kinematic enhancement in Eq. (1) and the traction-separation re-
lationship, the constitutive relationship of the RVE can be derived as [8,9]:
ଵ ଵ ିଵ
ોሶ ൌ ‫܉‬଴ ቂઽሶ െ ‫ܝܖ‬ሶ ቃ and ‫ܝ‬ሶ ൌ ቂ ‫܉ ்ܖ‬଴ ‫ ܖ‬൅ ۹ ୲ ቃ ‫܉ ்ܖ‬଴ ઽሶ (3)
ு ு

2.2 Identifying the initiation and orientation of localisation band

From the model formulation presented above, it can be seen that the localisation band
is directly incorporated into the constitutive model with its orientation and behaviour.
The use of a cohesive-frictional model to describe the behaviour inside this localisation
zone gives us a natural way to not only detect the onset and orientation of the band but
also to track its evolution during the course of loading.
Because the yielding criterion is defined in the localisation band as presented in Eq.
(2), the initiation of localisation and its orientation can be determined using this crite-
rion. Specifically, for a given stress state, all possible orientations are scanned to detect
the onset of localisation. For each potential plane, the normal and shear tractions in the
local coordinate system of the band (Fig. 2) are calculated as:
ଶ
‫ݐ‬௡ ൌ ݊௜ ߪ௜௝ ݊௝ ƒ†‫ݐ‬௦ ൌ ටฮߪ௜௝ ݊௝ ฮ െ ‫ݐ‬௡ ଶ (4)
where ݊௜ is the normal vector of the potential plane, and ߪ௜௝ is the stress state of the
material in index notation form. A localisation band will initiate if there exists a traction
vector ሺ‫ݐ‬௡ ‫ כ‬ǡ ‫ݐ‬௦ ‫ כ‬ሻ that maximises and violates the initial yield condition in Eq. (2), as:
‫ݕ‬ሺ‫ݐ‬௡ ‫ כ‬ǡ ‫ݐ‬௦ ‫ כ‬ሻ ൌ ƒš‫ ܖ׊‬ሼ‫ݕ‬ሺ‫ݐ‬௡  ǡ ‫ݐ‬௦  ǡ ‫ ܦ‬ൌ Ͳሻሽ ൒ Ͳ (5)
The orientation corresponding to this traction vector is the band orientation at that
material point and it is used for the rest of the calculation. This procedure to determine
the localisation onset is illustrated in Fig. 2 where three orientations are given as exam-
ples of the scanning process described above. The corresponding tractions (i.e., ‫ݐ‬௡  ǡ ‫ݐ‬௦  )
acting on these planes are plotted in traction space with respect to yield surface. In this
example, because the traction on plane 3, associated with normal vector ‫ܖ‬ଷ , lies on the
314 L. A. Le et al.

yield surface and thus satisfies the condition in Eq. (5), it marks the onset of localisation
and ‫ܖ‬ଷ is considered as the orientation of the localisation band.

Fig. 2. An illustration of identifying the initiation and orientation of localisation band

3 Numerical example
In this section, the capability of the proposed approach in identifying the onset and
orientation of localisation band is illustrated using the triaxial test results of Bentheim
sandstone conducted by Klein et al. [14].
3.1 Calibration of parameters related to the onset of localisation band
The procedure and illustration, described above, show that the onset and orientation of
localisation band totally depend on the yield-failure surface formulated in Eq. (2). As
also seen from the formulation, the cohesive-frictional model, including this yield func-
tion, is defined and directly linked with behaviour inside the localisation band. As a
result, parameters related to the shape of the initial yield surface (i.e., ߤ଴ and ݉) can be
calibrated directly from an experimental dataset of localisation onset.

(a) (b)
Fig. 3. Illustration of yield surface calibration: (a) experimental data in ‫ ݌‬െ ‫ ݍ‬plane and (b)
converted traction states and calibrated yield surface by the model
As the majority of the yield surface is in compression, the ideal dataset should be
obtained from triaxial compression tests, in which the stress states at yielding are rec-
orded along with the failure patterns. Specifically, from triaxial test results conducted
by Klein et al. [14], the differential stress ‫ ݍ‬ൌ ȁߪଵ െ ߪଷ ȁ and the hydrostatic stress ‫ ݌‬ൌ
ȁఙభ ାଶఙయ ȁ
, where localisation initiates, can be plotted alongside their failure plane orien-
ଷ
tation as shown in Fig. 3a. In association with these orientations of failure, stress states
Predicting onset and orientation of localisation bands … 315

are converted to stresses (tractions) in the local coordinate system of the localisation
band as illustrated in Fig. 3b. The two parameters ߤ଴ and ݉ are then calibrated to fit
the experimental data in the local coordinate system. With tensile strength ݂௧ ൌ ͳƒ
and compressive strength ݂௖ ൌ ͸Ͳƒ, the calibrated initial yield surface is illustrated
in Fig. 3b with two parameters: ߤ଴ ൌ ͲǤ͵ͷ and ݉ ൌ ͳǤͳ.

3.2 Prediction of localisation onset and orientation

The proposed model is used to predict the onset and orientation of localisation band in
triaxial compression test under three different confining pressure ߪଷ ൌ
Ͳǡ െͳͲǡ െ͵Ͳƒ. The stress states associated with the localisation onset, presented in
Fig. 4a, show a fair agreement between the predictions by the model and experimental
measurements.

Fig. 4. Numerical predictions: (a) stress states at the onset of localisation and (b-d) normalised
values of yield function in different failure planes alongside predicted failure orientation
Furthermore, values of normalised yield function with respect to different values of
potential orientation (i.e., ߠ ൌ Ͳ଴ ̱ͻͲ଴ ), plotted in Fig. 4b-d, show that the localisation
occurs when the maximum value of yield function reaches zero. The orientations of
localisation band at these points, also plotted in the figure, are comparable with those
observed in experiments. It is seen that with the increase of confining pressure (i.e.,
ȁߪଷ ȁ ൌ Ͳ̱͵Ͳƒ), the failure orientation decreases (from ͹ʹ଴ to ͸͵଴ ), as also ob-
served in experiments [1±3]. All these features are captured at constitutive level, thanks
to the inclusion of localisation band inside the constitutive model.
316 L. A. Le et al.

4 Conclusion
In this study, a cohesive-frictional model is used to predict both onset and orientation
of the localisation band. Since quantities in the cohesive-frictional model are defined
inside the localisation plane, the calibration of parameters becomes straightforward and
the model yields fairly good results in comparison with the experimental counterparts.
Further work is in progress to capture the full range of bifurcation, from localisation to
diffuse failure mode, to cover a wider scope in modelling geomaterials.

5 Acknowledgements
Giang D. Nguyen and Ha H. Bui gratefully acknowledge support from the Australian
Research Council via Projects FT140100408, DP170103793 & DP190102779.

References

1. Ingraham M D, Issen K A and Holcomb D J: Response of Castlegate sandstone to true tri-

axial states of stress J. Geophys. Res. Solid Earth 118 536±52 (2013)
2. Ma X and Haimson B C: Failure characteristics of two porous sandstones subjected to true
triaxial testing J. Geophys. Res. Solid Earth 121 6477±98 (2016)
3. Vachaparampil A and Ghassemi A: Failure characteristics of three shales under true-triaxial
compression Int. J. Rock Mech. Min. Sci. 100 151±9 (2017)
4. Rudnicki J W and Rice J R: Conditions for the localization of deformation in pressure-sensi-
tive dilatant materials J. Mech. Phys. Solids 23 371±94 (1975)
5. Das A, Tengattini A, Nguyen G D, Viggiani G, Hall S A and Einav I: A thermomechanical
constitutive model for cemented granular materials with quantifiable internal variables. Part
II - Validation and localization analysis J. Mech. Phys. Solids 70 382±405 (2014)
6. Nguyen G D, Nguyen C T, Nguyen V P, Bui H H and Shen L: A size-dependent constitutive
modelling framework for localised failure analysis Comput. Mech. 58 257±80 (2016)
7. Tengattini A, Das A, Nguyen G D, Viggiani G, Hall S A and Einav I: A thermomechanical
constitutive model for cemented granular materials with quantifiable internal variables. Part
I - Theory J. Mech. Phys. Solids 70 382±405 (2014)
8. Nguyen G D, Einav I and Korsunsky A M: How to connect two scales of behaviour in con-
stitutive modelling of geomaterials Géotechnique Lett. 2 129±34 (2012)
9. Nguyen G D, Korsunsky A M and Einav I: A constitutive modelling framework featuring
WZRVFDOHVRIEHKDYLRXUௗ)XQGDPHQWDOVDQGDSSOLFDWLRQVWRTXDVL-brittle failure Eng. Fract.
Mech. 115 221±40 (2014)
10. Le L A, Nguyen G D, Bui H H, Sheikh A H and Kotousove, Khanna, A: Modelling jointed
rock mass as a continuum with an embedded cohesive-frictional model. Eng. Geol. 228, 107±
120 (2017)
11. Le L A, Nguyen G D, Bui H H, Sheikh A H and Kotousove: Localised failure mechanism as
the basis for constitutive modelling of geomaterials. Int. J. Eng. Sci. 133 284±310 (2018)
12. Neilsen M K and Schreyer H L: Bifurcations in elastic-plastic materials Int. J. Solids Struct.
30 521±44 (1993)
13. Hill R: Elastic properties of reinforced solids: Some theoretical principles. J. Mech. Phys.
Solids 11 357±72 (1963)
14. Klein E and Reuschlé T: A model for the mechanical behaviour of bentheim sandstone in the
brittle regime. Pure Appl. Geophys. 160 833±49 (2003)
 Non-linear deformational analysis of reinforced concrete
frame

Tran Thi Thuy Van1 and Vu Thi Bich Quyen1

1
Hanoi Architectural University, Viet Nam
ttthvan.hau@gmail.com

Abstract. The paper presents the algorithm and some results of deformational non-
linear analysis of reinforced concrete frame system taking into consideration of non-
linearity of materials. The nonlinearity of materials means the use of nonlinear models
of concrete and reinforcement, stress-strain relationships of concrete and reinforce-
ment are taken according to diagrams illustrated in Russian codes. The algorithm of
deformational non-linear analysis has two stages. In the first stage, the relationship
between element cross-sectional stiffness and internal forces was established. In the
second stage, the structural analysis software SAP2000 was used for calculation of
internal forces with the results received in the first stage. The results were investigat-
ed with a gradual increase of load values according to the relationship between ele-
ment cross-sectional stiffness and internal forces. This helps to observe the places of
formation of plastic hinges. From there some comments and recommendations were
proposed to evaluate the criterion of limit state in structural analysis.

Keywords: Non-linear deformational analysis, reinforced concrete frame sys-

tem, plastic hinges, non-linear model

1 Introduction

Currently in Vietnam, the analysis and design of reinforced concrete structures are
carried out according to national construction code TCVN 5574-2012 [1] which has
provisions considering plastic deformation. However, in this code, there are no
specific instructions on the calculation method of internal forces including plastic
deformation of materials. So in fact, reinforced concrete structures are calculated with
assumptions that materials work in the elastic phase. Russian construction codes SP
52-101-2003 [2] and SP 63.13330.2012 [3] allow non-linear deformational analysis
for bending members using non-linear deformational models. From there, the
relationship between cross-sectional stiffness of bending elements and internal forces
appearing in these elements is established.
Structural analysis based on deformational models is discussed in the works of
large group of researchers [4-10]. In the proposed calculation methods, it is common
to use non-linear relations between stresses and strains. The main differences are the
use of different state equations or applying for different structures. In [4-9], the
method of non-linear deformational analysis for T-shape and I-shape bending
elements in road bridge superstructures with reinforced concrete and precast pre-

© Springer Nature Singapore Pte Ltd. 2020 317

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_48
318

stressed reinforced concrete is presented. The calculation method allows seeing the
redistribution of internal forces in bending elements of road bridge superstructures.
This paper presents how to apply the method of non-linear deformational analysis for
bending elements of the plane frame system. Applying the result of non-linear
deformational analysis into the analysis of plane frame system allows observing the
redistribution of internal forces in the whole system. That explains the load carrying
capacity of the frame system if there are beams with the decrease of cross-sectional
stiffness, or when the system is subjected to the load greater than the designed load.

2 Non-linear deformational analysis

2.1 Theoretical background

The theoretical basis of the non-linear deformational analysis is the use of the non-
linear deformational model in SP 52-101-2003 [2] and in SP 63.13330.2012 [3], the
kinematical condition of the hypothesis of the flat section and the stress-strain
relationship of compressive concrete and reinforcement corresponds to the diagram
illustrated in Fig. 1 and written in equations (1), (2).
a) b)
s

s0 s2 Rs s2
b

b2 b0 b1 b s2 s0 s
s0 s2

b1

b2 b0 Rb s0 s2 Rsc

Fig. 1. Diagrams İ Iı illustrated in Russian code [2, 3]: a ± for compressive con-
crete, b ± for reinforcement
|ıb1|=0.6Rb, |ıb0|=Rbİb1 ıb1/Ebİb0 íİb2 í (1)
ıs0=Rs, |ısc0|=Rscİs0 ıs0/Esİsc0 ısc0/Es, İs2=0.025, (2)
In which, ȿb, Rb ± Elastic modulus and strength of compressive concrete;
Es, Rs, Rsɫ ± Elastic modulus and strengths of tensile and compressive reinforcement.
The working capacity of the tensile concrete is not included ıbt = 0).
The method of non-linear deformational analysis was mentioned in [4-7]. The relative
strain İb, z = x of the compressive concrete is taken as an independent variable (x-height
of the compressive zone; z-distance measured from the edge of compressive zone to
neutral axis, Fig. 2.). In accordance with the hypothesis of flat sections used in the
calculation, the distribution of relative strains İbİsİsc by the height of the section is
linear. The problem is solved in reverse way; the acting load is a bending moment.
FURPWKHVHWRIVWUDLJKWOLQHVİ f (z), the real (calculated) line is chosen by satifying
the following condition:
F = Fb + Fs + Fsc = 0 (3)
Where, Fb, Fs, Fsc - forces in compressive concrete, tensile and compressive
reinforcements, determined by (5), (6), (7).
Non-linear deformational analysis of reinforced concrete frame 319

The above-mentioned idea allows to build a diagram or make a table, where

eDFK YDOXH RI İb, z = x corresponds to the unique values of M and x. It is possible to
determine the distribution of relative strains İbİsİscVWUHVVHVıbıs, ısc by the cross-
section height, curvature 1/ȡ İs± İb, z = x) / h0, cross-sectional stifIQHVV0ȡ7KH
last parameter M/(1/ȡLVXVHGfor frame analysis in order to reflect the distribution of
cross-sectional stiffness along the length of the beam with a gradual increase of
applying load.
a) b)
bɩ z z
as'

b,z=x b,z=x
hɩ

sc
sc
As'
x

x
b,z

z
h0

h
As
s s
as

Fig. 2. Beam cross-section and graph of non-linear deformational analysis of reinforced con-
crete bending element.

2.2 Non-linear deformational analysis procedure for bending reinforced

concrete elements

The non-linear deformational analysis is established in the following procedure:

Step 1: Given input parameters: geometrical sizes of the cross section b, b f, h, hf,
reinforcement parameters As, Asc, as, asc, h0, material characteristics Eb, Es, Rb, Rs
Step 2: Determinate strains in compressive concrete by the condition:
Hb,z Hb,z x ˜  z / x  (4)
And stresses are determinated by the relations [2] according to Fig.2:
Step 3: Determinate the total forces in compressive concrete, tensile and compres-
sive reinforcements.
x
When x d h f o Fb (H b , x) bf ˜ ³ Vb  H b,z  ˜ dz (5)
0
x hf x
When h f d x o Fb (H b , x) b˜ ³ Vb  H b,z  ˜ dz  bf ˜ ³ Vb  H b,z  ˜ dz
0 x hf

Fsc  Hb , x  Vsc  Hsc  Hb , x  ˜ Asc (6)

Fs  Hb , x  Vs  Hs  Hb , x  ˜ As (7)
Step 4: Determinate the height of the compressive concrete x by solving the fol-
lowing equation:
Fb  Hb , x   Fsc  Hb , x   Fs  Hb , x  0 (8)
Step 5: Determinate the bending moment Mex of all above determined forces about
the neutral axis of the cross- section:
x
When x d h f o M b (Hb , x) bf ˜ ³ Vb  H b,z  ˜ z ˜ dz (9)
0
x hf x
When h f d x o M b (Hb , x) b˜ ³ Vb  H b,z  ˜ z ˜ dz  bf ˜ ³ Vb  H b,z  ˜ z ˜ dz
0 x hf
320

Msc  Hb , x  Fsc  Hb , x  ˜  x  asc  (10)

Ms  Hb , x  Fs  Hb , x  ˜  x  as  h  (11)
Mex Mb  Hb , x   Msc  Hb , x   Ms  Hb , x  (12)
Step 6: Determinate the bending curvature ȡ İb,z=x /x, and cross-sectional stiff-
ness of separate element with length dl: Mexȡ.
Step 8: Draw diagrams of relationships between above-determined parameters.
Based on the proposed procedure the calculation program for non-linear deforma-
tional analysis for bending elements of plane frames is established using MathCad
calculation software.

3 Examples

Nonlinear deformational analysis of plane frame with geometry parameters given in

Fig. 3a is carried out in 2 stages. Material characteristics: Elastic modulus of concrete
ȿb=3 ˜ 104 Ɇ3ɚElastic modulus of reinforcement Es=2.1 ˜ 105Ɇ3ɚStrength of com-
pressive concrete Rb=14. Ɇ3ɚ strength of reinforcement Rs  Ɇ3ɚ UHLQIRUFe-
ment in the beam end section and the middle section of the beam is given in Fig.3b, c.
Given following parameters l1=l2=l3=8m, h1=3.6m, h2=3.6m, q1=q2=12kN/m,
P1=15kN, P2=18kN.
a) b) c)

1 Ø25 2 Ø25
2 Ø25

2 Ø25 Ø8
a150
2 Ø14 2 Ø14

Ø8 1 Ø20
a150
2 Ø25 2 Ø25

Fig. 3. Example of the plane frame and reinforcement in the beams: a ± plane frame, b - rein-
forcement in the beam end section, c ± reinforcement in the middle section of the beam.
Stage 1: Calculation results of the nonlinear deformational analysis for above-
illustrated beams are shown in Fig. 4.
Stage 2: The last diagram M/(1/ρ)= f(M) is applied in to plane frame analysis to
reflect the change of cross-sectional stiffness along the length of beams when the load
increases gradually until forming plastic hinge at the certain section of beams. The
change of cross-sectional leads to the redistribution of internal forces of elements in
the system and allows the entire system to continue carrying the load despite the fact
that the bending moment reached the limit value. The results of the calculations are
shown in Table 1. From the results illustrated in Table 1, it can be seen that, the
results of the non-linar deformational analysis of reinforced concrete beam in plane
Non-linear deformational analysis of reinforced concrete frame 321

frame allow increasing the load carrying capacity proximately 30% in comparison
with result obtained by an ordinary calculation method.
a) b)

d)
c)

Fig. 4. Calculation results of deformational analysis for beams in frames: a ± M=f(İb); b -

1/ȡ= f(İb); c ± M/(1/ȡ)= f(İb); d - M/(1/ȡ)=f(M).

Table 1. Values of bending moment of the beam in accordance with deformational analysis
results in relation with values of bending moment in the limit state.
Bending moment of the beam in relation with value in limit state
q1=0.2qlim q1=0.4qlim q1=0.8qlim q1=1.1qlim q1=1.3qlim
Beam 1 End M=0.23 M=0.65
M=0.3 Mlim M=0.83Mlim M=0.99Mlim
section Mlim Mlim
Middle M=0.22 M=0.66
M=0.35Mlim M=0.86Mlim M=0.94 Mlim
section Mlim Mlim
Beam 2 End M=0.24
M=0.6 Mlim M=0.65Mlim M=0.82Mlim M=0.98 Mlim
section Mlim
Middle M=0.22
M=0.32Mlim M=0.66Mlim M=0.85Mlim M=0.96 Mlim
section Mlim
Beam 3 End M=0.23 M=0.33
M=0.6 Mlim M=0.87Mlim M=0.96 Mlim
section Mlim Mlim
Middle M=0.21 M=0.34
M=0.6 Mlim M=0.9 Mlim M=0.95 Mlim
section Mlim Mlim
Beam 4 End M=0.25 M=0.38
M=0.7 Mlim M=0.86Mlim M=0.97 Mlim
section Mlim Mlim
Middle M=0.22 M=0.37 M=0.63
M=0.85Mlim M=0.96 Mlim
section Mlim Mlim Mlim
Beam 5 End M=0.24 M=0.32
M=0.6 Mlim M=0.84Mlim M=0.96 Mlim
section Mlim Mlim
Middle M=0.21 M=0.65
M=0.3 Mlim M=0.83Mlim M=0.95 Mlim
section Mlim Mlim
Beam 6 End M=0.26 M=0.36 M=0.67
M=0.86Mlim M=0.94 Mlim
section Mlim Mlim Mlim
Middle M=0.20 M=0.35
M=0.69Mlim M=0.88Mlim M=0.97 Mlim
section Mlim Mlim
322

4 Conclusions

Non-linear deformational analysis for bending elements in reinforced concrete plane

frames is applied to establish the relationship between internal forces and cross-
sectional stiffness. The result obtained by the proposed method allows illustrating the
development of the stress-strain state at any period of the loading process. This has an
important meaning in the assessment of the load carrying capacity of the operating
system.
The results of nonlinear deformational analysis applying to plane frames analysis
allow seeing the redistribution of internal forces in the whole system. This explains
the load carrying capacity of the system when the system is subjected to load greater
than the design load or under the condition of that there is one or several beams with
reduction of cross-sectional stiffness.

References
1. Concrete and reinforced concrete structures. National construction design code TCVN
5574-2012 (2012).
2. The Russian code of rules for designing and construction. Concrete and reinforced con-
crete structure with precast-prestressed reinforcement. SP 52-101-2003/ FGUP SPP.-2004
- 53s
3. The Russian code of rules for designing and construction. Concrete and reinforced con-
crete structure. SP 63.13330.2012 / FGUP SPP. (2015).
4. D.M. Shapiro, A.V. Agarkov, Tran Thi Thuy Van. Spatial non-linear deformational analy-
sis of road bridge multy-beam superstructures. Scientific heralds of Voronezh State Uni-
versity of architecture and civil engineering. Serie of construction and architecture ± Voro-
nezh, Russia (2008), p. 29 -37.
5. D.M. Shapiro, Tran Thi Thuy Van. Analysis of road bridge multy-beam superstructures of
the standard design in 1957. Scientific heralds of Voronezh State University of architecture
and civil engineering ± Voronezh, Russia (2007), p. 63 -70.
6. Tran Thi Thuy Van. Non-linear deformational analysis applying in precast pre-stressed
road bridge superstructures. Proceedings of XIV National Scientific Conference on Solid
Mechanics (2018).
7. Tran Thi Thuy Van, D.M. Shapiro. Non-linear deformational analysis for precast pre-
stressed concrete beam systems. E3S Web of Conferences (Vol. 97), XXII International
6FLHQWLILF&RQIHUHQFH³&RQVWUXFWLRQWKH)RUPDWLRQRI/LYLQJ(QYLURQPHQW´)250-2019)
8. D.M. Shapiro. Nonlinear deformation calculation of spatial reinforced concrete spans
highway bridges / D.M. Shapiro, A.P. Tyutin / Building mechanics and construction
(2013), p. 102 - 108.
9. D.M. Shapiro. Nonlinear deformation and bearing capacity of the bridge plate-reinforced
concrete beam spans / D.M. Shapiro, A.P. Tyutin / Building mechanics and construction
(2014), p.78 - 87.
10. S.V.Gorbatov, S.G. Smirnov. Strength calculation of eccentrically compressed reinforced
concrete elements of rectangular cross section based on a non-linear deformation model.
Scientific heralds MGSU, 2011, p. 72-76.
 Reliability Evaluation of Eurocode 4 for Concrete-Filled
Steel Tubular Columns

Huu-Tai Thai and Son Thai

Department of Infrastructure Engineering, The University of Melbourne, Parkville, VIC 3010,

Australia
tai.thai@unimelb.edu.au

Abstract. This paper evaluates the reliability of the design provisions given in
Eurocode 4 (EC4) for concrete-filled steel tubular (CFST) columns under axial
compression. The evaluation is based on the experimental results of 2,224 tests
on short and slender CFST columns (1,245 circular sections and 979 rectangular
sections). Monte Carlo simulation is used to estimate probability of failure and
the reliability index based on the randomness in dead and live loads, steel yield
stress, concrete compressive strength and the error of the EC4 resistance model.
The evaluation is carried out for both specimens within and beyond code limits
of material strengths and section slenderness. The results of parametric studies
indicate that the reliability index of EC4 is in the range from 2.3 to 4.2 which is
quite below the target reliability value of 3.8 prescribed by EC4.

Keywords: CFST Columns, Test Database, Eurocode 4, Reliability Analysis.

1 Introduction

CFST structures have been increasingly used in multi-storey buildings, bridges and
other infrastructure due to their excellent mechanical performance such as high
strength, high ductility and large energy absorption capability. In order to fulfil the need
for sustainable construction, high strength steel and high performance concrete have
been developed. The practical applications of high strength CFST columns with non-
compact and slender sections is therefore expected to increase in high-rise buildings
due to reducing the consumptions of construction materials [1]. For example, Obayashi
Technical Research Institute Main Building in Tokyo, Japan used a CFST column with
yield strength of steel up to 780 MPa and compressive strength of concrete up to 160
MPa.
EC4 [2] only provides design guidelines for CFST columns made from normal strength
concrete of compressive strength of 50 MPa and normal strength steel of yield stress of
460 MPa. An extension of EC4 to concrete cylinder strength up to 90 MPa and steel
yield stress up to 550 MPa was also proposed by Liew and Xiong [3]. It should be noted
that the local buckling of the steel section of CFST columns is also omitted in EC4 by
limiting the section slenderness ratio D/t to 90(235/fy) for circular sections and B/t to
ͷʹඥʹ͵ͷȀ݂௬ for rectangular sections with B, D and t being the diameter, depth and

© Springer Nature Singapore Pte Ltd. 2020 323

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_49
324 H.-T. Thai and S. Thai

thickness of hollow steel sections, respectively, and fy being the yield stress of the steel
tube. It means Class 4 slender section is not allowed in EC4.
The test database collected recently by Thai et al. [4] indicated the availability of testing
data of CFST columns with concrete strength up to 186 MPa and steel yield stress up
to 853 MPa. This test data can be used to evaluate the applicability of EC beyond its
limits on material strengths and section slenderness. This paper will evaluate the relia-
bility of EC for CFST columns under axial compression. The reliability analysis is per-
formed using Monte Carlo simulation with the probabilistic model of the model error
obtained from 2,224 test results collected by Thai et al. [4]. The influences of material
strength and section slenderness on the reliability index of EC are also investigated.

2 EC4 Predictions and Comparison with the Test Results

2.1 Design Equations

In EC4, the section strength of CFST columns under axial compression is calculated
based on the strengths of steel and concrete components. For rectangular sections, the
confining effect is ignored. Therefore, the section strength is obtained as

N us As  f y / J M   Ac  f c' / J C  (1)

For circular sections, if both relative slenderness ratio O d 0.5 and the eccentricity ratio
e D d 0.1 are satisfied, the confining effect is included via the coefficients Ks and Kc as

§ t fy ·
N us As  f y / J M K s  Ac  f c' / J C  ¨1  Kc ¸ (2)
© D f cc ¹

where As and Ac are the areas of the steel tube and concrete, respectively. fy and I¶c are
the yield stress of the steel tube and cylinder characteristic strength of concrete, respec-
tively. J M and J C are partial factors for steel and concrete, respectively. The calcula-
tion of Ks and Kc can be found in EC4 [2]. The member strength of CFST columns is
calculated by multiplying the section strength with a reduction factor F [2] as

N uc F N us (3)

2.2 Test Database

The test database considered in this study includes 2,224 tests on short and slender
CFST columns with 1,245 specimens on circular sections and 979 specimens on rec-
tangular sections. This test database is extracted from over 3,100 tests recently collected
by the authors [4] based on the work of Goode [5]. Only columns under axial compres-
sion are considered. There are 328 specimens (14.8%) with steel yield stress over 460
MPa and 836 specimens (37.6%) with concrete cylinder strength over 50 MPa. The
number of tests beyond the slenderness limit is 550 (24.7%).
Reliability Evaluation of Eurocode 4 for Concrete-Filled … 325

2.3 Test and Prediction Comparison

Fig. 1 shows the comparison between test results and EC4 prediction versus the steel
yield stress fy and concrete cylinder strength I¶c. In general, EC4 is conservative and
underestimates the capacity of CFST columns with the mean of 1.08 and a coefficient
of variation (CoV) of 0.16. EC4 can safely predict the strength of 1,481 columns (67%).
The histogram and statistical properties (mean, CoV and distribution type) of Test/EC4
prediction ratio are illustrated in Fig. 2 for circular and rectangular columns.

≤460MPa >460MPa ≤50MPa >50MPa

85.2% 14.8% 62.4% 37.6%

Fig. 1. Comparison of Test/EC4 ratio versus steel and concrete strengths.

Mean = 1.09 Mean = 1.07

CoV = 0.15 CoV = 0.18
Sample = 1,245 Sample = 979
Gumbel distribution Gumbel distribution

Fig. 2. Histogram and statistical properties of Test/EC4 ratio.

3 Reliability Analysis

The reliability index E is related to the probability of failure Pf as E ) 1  Pf  where

) is the standard normal cumulative distribution function. The value of Pf can be
326 H.-T. Thai and S. Thai

predicted using Monte Carlo simulation as Pf P G  0 n f / n where nf is the num-

ber of simulations that G  0 and n is the total number of simulations. The limit state
function is defined as G R  Q with R and Q being the resistance and total dead and
live loads, respectively. To account for bias correction, the resistance R is obtained by
multiplying the code prediction Nuc with the model error (ME) as R ME u Nuc . The
value of Nuc is obtained from Eq. (3) (ignoring partial factors for materials i.e.,
J M J C 1.0 ) by considering the randomness in material properties. The nominal val-
ues of dead load Dn and live load Ln can be calculated from the design resistance NRd
for a fixed load ratio of Ln/Dn as N Rd 1.35Dn  1.6 Ln . The design resistance NRd is
obtained from Eq. (3) with partial factors for steel J M 1.0 and concrete J C 1.5 . In
this study, material, load and model error are treated as random variables, whilst the
randomness in geometry is ignored due to its small variability. The statistical properties
and distribution of random variables are summarized in Table 1. It is noted that the
statistical properties of ME are obtained in this paper based on the experimental results
of 2,224 tests (see Fig. 2). Fig. 3 illustrates the results of Monte Carlo simulation for a
typical CFST circular columns (t = 5mm, D = 200mm, L = 4m, fy = 350MPa, f'c =
40MPa, Ln/Dn = 2.0).

Table 1. Random variables considered in the reliability analysis.

Properties Variables Mean/Nominal CoV Distribution References

Material fy 1.10 0.06 Normal [6]
f'c 1.08 0.15 Normal [7]
Load Dn (dead) 1.05 0.10 Normal [8]
Ln (live) 1.00 0.25 Gumbel [8]
Model error ME (circular) 1.09 0.15 Gumbel Fig. 2
ME (rectangular) 1.07 0.18 Gumbel Fig. 2

Pf = 0.0018
o Safe
Load Q E = 2.92 o Failure
Sample = 106

Resistance R

Fig. 3. Monte Carlo simulation for a typical CFST column.

Reliability Evaluation of Eurocode 4 for Concrete-Filled … 327

4 Reliability Results and Discussion

In this parametric study, the influences of material strengths (f y and I¶c), section slen-
derness ratio (D/t or B/t) and live load-to-dead load ratio (Ln/Dn) are investigated. In
each case study, a typical specimen with t = 5mm, D=B=H=40t, L = 3D (short column)
or 20D (long column), fy = 350MPa, f'c = 40MPa, Ln/Dn = 2.0) is considered. It can be
seen that increasing fy or Ln/Dn will reduce the reliability index of EC4 as shown in Fig.
4(a) and (d), whereas increasing I¶c or section slenderness will increase the reliability
index as shown in Fig. 4(b) and (c). It means that EC4 can safely extend to CFST col-
umns with high strength concrete and slender sections. But it is not true for the case of
high strength steel. Fig. 4(d) also indicates that the reliability index of circular section
is greater than that of rectangular section. This is due to the fact that the CoV of the
model error of circular section is smaller than that of rectangular section (see Fig. 2).
In other words, EC4 equation for circular section Eq. (2) is more conservative that that
for rectangular section Eq. (1). There is no clearly trend on the effect of short and long
columns on the reliability index. In general, the reliability index of EC4 for CFST col-
umns is in the range from 2.3 to 4.2 for a wide range of geometric and material param-
eters as well as loading ratios considered in this study. The minimum value of the reli-
ability index of 2.3 is quite below the target reliability value of 3.8 prescribed by EC4.

(a) fy effect (b) I¶c effect

(c) Section slenderness effect (d) Ln/Dn effect

Fig. 4. Effect of fy, I¶c, section slenderness and live load-to-dead load ratio
328 H.-T. Thai and S. Thai

5 Conclusions

This paper presented an investigation on the reliability of EC4 in predicting the ultimate
load-carrying capacity of CFST columns under axial compression. A database of over
2,200 tests on CFST columns was collected from the open literature to calibrate the
validity of EC4 design equations and determine the probabilistic model of the EC4 code
prediction. Monte Carlo simulation was then adopted to calculate the reliability index
of CFST columns with the randomness in model error, dead load, live load, steel yield
stress, concrete compressive strength. The obtained reliability index of EC4 varies from
2.3 to 4.2 for a wide range of dead load-to-live load ratio and geometric and material
parameters considered in the parametric study. The results indicate that EC4 cannot
meet a target reliability value of 3.8. Therefore, it is necessary to re-examine the design
equations of EC4 for axially loaded CFST columns.

References

1. Liew J.Y.R., Xiong M., Xiong D.: Design of Concrete Filled Tubular Beam-columns with
High Strength Steel and Concrete. Structures 8,213-26 (2016)
2. Eurocode 4. Design of composite steel and concrete structures Part 1-1: general rules and
rules for buildings. Brussels, Belgium: CEN. 2004.
3. Liew J.R., Xiong M. Design guide for concrete filled tubular members with high strength
materials - an extension of Eurocode 4 method to C90/105 concrete and S550 steel: Research
Publishing; 2015.
4. Thai S., Thai H.-T., Uy B., Ngo T.: Concrete-filled steel tubular columns: Test database,
design and calibration. Journal of Constructional Steel Research 157,161-81 (2019)
5. Goode C.D.: Composite columns - 1819 tests on concrete-filled steel tube columns compared
with Eurocode 4. The Institution of Structural Engineers 86(16),33-8 (2008)
6. Bartlett F.M., Dexter R.J., Graeser M.D., Jelinek J.J., Schmidt B.J., Galambos T.V.: Updating
standard shape material properties database for design and reliability. Engineering Journal-
AISC 40(1),2-14 (2003)
7. Bartlett F.M., Macgregor J.G.: Statistical analysis of the compressive strength of concrete in
structures. ACI Materials Journal 93(2),158-68 (1996)
8. Ellingwood B., Galambos T.V.: Probability-based criteria for structural design. Structural
Safety 1(1),15-26 (1982)
 Part III
Sustainable Construction Materials and
Technologies
 Preparation of low cement ultra-high performance
concrete

Norzaireen Mohd Azmee1[norzaireen@gmail.com] and Nasir Shafiq1[nasirshafiq@utp.edu.my]

1
Universiti Teknologi PETRONAS, MALAYSIA

Abstract. Technology advancement in concrete industry and expanding interest for high
quality construction materials have prompted the development of ultra-high performance
concrete (UHPC). Despite of many advantages gained using this performance-based con-
crete, however, conventional UHPC recipe raises many concerns especially on sustaina-
bility issues. Producing UHPC, requires relatively high amount of cement content and of-
ten the compositions are not optimized. Low cement UHPC was proposed to minimize
the economic and environmental disadvantages of current UHPC by incorporating high
content of supplementary cementing materials (SCM). An experimental program was car-
ried out to evaluate the effect of SCM combinations on workability and compressive
strength. It was found that good consistency and highest strength of low cement UHPC
could be achieved with binary combination of fly ash and ultrafine calcium carbonate as
SCM replacing up to 50% of cement. Enhancement of compressive strength as early as 7
days at 12.5% followed by 8% at 28 days and 20% at 90 days were observed in this low
cement UHPC.

Keywords: Ultra-high performance concrete, High volume fly ash, Ultrafine CaCO3,
Compressive strength.

1 Introduction

Concrete is the most consumed material after water and worldwide about 10 billion tons of
concrete are produced annualy [1]. Over the 30 years, remarkable advances have been made in
concrete technology. Recent advancement of UHPC showed very exciting results in both
strength and durability [2]. Despite many advantages gained using UHPC, however, conven-
tional UHPC recipe raises many concerns especially on sustainability issues due to its extreme-
ly high usage of cement content, high production cost, and sensitive in-situ construction [3].
Production of cement requires high energy consumptions which contributes about 7% of CO2
emissions [4]. On top of that, the cement manufacturing also requires heavy usage of natural
resources that disturb the eco-system. Cement content in UHPC can reach as high as 1200kg/m3
compared to 300-500kg/m3 for conventional concrete [3,4]. Using higher content of SCM in
UHPC would lower the amount of cement usage, reduce the cost with lower emission. Accord-
ing to past researches, due to the low water to binder ratio (W/B) used in UHPC, only half of
cement gets hydrated [5]. The remaining unhydrated cement particles works as micro-
aggregates making it to be the most expensive fillers.

© Springer Nature Singapore Pte Ltd. 2020 331

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_50
332 N. M. Azmee and N. Shafiq

UHPC with SCM, such as SF, slag and FA have been reported as an excellent solution from
durability point of view. Among all, SF is the most famous pozzolanic material particularly for
developing UHPC with maximum replacement of 30% by weight of cement [6]. SF plays im-
portant roles in UHPC by filling up voids between cement grains, enhances the rheological
characteristics and forms hydration products through pozzolanic activity [7]. Another important
effect of the SF in UHPC is the improvement of the interfacial transition zones (ITZ) between
binder and steel fibers which increases the mechanical strength and enhances the microstruc-
ture. The only issue faced when incorporating SF in UHPC is the high price of SF. Great atten-
tions have been given in search of substituting SF and part of cement to produce UHPC with
less cost and lower emission while providing the equivalent properties. Available in huge quan-
tities worldwide as waste material with very low cost, FA is gaining its reputation as the best
option to replace SF and part of cement in UHPC. FA has been commonly used to replace part
of cement in normal to high strength concrete with an average percentage replacement of 30%
[2,3,8]. In the preparation to develop low cement UHPC, high volume fly ash (HVFA) in the
replacements of 30-70% were used in this study. While, FA has been reported to negatively
influenced the early age of concrete strength [9]. Recent findings have shown that HVFA works
better when concrete is prepared using low W/B [5,10]. Lam et al. [11] demonstrated a 13%
improvement for 28-day compressive strength when W/B was reduced from 0.5 to 0.25. Be-
sides reducing the W/B, incorporation of ultrafine particles would also benefit the strength
development of HVFA concrete. These particles will act as nuclei for cement to accelerate the
hydration process, to reduce permeability and to densify both the microstructure and ITZ [12].
The combination between FA and ultrafine particles can tightly bond the hydration product and
accelerates the pozzolanic reaction improving its early strength [13]. This present paper aims at
the development of an eco-efficient UHPC using HVFA and ultrafine particles in providing
sustainable and yet economical solutions for the growing demand of UHPC in the industry.

2 Experimental Work

2.1 Material and Mix Proportions

Locally produced FA, SF and UFCC with respected particle mean size of 24.4m, 0.16m and
3.5m was used as SCM. Crushed granite with maximum size of 10mm, river sand with fine-
ness modulus of 2.8 and Polycarboxylic Ether based superplasticizer (SP) were selected and
used in all mixes. The experimental program consisted of mixing, casting and testing 150 cubes
of size 100mm. A 60-L pan mixer was used to prepare fresh UHPC in this study. All concrete
mixes were mixed for about 20 minutes. The consistency testing of fresh UHPC was measured
using standard 300mm slump cone. After fresh testing and casting was completed, samples
were well covered by plastic sheet. Samples were then demoulded after 24 hours and shifted to
drying shelves for ambient curing until the specified testing time. Details of the mix proportions
are given in Table 1. Water to binder ratio was kept constant to 16% throughout the mix, the
usage of SP varies from 1.5-4% depending on the combination type of SCM used. Compressive
strength and microstructure of UHPC mixture were investigated. Samples of SEM analysis was
produced by taking small pieces from the concrete fractured specimen. Prior to testing, the
broken pieces were soaked using acetone and oven dried at 60°C to stop the hydration.
Preparation of low cement ultra-high performance concrete 333

Table 1. UHPC mix proportions

% Binder % Fiber
Phase Mix ID Binder CA Sand
OPC FA SF UFCC SE EH
A MA1-MA3 500-900 100 - - - 930-1195 620-797 - -
B MB1-MB6 900 100 - - - 930 620 0-2 0-2
C MC1-MC5 900 30-70 30-70 - - 930 620 1 -
D MD1-MD5 900 30-90 0-60 10 - 930 620 1 -
E ME1-ME6 900 50-60 30-40 - 2.5-10 930 620 1 -
CA = Coarse aggregate; SE = Straight fiber; EH = Hooked-end fiber

3 Results and Discussions

3.1 Influence of Binder Content on Strength and Fluidity of UHPC

Mix MA1 with 500 kg/m3 binder exhibited an extremely dry fresh mixture with insufficient
paste that was hardly workable. Improvement in slump flow together with the hardened con-
crete appearance can be seen as total binder increased from 500kg/m3 to 900kg/m3 and as ratios
for mortar to CA volume increased. Slump spread for mix MA3 was 2.8 times higher than mix
MA1. Table 2 shows that, at fixed W/B, the 28 days compressive strength improved by 13%
and 88% as the binder content increased from 500kg/m3 and 700kg/m3 to 900kg/m3 respective-
ly suggesting higher paste volume is required in attaining high fluidity and ultra-high strength
behavior.

Table 2. Influence of binder on strength and fluidity of UHPC

Binder Compressive strength
Slump spread Targeted slump
Mix content (MPa)
(mm) spread (mm)
(kg/m3) 7d 28d
MA1 500 49 56 300
MA2 700 59 63 450 > 750
MA3 900 80 105 850

3.2 Influence of Binder Content on Strength and Fluidity of UHPC

The beneficial effect of the fiber addition to UHPC on its compressive strength is reported by
various researchers [14]. In sub phase B, compressive strength and fluidity test results of UHPC
mixes containing different fiber type and content (0,1 and 2%) tested at 28 days are presented
in Table 3. Addition of EH fiber with aspect ratio of 83.33 in UHPC mixes at any fiber content
of 0.5%, 1% and 2% were found to be unsuitable. UHPC mixes with EH fiber inclusions have
high tendency of fiber clumping and segregation with slump spread of less than 300mm. Earlier
study done on similar UHPC mixes having maximum particle size of 1.2mm showed that the
optimum fiber used are the combination of EH and SE fiber in the total quantity of 2% by vol-
ume [14,15]. However, in this study, inclusion of 10mm coarse aggregate limits the allowable
fiber content to be less than 2% and fiber length should be kept maximum of 20mm. UHPC
with more than 1% EH fiber experienced severe fiber balling/clumping and unworkable mixes,
thus no specimens were casted for latter mixes. It can be observed from Table 4 that the maxi-
mum 28 days compressive strength of UHPC is from Mix MB4 with 1% of SF fiber at
334 N. M. Azmee and N. Shafiq

125MPa. A reduction of 16% and 22% can be observed in the 28 days compressive strength
once 0.5% of SF was partially and fully replaced by EH fiber.

Table 3. Effect of fiber on strength and fluidity of UHPC

28d Compressive
Slump spread
Mix strength Visual inspection
(MPa) (mm)
MB1-MB3 - < 300 Severe fiber balling and mixes are not workable
MB4 125 710 Acceptable
MB5 98 420 Fiber balling and segregation
MB6 105 560 Fiber balling and segregation

3.3 Influence of SF and UFCC on binary blended FA-OPC UHPC

The effect of HVFA in combination with SF/UFCC on compressive strength of UHPC was
studied. Figure 1 shows that addition of FA reduced the heat of hydration thus slower down the
setting time that caused reduction in the early age strength, which explained the slower strength
development for all mix with only FA (Phase C). It is clear that combination of FA and 10% SF
(mixes in Phase D) at similar replacement provided higher early strength with improvement
that ranges from 20% to 58% except for mix MD2. SF acts as reactive pozzolan and effective
void filler to improve cement paste cohesion and aggregate particle adhesion. This can be ex-
plained by better dispersion of cement grains, which leads to an improved gel phase’s devel-
opment [16]. Highest compressive strength of 95MPa and 135MPa can be seen at 7 and 28 days
respectively on mix (ME3) with 40% FA and 10% UFCC. It can be seen from Figure 5 that the
7-day compressive strength of mixes in Phase E increased by about 25% due to the addition of
UFCC that ranges from 2.5% to 10%. The strength enhancement due to the inclusion of UFCC
in UHPC containing HVFA is acceptable when compared to results obtained by [11,16]. In
their studies, they explained that inclusion of CC powder in the size of less than 1m improves
the microstructure by forming additional C-S-H gels and decreases the CH and Ca2SiO4 of
HVFA concretes. Not only that the CC powder led to much denser microstructure, it also re-
sponsible in changing the formation of the hydration products which contributed to the im-
provement of early age and later age compressive strength and durability properties. UFCC has
high reactivity and effective filler effect in reducing the pore space and its connectivity inside
the concrete [11].

Fig. 1. Compressive strength results of UHPC at various SCM

Preparation of low cement ultra-high performance concrete 335

3.4 Microstructure properties on hydrated UHPC

SEM was used to investigate the microstructure of the UHPC mixes (MB4, MC3, MD3 and
ME3) tested at 90 days. Figure 3 shows, SEM images of a very dense structure in the hardened
pastes with very few air pores. Homogeneous morphology of calcium silicate hydrate (C–S–H)
gel as main hydration product was observed with no sign of calcium hydroxide (CH) and
ettringite (AFm) products. The extremely low W/B and high powder content used in UHPC
resulted in a very low porosity, restricting the available space for CH crystals to grow. Presents
of unreacted FA and UFCC particles can be seen in Figure 3(b) which served as micro aggre-
gates that contributed to the strength of the cementitious material [5,8]. Synergistic effect of FA
and UFCC has lower down the capillary pores for mix ME3 compared to other mixes. This is
consistent with the compressive strength results, which reach the highest compressive strength
of 135MPa at 28 days. Figure 4(b) reveals that the UHPC prepared with very low W/B, and
with hardened paste made of cement incorporating HVFA at 40% and 10% UFCC, and coarse
aggregate with maximum size of 10mm, has a very compact ITZ structure compared to mix
without any replacement (Figure 4(a)). No distinct porosity is found in the ITZ of mix ME3,
which is conducive to UHPC preparation. A homogenous morphology of the paste structure
and the compact ITZ structure are the theoretical rationales for preparing UHPC without re-
moving coarse aggregate. The dense microstructure of mix ME3 was extended to aggregate
boundary and as a result the well-known microstructure gradient of the cement paste towards
the transition zone was almost absent. This work observed even when no pressure was applied
to fresh concrete of mix ME3.

(a) (b)

Fig. 2. SEM images (x1000) for hardened UHPC paste: (a) Mix MB4 and (b) Mix ME3

(a) Paste (b) Paste

Aggregate
Aggregate

Fig. 3. SEM images (x2000) of ITZ in hardened UHPC: (a) Mix MB4 and (b) Mix ME3

4 Conclusions

1. UHPC with compressive strength of up to 135MPa at 28 days was successfully pre-

pared without special curing treatment using only 450kg/m3 cement with combina-
tions of FA and UFCC.
336 N. M. Azmee and N. Shafiq

2. The addition of 10% UFCC accelerated the pozzolanic reaction and increased the 7-
day compressive strength of UHPC containing 40% and 50% FA by about 10-20%.
At 28 days, the increase in compressive strength was between 8-25%. It is noted that
UFCC with mean particle size of 3.5m can be used as a cheaper alternative to fully
replace SF and compensates the low early strength of HVFA concrete. UHPC con-
taining both UFCC and HVFA at 50% replacement level offers great potential for use
in concrete construction where both short and long term compressive strength are
considered.
3. SEM image analysis show the incorporation of 10% UFCC, as a partial cement re-
placement in UHPC with HVFA densified the microstructure which led to the im-
provement of compressive strength. The particle packing effect of UFCC plays an
important role in reducing the volume of pores and enhancing the ITZ.

References
1. Patel V., Shah N.: Survey of HPC Developments in Civil Engineering Field. Open Journal of Civil
Engineering, pp. 69-79, (2013).
2. Schmidt M., Fehling E.: UHPC: Research Development and Application in Europe. American Con-
crete Institute, 228(4):51-78 (2005).
3. Azmee NM., Nuruddin MF.: Effect of HVFA in UHPC on Compressive Strength. Engineering Chal-
lenges for Sustainable Future, London: Taylor and Francis Group, pp. 463-466 (2016).
4. G. Habert G., E. Denarie E., et al.: Lowering the Global Warming Impact of Bridge Rehabilitations by
Using UHPFRC. Cement and Concrete Composites, no.38, pp. 1-11 (2013).
5. Poon CS., Lam L., Wong YL.: A study on high strength concrete prepared with large volumes of low
calcium fly ash. Cement Concrete Research, no.30, pp. 447–455 (2000).
6. Wille K., Naaman AE., et al.: UHPC and FRC: achieving strength and ductility without heat curing.
Materials and Structures, no. 45, pp. 309-324 (2012).
7. Chung DDL.: Review: improving cement-based materials by using silica fume. Journal Material Sci-
ence, no.37, pp.673–82 (2002).
8. Berry EE., Hemmings RT., et al.: Hydration in HVFA concrete binder. ACI Material. Journal, no.91,
pp.382–389, 1994.
9. Babu KG., Rao GSN.: Efficiency of fly ash in concrete with age. Cement Concrete Research, no.26,
pp.465-474 (1996).
10. Wang C., Yang C., Liu F., Wan C., Pu X.: Preparation of UHPC with common technology and mate-
rials. Cement concrete composite, no.34, pp. 538-544 (2012).
11. Lam L., Wong YL., Poon CS.: Effect of fly ash and silica fume on compressive and fracture behaviors
of concrete. Cement Concrete Research, no.28, pp. 271–283 (1998).
12. Sanchez F., Sobolev K.: Nanotechnology in concrete – a review. Construction Building Materials, pp.
2060–2071 (2010).
13. Shaikh F., Supit S., Sarker P.: A study on the effect of nano silica on compressive strength of HVFA
mortars and concretes. Materials and Design, pp. 433–442 (2014).
14. Bonneau O., Lachemi M., et al.: Mechanical properties and durability of two industrial reactive pow-
der concretes. ACI Materials Journal, 94(4):286–290 (1997).
15. Nuruddin MF., Azmee NM., Chang KY.: Effect of MIRHA and Fly Ash in Ductile Self-Compacting
Concrete on Abrasion and Impact Performance. Applied Mechanics and Materials, (2014).
16. Xu A., Sarkar SL.: Microstructural Developments in HVFA Cement System. Journal Material Civil
Engineering 6:117-136 (1994).
 Matrix dependent piezoresistivity responses of high
performance fiber-reinforced concretes

Duy-Liem Nguyen1, Thi-Bich-Nga Vu2,*, Huynh-Tan-Tai Nguyen1, Van-Ben Nguyen1

1HoChi Minh City University of Technology and Education, Ho Chi Minh City, Vietnam
2Hong Bang International University, Ho Chi Minh City, Vietnam

(liemnd@hcmute.edu.vn, *Corresponding email:

ngavtb@hiu.vn,tainht@hcmute.edu.vn, nvben165@gmail.com)

Abstract. An experiment was conducted to investigate the matrix dependent pi-

ezoresistivity responses of high performance fiber-reinforced concretes
(HPFRCs) under direct tension. Three comparative HPFRCs were produced from
following matrice: M1 was the controlled matrix while amount of cement in M2
or M3 was partly replaced by CB or GGBS, respectively. All HPFRCs contained
same amount of twisted steel fibers 2% volume fraction. The investigated
HPFRCs exhibited strain-hardening and both self strain-sensing and self damage-
sensing abilities. The replacement of cement by CB and GGBS produced high
enhancements of first-cracking strengths but great reductions of gauge factors
within first-cracking points. Besides, as CB and GGBS were used in M2 and M3,
respectively, there were not clear effects on post-cracking strengths but slight
enhancements of gauge factors within post-cracking points were observed.

Keywords: High-Performance Fiber-Reinforced Concretes, Gauge Factor,

Carbon Black, Furnace Slage, Self Sensing

1 Introduction

The serious deterioration of civil constructions may cause catastrophic collapse during
long-time service. This matter attracts much researcher performing on structural health
monitoring to detect damage and structural resistances of the constructions periodi-
cally. Conventional techniques for monitoring structural health are to use attached or
embedded sensors, these techniques are evaluated to be still limited with their high cost
and low durability [1]. To use cement-based material as a self-sensor is a new technique
that can overcome these limitations, i.e., materials can sense strain/stress and dam-
age/crack by measuring the electrical resistivities of them under external loads, no need
for replacing sensors and no concern for affecting the structure by embedding sensors.
High performance fiber-reinforced concretes (HPFRCs) demonstrated their high me-
chanical behaviors and crack resistances, e.g., compressive strength more than 80 MPa,
post-cracking tensile strength more than 10 MPa, strain capacity more than 0.4% with
multiple micro-cracks during strain-hardening [2-4]. Additionally, HPFRCs have re-
cently been reported as smart materials which can produce high damage-sensing capac-

© Springer Nature Singapore Pte Ltd. 2020 337

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_51
338 D.-L. Nguyen et al.

ity [2-6]. Fig. 1 shows the typical piezoresistivity response of HPFRCs, the self dam-
age-sensing ability was investigated from start of load to the post-cracking point (  pc ,
 pc ) while the self strain-sensing ability was studied within the first-cracking point (
 cc ,  cc ). There is a great demand in developing both strain-sensing and damage-sens-
ing capacity of HPFRCs since some structural members have been designed to serve in
elastic state while others have been served in inelastic state with crack occuring. Both
carbon black (CB) and ground granulated blast furnace slag (GGBS) are very fine ma-
terials, thus the use of CB and GGBS for partial cement-replacement is expected to
enhance mechanical resistances of HPFRCs by minimizing air voids inside concretes.
Moreover, GGBS is a by-product from the blast-furnace of iron while a large amounts
of carbon-dioxide gas emit into the atmosphere during the production of cement that
plays a major contributor for green house effect and global warming. Therefore, the use
of GGBS as a cement alternative will bring much benefit in environment. This situation
motivated an experimental test reported in this research, which concentrates on the ef-
fect of carbon black (CB) and ground granulated blast furnace slag (GGBS) on piezo-
resistivity responses under direct tension of HPFRCs. The study was expected that CB
or/and GGBS would produce high enhancements of tensile resistances as well as self-
sensing capacities of HPFRCs.
resistivity
Electrical

Point A: First-cracking
pc ≥ cc
Tensile
stress
Point B: Post-cracking

Hardening branch B (pc , pc )

Crack generating Softening branch
A Horizontal branch
(cc , cc )
Linear branch
No Crack

cc pc C
Strain-sensing w

Damage-sensing

Fig. 1. Typical piezoresistivity response of HPFRCs

2 Experiment

2.1 Materials and Mix Design

Three matrice were designed to investigate the effects of CB and GGBS on the piezo-
resistivity responses of HPFRCs under direct tension as follows: M1 was the controlled
matrix while amount of cement in M2 or M3 was partly replaced by an additive: CB or
GGBS, respectively. The CB content of M2 was 1% of the cement by weight, whereas
the GGBS content of M3 was 25% of the cement by weight. The total weight content
of cement and additive were kept constant as provided in Table 1. The compressive
Matrix dependent piezoresistivity responses … 339

strengths of plain mortar matrice (with no fiber) of M1, M2, M3 were examined to be
89, 92, 109 MPa, respectively, while their resistivities were detected to be 168.11,
150.30 and 155.94 kΩ-cm, respectively. As shown in Table 1, CB and GGBS reduced
the resistivity of M1 by 10.6% and 7.3%, respectively owing to their lower electrical
resistivities than cement replaced [4]. Matrix M1, M2, M3 were added same 2% twisted
steel fiber content by volume. The twisted fiber used had its length of 30 mm with three
ribs, equivalent diameter of 0.3 mm with triangular cross section and a density of 7.9
g/cm3. The used twisted fiber had tensile strength of 2428 MPa, elastic modulus of 200
GPa and electrical resistivity of 1.94×10-8 k-cm. All investigated specimens were
cured in a water tank at 25°C for 14 days. After curing, the specimens were removed
from the water tank then dried at 70°C in an oven for 12 h; this task helps reduce the
humidity effect on electrical resistivity.
Table 1. Compositions, compressive strengths, electrical resistivities of the studied matrices

Cement Silica Silica Fly Super- Carbon f 'c m

GGBS Water
(Type 3) fume sand ash plasticizer black (MPa) (kΩ -cm)
M1 0.800 0.07 1.00 0.20 0.04 - - 0.26 89 168.11
M2 0.792 0.07 1.00 0.20 0.04 0.008 - 0.26 92 150.30
M3 0.640 0.07 1.00 0.20 0.04 - 0.16 0.26 109 155.94

2.2 Test Set-up And Procedure

(a) Specimen geometry (b) Tensile test set-up

Fig. 2 Specimen geometry and test set-up

All tests were conducted with mixture age of 18 days. For measuring the electrical re-
sistivities of the specimens, a layer of silver paste was first applied onto the surfaces of
each specimen to enhance the electrical conductivity between the copper tapes and
specimen; the copper tapes were attached on the silver-paste layer as shown in Fig 2.
340 D.-L. Nguyen et al.

3 Testing Result And Discussion

a) M1, strain up to first crack b) M1, strain up to post crack

c) M2, strain up to first crack d) M2, strain up to post crack

e) M3, strain up to first crack f) M3, strain up to post crack

Fig. 3 Piezoresistivity responses of investigated HPFRCs at first and post crack

A gauge (or strain) factor is defined as the fractional changes in the electrical resistance
per unit strain and is applied to evaluate the self strain-sensing or self damage-sensing
capacity of HPFRCs. For a strain-hardening HPFRCs, the first-cracking gauge factor
Matrix dependent piezoresistivity responses … 341

( GFcc ) would be derived based on the difference in the electrical resistance between the
start of loading and the first-cracking point while the post-cracking gauge factor ( GFpc )
would be computed based on the difference in the electrical resistance between the start
of loading and the post-cracking point. Eqs (1) & (2) can be used to compute the GFcc
and GF pc of HPFRCs, respectively.

GFcc 
R R0  R  Rcc  R0   R0  Rcc    0  cc   1  0 / cc 
 0
(1)
  cc  0  R0   cc 0   cc  cc

GFpc 
R R0

 R0  R pc  R0   R0  R pc    0   pc   1  0 /  pc  (2)
  pc  0  R0   pc  0   pc  pc
where R0 (or  0 ), Rcc (or cc ) and R pc (or  pc ) are the corresponding electrical re-
sistances (or resistivities) at the start of loading, the first-cracking point and the post-
cracking point, respectively.
Table 2 Electro-tensile parameters of HPFRCs

At first crack At post crack

Matrix Spe.
 cc (%)  cc (MPa) GFcc  pc (%)  pc (MPa) GFpc
Sp1 0.039 4.19 1155.06 0.57 12.60 158.95
M1 Sp2 0.037 5.53 811.95 0.54 12.14 168.59
(con- Sp30.040 3.56 719.45 0.55 12.84 142.07
trolled) 0.039 4.43 895.49 0.55 12.53 156.54
Ave. (0.002) (1.01) (229.51) (0.013) (0.36) (13.43)
Sp1 0.046 7.49 309.18 0.59 12.51 125.58
Sp2 0.047 6.60 481.89 0.47 11.08 197.29
0.047 7.05 395.53 0.53 11.79 161.44
M2 (CB) Ave.
(0.001) (0.63) (122.13) (0.058) (1.01) (50.71)
Sp1 0.036 7.12 110.40 0.45 12.39 173.67
Sp2 0.039 7.57 71.60 0.45 13.17 165.27
Sp3 0.031 7.33 199.56 0.42 11.77 163.45
M3 0.035 7.35 127.18 0.44 12.44 167.46
Ave. (0.004) (0.24) (65.61) (0.018) (0.70) (5.45)
(GGBS)
The standard deviations are provided within parentheses
Fig. 3 shows the piezoresistivity responses of three investigated HPFRCs generated
from M1 (Figs. a,b), M2 (Figs. c,d), M3 (Figs. e,f). In each mixture, the curves of rela-
tive reduction in electrical resistivity versus strain and tensile stress versus strain were
performed. All curves of tensile stress versus strain of HPFRCs demonstrated strain-
hardening whereas all curves of reduction in electrical resistivity versus strain of them
expressed the strain-softening. Tables 2 provides the electro-tensile parameters of in-
vestigated HPFRCs at the first-cracking and the post-cracking points. As shown in Ta-
ble 2, the first-cracking tensile strengths were significantly enhanced when CB and
GGBS were partly replaced cement with order as follows: M3 (7.53 MPa) >M2 (7.05
MPa) >M1 (4.43 MPa). However, the GFcc exhibited the opposite order as follows: M1
342 D.-L. Nguyen et al.

(985.49) >M2 (395.57) >M3 (127.18). At the post-cracking points, the trend of  pc were
not clear, M2 & M3 produced slight reductions in post-cracking strengths: M1 (12.53
MPa) >M3 (12.44 MPa) > M2 (11.79 MPa). There were slight enhancements of GFpc
as CB and GGBS were used for M2 and M3 as follows: M3 (167.46) >M2 (161.44)
>M1 (156.54 MPa). The  pc was much higher than the  cc (1,67-2.84 times) regardless
of matrix type whereas the GF pc was generally lower than the GFcc , except for M3. All
investigated matrice produced high gauge factors, from 127.18 to 895.49, within both
the first crack and the post crack. These values are much higher than the commercially
conventional gauge factor around 2 [1]. The highest GFcc was 895.49, produced by M1
while the highest GF pc was 167.46, produced by M3.

4 Conculusion

1) The replacement of cement by CB or GGBS produced significant enhancements of

first-cracking strengths but significant reductions of gauge factors within first-cracking
points. The effects of CB and GGBS on post-cracking strengths were not clear. The
replacements of cement by CB and GGBS produced slight enhancements of gauge fac-
tors within post-cracking points.
2) All matrice produced tensile strain-hardening behaviors and clear reductions in elec-
trical resistivity during tension. The investigated HPFRCs exhibited not only high self
strain-sensing capacities but also high self damage-sensing capacities.

Acknowledgments

This work is funded by Hong Bang International University under grant code GV1905.

References
1. Chung, D.D.L.: Self-monitoring structural materials. Mater Sci Eng: R: Reports 22(2) 57 –
78 (1998).
2. Naaman, A.E., Reinhardt, H.W.: Proposed classification of HPFRC composites based on
their tensile response. Mater. Struct. 39 547 – 555 (2006).
3. Nguyen, D.L., Song, J., Manathamsombat, C., Kim, D.J.: Comparative electromechanical
damage-sensing behavior of six strain-hardening steel-fiber-reinforced cementitious com-
posites under direct tension. Composites: Part B; 69, pp. 159-168 (2015).
4. Song, J., Nguyen, D.L., Manathamsombat, C., Kim, D.J.: Effect of fiber volume content on
electromechanical behavior of strain-hardening steel-fiber-reinforced cementitious compo-
sites. Journal of Composite Materials, doi:10.1177 /0 021998314568169 (2015).
5. Nguyen, D.L. and Kim, D.J.: Self damage sensing of fiber reinforced cementitious compo-
sites using macro-steel- and micro-carbon-fibers, Proceedings of The 6th International Con-
ference on Engineering and Applied Sciences, Hong Kong, June 8-10, pp 144-152 (2016).
6. Wen, S.H., Chung, D.D.L.: A comparative study of steel- and carbon-fiber cement as piezo-
resistive strain sensors. Adv Cem Res. 15(3): 119–28 (2003).
 Influence of elastic modulus under uniaxial tension and
compression on the first-cracking flexural properties of
UHPFRCs

Duy-Liem Nguyen1,*, Minh-Thuan Duong1

1 Ho Chi Minh City University of Technology and Education, Ho Chi Minh City, Vietnam
(liemnd@hcmute.edu.vn)

Abstract. This paper investigated the influence of elastic modulus under uniaxial
tension and compression on the first-cracking flexural properties of ultra-high-
performance fiber-reinforced concretes (UHPFRCs). The elastic modulus of a
uniaxial stress versus strain response curves is defined as the slope of linear por-
tion within the first crack and it refers to the stiffness of the tested material. Un-
like metal exhibiting same elastic modulus in both tension and compression, the
elastic modulus of UHPFRCs, observed from experimental tests in previous stu-
dies, were considerably different, and, they much influenced on the first-cracking
flexural resistances from sectional analysis. Relationships between compressive
strength and elastic modulus of UHPFRCs were also investigated and discussed.

Keywords: Ultra-High-Performance Concrete, Elastic Modulus, First Crack,

Flexural Resistance, Uniaxial Tension

1 Introduction
Ultra-high-performance fiber-reinforced concretes (UHPFRCs) is one of promising en-
gineered construction materials due to its high strength, toughness, durability and high
crack resistance [1-3]. The superior mechanical and material properties of UHPFRCs
are extremely attractive to structural engineers to apply them in civil and military infra-
structures for enhancing their resistance under severe environments and loads. Alt-
hough the flexural modulus of rupture (MOR) or the tensile post-cracking resistance of
UHPFRCs with its work-hardening accompanied by multiple micro-cracks is intensely
focused, the first-cracking resistance of UHPFRCs should be also studied to deeply
understand this material. Elastic modulus is one of mechanical parameters of material
which is defined as slope of the initial linear portion of a stress versus strain response
curve. The elastic moduli of uniaxial tensile and compressive responses are generally
identical for metal whereas they may be different for concrete or UHPFRCs owing to
much lower resistances under tension of these cement-based materials. In addition,
there is strong correlation between tensile and flexural behaviors of UHPFRCs [4].
Therefore, the difference of elastic moduli of uniaxial tensile and compressive re-
sponses of UHPFRCs may influence on the flexural first-cracking resistance of UHP-
FRCs. This situation was the motivation for this study focusing on the influence of
elastic moduli under uniaxial tension and compression on the first-cracking flexural
properties of UHPFRCs.

© Springer Nature Singapore Pte Ltd. 2020 343

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_52
344 D.-L. Nguyen and M.-T. Duong

2 Experiment

2.1 Materials
Table 1. Composition of mortar matrix of UHPFRC
Cement Silica Silica Silica Superplas- Water
(Type 1) fume sand powder ticizer
1,00 0.25 1.10 0.30 0.04 0.20

Table 1 provides composition of UHPFRC plain mortar. The UHPFRC plain mortar
was added a blend of 1.0% volume fraction of micro smooth fiber and 1.0% volume
fraction of macro twisted fiber. The twisted fiber had 30 mm length, 0.3 mm diameter
and tensile strength of 2428 MPa, while micro smooth fiber had 13 mm length, 0.2 mm
diameter and tensile strength of 2788 MPa. The compressive specimen was cylindrical-
shaped with diameter×height of 100×200 mm with gauge length of 100 mm, whereas
the tensile specimen was bell-shaped with rectangular section of 25×50 mm within
gauge length of 125 mm. The detail of mixing UHPFRC and curing specimens can be
referred to previous researches of the first author [2,3].
2.2 Test Set-up And Results

Compressive
specimen

Three LVDTs

a) Tensile test b) Compressive test

Fig. 1 Test set-up
All tests were conducted using a UTM machine under displacement control, as shown
in Fig 1. The speed of loading was 1 mm/min for all specimens. During the tensile test,
the load history was measured using a load cell attached on the specimen top while the
elongation history of the specimen was obtained by averaging two deformation histo-
ries measured from linear variable transformers (LVDTs) shown in Fig. 1a. During the
compressive test, the load history was measured by UTM while the elongation history
of the specimen was obtained by averaging three LVDTs shown in Fig. 1b.
Fig.2a and Fig.2b show the uniaxial tensile response [2] and the uniaxial compressive
response of UHPFRC [3], respectively: the studied UHPFRC demonstrates strain-hard-
ening performance under tension and can be modeled as linear performance under com-
pression because of dispersed and brittle behavior beyond the peaks. The mechanical
Influence of elastic modulus under uniaxial tension … 345

resistances of UHPFRC, including elastic moduli, under uniaxial tension and compres-
sion are provided in Table 2. According to Table 2, the ratio of elastic modulus, Et / Ec
, was about 6.03 times and Ec,UHPFRC  26.83  UHPFRC
1.5
fc',UHPFRC , where  UHPFRC is density of
UHPFRC,  UHPFRC  2600 kg/m3. This equation is quite different from that of NC:
3
Ec, NC  0.043  1.5
NC fc', NC , with  UHPFRC  2500 kg/m .

a) Uniaxial tensile response b) Uniaxial compression response

Fig. 2 Uniaxial stress versus strain responses of UHPFRC
Table 2. Mechanical resistances of UHPFRC
Tensile first-crack Tensile post-crack Compressive Utimate
cc (%) σcc (MPa) Et (MPa) pc (%) σpc (MPa) σcu (%) f’c (MPa) Ec (MPa)
0.025 7.56 272620 0.58 11.78 0.42 161.53 45210

3 Influence Of Elastic Modulus Under Uniaxial Tension And

Compression On The First-Cracking Flexure Of UHPFRCs
3.1 Section Analysis

Figs. 3a and 3b show the simplified model of uniaxial stress versus strain responses of
UHPFRC under uniaxial tension and compression, respectively. The uniaxial tensile
response of UHPFRCs is simplified under a trilinear model while the uniaxial compres-
sive response of UHPFRC is proposed to be linear as described in Fig. 3b because the
measured compressive response curve of UHPFRCs prior to the peak is almost linear
and the response beyond the peak can not be used [4]. In this study, a beam with rec-
tangular section having depth h , width b would be analyzed under the positive mo-
ment. Fig. 4 provides the distribution of the stress, strain along the depth of section.
346 D.-L. Nguyen and M.-T. Duong

Location of neutral axis ( c  kh ), curvature of beam (  ) and the first-cracking mo-

ment ( M cr ) are built by Eqs. (1-5).
 t  Et  t ,  c  Ec c ,    c   t ,    k (1)
c t
kh (1  k ) h (1  k )
 1 1 1 k 1 k2
C  2  cbkh  2  c Ecbkh  2  t (1  k ) Ec bkh  2 Ec bh t (1  k ) (2)

 y  2 kh
 C 3
 1 1
T  2  t b(1  k ) h  2 Et bh t (1  k ) (3)

 y  2 (1  k )h
 T 3

Static equilibrium for horizontal forces: C  T


k2 E
 t  constant  X  k
Et (4)
(1  k ) 2 Ec Ec  Et
1  2(1  k )h 2kh  1 (5)
M cr  M T  M C  T ( yT  yC )  Et bh t (1  k )    Et bh 2 t (1  k )
2  3 3  3

Stress Stress
pc, pc) cu, f’c)
Post-cracking point
Model
cc, cc)
Experiment
First-cracking point

Et Ec

Strain Strain

a) Uniaxial tensile response b) Uniaxial compression response

Fig. 3 Simplified model of uniaxial stress versus strain responses of UHPFRC
c c
C
Ec kh
Neutral
axis
h t Mcr
Et
T (1-k)h

t t
b

Fig.4 Strain and stress distribution along the depth of rectangular section
Influence of elastic modulus under uniaxial tension … 347

Since Et  Ec thus k  0.5 , i.e., the neutral axis near bottom and  t   c
Condition for the first crack occurs at bottom of beam:  t   cc and    k
  cu
c t
(1  k )

(4)   Et   and M  1  bh2 (1  k ) (6)

cr cc
cc cu
Ec 3
Condition for the top of beam fails:  c   cu and    (1  k )  
t c cc
k
(4)   Et   and M  1 f 'bh2 k (7)
cr c
cc cu
Ec 3

3.2 Application For The Investigated UHPFRC

Considering a beam with depth h=400 mm, width b=200 mm, tensile and compressive
properties of UHPFRC as provided in Table 2.
We have k  Et
 0.71
Ec  Et
Checking condition (6):  Et  0.025% 272.62  0.061%    0.42% : OK
cc cu
Ec 45.21
 The crack of beam occurs at bottom and M  1  bh2 (1  k ) =23.34 kN.m
cr cc
3

4 Conclusion

This investigation provides useful results from experimental test combined section
analys of the beam using UHPFRC. The following conclusions can be drawn from the
study:

1) Elastic modulus of uniaxial tensile response was higher the that of uniaxial compres-
sive response of UHPFRC. Thus the location of the neutral axis would be nearer the
bottom than the top.
2) That the beam produces the first crack at bottom or fails at top depends much upon
tensile first-cracking properties and compressive utimate properties. Condition for the
first crack occurs at bottom of beam:  Et  
cc cu
Ec
3) A model and equations were proposed to predict the first-cracking moment of beam
using UHPFRC regarding its uniaxial tensile and compressive responses.

Acknowledgments
This research is supported by Ho Chi Minh city University of Technology and Educa-
tion (Vietnam).
348 D.-L. Nguyen and M.-T. Duong

References
1. Nguyen, D.L., Ryu, G.S, Koh, K.T, Kim D.J.: Size and geometry dependent tensile behavior
of ultra-high-performance fiber-reinforced concrete. Composites: Part B 58, pp. 279-292
(2014).
2. Nguyen, D.L., Kim, D.J., Ryu, G.S, Koh, K.T.: Size effect on flexural behavior of ultra-
high-performance hybrid fiber-reinforced concrete. Composites: Part B 45, pp. 1104-1116
(2013).
3. Nguyen, D.L. and Kim, D.J.: Sensitivity of various steel-fiber types to compressive behavior
of ultra–high–performance fiber–reinforced concretes. Proceedings of AFGC-ACI-fib-
RILEM International Symposium on Ultra-High Performance Fibre-Reinforced Concrete,
UHPFRC 2017, PRO 106 - RILEM Pubications, October 2-4, Montpellier, France, pp 45-
52 (2017).
4. Nguyen, D.L., Thai, D.K., Kim D.J.: Direct tension-dependent flexural behavior of ultra-
high-performance fiber-reinforced concretes. Journal of Strain Analysis for Engineering De-
sign, Vol. 52(2) 121–134 (2017).
 Investigation on shear resistances of short beams using
HPFRC composited normal concrete

Duy-Liem Nguyen1,*, Van-Toi Do1, Minh-Phung Tran2 and Luu Mai3

1 Ho Chi Minh City University of Technology and Education, Ho Chi Minh City, Vietnam
2 Thu Dau Mot University, Binh Duong Province, Vietnam
3 Ho
Chi Minh City University of Transport, Ho Chi Minh City, Vietnam
(liemnd@hcmute.edu.vn, dovantoi.nv@gmail.com,
phungtm@tdm.edu.vn, mluuqn@gmail.com)

Abstract. This paper presents an investigation on shear resistances of short

beams using high performance fiber-reinforced concrete (HPFRC) composited
normal concrete (NC). Total twelve short beams with identical span-length/depth
ratio of 2 were tested under three-point bending. HPFRC and NC had their com-
pressive strength of 80 MPa and 20 MPa, respectively. HPFRC layer was placed
in extreme top or bottom fiber with various thicknesses; for each thickness, steel
reinforcing-bar and no reinforcing-bar embedded inside beams were examined.
The results of experimental test were evaluated and discussed.

Keywords: High-Performance Fiber-Reinforced Concretes, Shear Resistance,

Composited Beam, Shear Strain, Short Beam

1 Introduction
High performance fiber-reinforced concrete (HPFRC) has revealed its high mechanical
and cracking resistances with work-hardening behavior accompanied by multiple mi-
cro-cracks [1,2]. However, much flexural members have been designed using more than
one type of concrete as a composite system, e.g., deck slab has commonly been used
low-strength concrete while main beam has been used high-strength concrete. Besides,
the combination between normal concrete (NC) and HPFRC may minimize the con-
struction cost because HPFRC is much expensive than NC [3]. In previous study [4],
the moment resistances of composited beams with span-length/depth ratio of 3 using
HPFRC and NC jointly was reported. And, this investigation is the next step to explore
shear resistances of composited beams with span-length/depth ratio of 2. The shorter
span-length of beam was programed in this study for purpose of shear failure because
the shorter beam is, the more probable failure occurs due to shear force.

2 Experiment

2.1 Materials And Beam Design

© Springer Nature Singapore Pte Ltd. 2020 349

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_53
350 D.-L. Nguyen et al.

NC HPFRC

Composited beam

Beam 1A 50 Beam 2A Beam 3A Beam 4A Beam 5A Beam 6A

NC NC HPFRC

75

75
NC

100
150

150
HPFRC NC
100

HPFRC HPFRC NC

75

75
HPFRC

50
150 150 150 150 150 150

Beam 1B Beam 2B Beam 3B Beam 4B Beam 5B Beam 6B

50

NC NC
75

75
NC HPFRC
100
150

150
HPFRC NC
100

HPFRC HPFRC NC
75

HPFRC

75
50

150 150 150 150 150 150

Fig. 1 Layout of experimental test

Bottom cover to=25 P Stirrup Ø6 2Ø6

Top cover d’=25
150

NC

2Ø12
300
150
400

a) Steel reinforcement in series of type A b) Formwork preparation

Fig. 2 Steel reinforcement and formwork preparation
Table 1. Composition of mortar matrix of HPFRC
Cement Silica Silica Fly Super- Water
(Insee, PC40) fume sand ash plas-ticizer
0.80 0.07 1.00 0.20 0.04 0.26

Fig. 1 shows the layout of experimental test in this study with location and thickness of
HPFRC layer of all studied beams. The HPFRC was added a blend of 0.5% volume
fraction of micro smooth fiber and 1.0% volume fraction of macro hooked fiber. Table
1 provided composition of HPFRC plain mortar. The hooked fiber had 35 mm length
and 0.5 mm diameter while micro smooth fiber had 13 mm length and 0.2 mm diameter.
The compressive strength of HPFRC and NC using cylindrical-shaped specimen of
100×300 mm were 80 MPa and 20 MPa, respectively. All beams had identical dimen-
sion, width×height×span length of 150×150×300 mm, beam length of 400 mm. Fig. 2
shows the formwork preparation and steel reinforcement placed in Beam 1B to Beam
6B (series type B), they include 2 bars in tension with diameter of 12mm ( As =2.26
Investigation on shear resistances of short beams … 351

cm2, cover t0 =25mm) and 2 bars in compression with diameter of 6 mm ( As' =0.57 cm2,
cover d ' =25mm), yield strength of tensile steel f y =400 MPa, yield strength of stirrup
and compressive steel f y' =240 MPa. There was no steel reinforcement placed in Beam
1A to Beam 6A (series type A). NC and HPFRC were mixed immediately after each
other within 30 min. All specimens were cured in a water tank at 25°C for 28 days.
2.2 Test Set-up And Procedure
All tests were conducted under three-point bending as described in Fig 3a. The speed
of loading during the test was 1 mm/min for all beams. The load and deflection were
measured by UTM machine. As presented in Fig. 3b, the moment ( M  PL / 4 ) and shear
force ( V  P / 2 ) at mid-span under 3PBT, where P and  are the applied load and de-
flection at mid-span, respectively; L is span-length while b and h are width and height
of beam section, respectively.
L/2 L/2

P
X
O x 
Y

Diagram M due to P

M=PL/4

Diagram V due to P

V=P/2

a) Photo of test set-up b) Distribution of M and V for simple span

Fig. 3 Test set-up and distribution of moment and shear force

3 Testing Result And Discussion

3.1 Applied Load Versus Deflection Behavior Of Investigated Beams

Fig.s 4a & 4b show the load versus deflection responses of six beam of series A and six
beam of series B, respectively. Table 3 provides the parameters of tested beams includ-
ing load-carrying capacity (maximum load, Pmax ) and its corresponding deflection at
mid-span (  ). It was noticed that the measured deflections of studied beams might be
affected by creep and shrinkage owing to different concretes mixed. The orders in term
of Pmax were shown in Table 3 according to series A and B as follows:
+ In series A: Beam 1A (100.39 kN) > Beam 2A (93.90 kN) > Beam 3A (80.47 kN) >
Beam 4A (58.89 kN) > Beam 6A (57.63 kN) > Beam 5A (26.55 kN).
+ In series B: Beam 1B (233.57 kN) > Beam 2B (188.59 kN) > Beam 3B (158.49 kN)
> Beam 4B (148.90 kN) > Beam 6B (106.59 kN) > Beam 5B (89.04 kN).
352 D.-L. Nguyen et al.

There was a consistent in orders of series A and B, and, reductions in Pmax from Beam
1 to Beam 5 was completely suitable because HPFRC is stronger than NC and lower
thickness of HPFRC placed at beam bottom leads to low Pmax . Besides, compared with
Beam 3, Beam 6 had same thickness of HPFRC but produced a lower Pmax , this means
that HPFRC placed in tensile zone is more effective than in compessive zone.

a) Load-deflection response of series A b) Load-deflection response of series B

Fig. 4 Load versus deflection responses of investigated beams
3.2 Crack Due To Shear Force
Evidence of shear failure would be identified through inclined crack regarding theory
of elasticity. Inclined cracks were observed in all beams of series A as shown in Fig. 5
although their slopes were different according to beam type. The crack shape in series
B was not clear as shown in Fig. 6. There was neither slide nor disconnection between
NC and HPFRC layer in all tested specimens.

a) Beam 1A b) Beam 2A c) Beam 3A

d) Beam 4A e) Beam 5A f) Beam 6A

Fig.5 Photos of cracks appeared in series A
 Investigation on shear resistances of short beams … 353

a) Beam 1B b) Beam 2B c) Beam 3B

d) Beam 4B e) Beam 5B f) Beam 6B

Fig.6 Photos of cracks appeared in series B

3.3 Comparison Of Shear Recistance Regarding Theory And Experiment

Micro fiber
V
NC

HPFRC Ffb h

V
Macro fiber
Fig.7 Model of fiber bridging crack
Table 3 – Comparison of shear resistance regarding theory and experiment

Beam Pm ax  Test Theory1 Ratio1 Due to fiber Theory2 Ratio2

name (kN) (mm) Vu (kN) Vn (kN) Vu / Vn R fiber (kN) Vn fiber (kN) Vu / Vn fiber
1A 100.39 1.71 50.20 26.73 1.88 22.07 48.80 1.03
2A 93.90 1.87 46.95 22.27 2.11 14.72 36.99 1.27
3A 80.47 1.75 40.24 20.04 2.01 11.04 31.08 1.29
4A 58.89 1.74 29.45 17.82 1.65 7.36 25.17 1.17
5A 26.55 1.07 13.28 13.36 0.99 0.00 13.36 0.99
6A 57.63 2.13 28.82 20.04 1.44 11.04 31.08 0.93
1B 233.57 3.88 233.57 135.30 1.73 22.07 157.37 1.48
2B 188.59 3.37 188.59 130.84 1.44 14.72 145.56 1.30
3B 158.49 3.26 158.49 128.62 1.23 11.04 139.65 1.13
4B 148.90 2.70 148.9 126.39 1.18 7.36 133.75 1.11
5B 89.04 4.49 89.04 121.94 0.73 0.00 121.94 0.73
6B 106.59 3.57 106.59 128.62 0.83 11.04 139.65 0.76

In this study, the effect of steel fiber embedded on shear resistance was proposed using
Eqs. (1) & (2) based on [1,2,5].

Vnfiber  min(0.166 fc' bv dv  Rs  R fiber , fc' bv dv ) (1)

R fiber    F fb  cos      2  bh / tan    pc (2)
354 D.-L. Nguyen et al.

Where dv  0.8h is effective height of section, bv is effective width of section,

Rs  Av f y dv / s is shear resistance from stirrup spacing of s . Av and f y are the area and
yield strength of stirrup, respectively. R fiber is shear resistance from fiber bridging the
inclined crack with mechanism shown in Fig. 7. Assume that   45o and  pc  9.81 MPa
[4].  is a proposed factor considering the complicated function of fiber, e.g., the em-
bedded fiber not only resists pull-out force but also slip force, wall effect of direct-
tensile specimen, thus the value of  is proposed to be 0.2 in this study. Table 4 pro-
vides the comparative shear resistances regarding estimation and experiment with two
cases: a) not consider the contribution of fiber bridging crack (Theory1) and b) consider
it (Theory2). The estimated shear resistance in case b) was more consistent with exper-
iment results, specially for series A with no steel reinforcement.

4 Conclusion
1) Series A having no steel reinforcement failed due to shear force with their inclined
cracks whereas series B were not clear with various crack shapes. There was neither
slide nor disconnection between NC and HPFRC layer in both series A and B.
2) As the thickness of HPFRC layer increased, the load-carrying capacity of beam was
enhanced. The HPFRC placed in beam bottom produced higher load-carrying capacity
than that placed in the beam top, with use of same thickness of HPFRC.
3) A model and equations were proposed to explain and to predict shear resistance of
composited beams using NC-HPFRC jointly. The effect of fiber bridging shear crack
on shear resistance was noticeable and should be considered in design.

Acknowledgments
This research is funded by Vietnam National Foundation for Science and Technology
Development (NAFOSTED) under grant number 107.01-2017.322.

References
1. Naaman, A.E., Reinhardt, H.W.: Proposed classification of HPFRC composites based on
their tensile response. Mater. Struct. 39 547 – 555 (2006).
2. Naaman, A.E.: Tensile strain-hardening FRC composites: historical evolution since the
1960, Adv. Constr. Mater. Part II (2007) 181–202.
3. Špak, M., Kozlovská, M., Struková, Z., Bašková, R.: Comparison of Conventional and Ad-
vanced Concrete Technologies in terms of Construction Efficiency. Advances in Materials
Science and Engineering, Volume 2016, Article ID 1903729, 6 pages, http://dx.doi.org/
10.1155/ 2016/ 1903729 (2016).
4. Nguyen, D.L., Tong, D.N: Bending resistance of steel-bar reinforced concrete beam with
extreme compression zones using high-performance composite. Proceedings of the 4th Con-
grès International de Géotechnique - Ouvrages –Structures, CIGOS 2017, October 26-27,
Ho Chi Minh City, Vietnam, published by Springers in the volume 8, 2017, pp89-99.
5. AASHTO LRFD – 2014, LRFD Bridge Design Specifications of America.
 Application of Fluidized Power Coating for Propellers

Tuan Phan Anh, Huong Pham Thi Thanh

School of Transportation Engineering, Hanoi University of Science and Technology, Vietnam

tuan.phananh@hust.edu.vn

Abstract. Fluidized bed powder coating is a process where a pre-heated part is

dipped into a tank of fluidized thermoplastic powder, which is typically PVC or
nylon. This work creates a beautiful and durable conformal coating for the sur-
face of metal parts, machinery and equipment with a plastic film to protect
against corrosion and environmental effect. In the shipbuilding industry, the flu-
idized bed powder coating method has been applied to the interior details, shell
machinery and marine equipment. However, the application to the surface of
the propeller has not been popular. This paper presents a study on the applica-
bility of fluidized bed powder coating to surfaces of propeller. The authors have
conducted research and experiments for a three-blade ship propeller. Research
has compared the performance of the initial propeller and itself after applying
the fluidized bed powder coating method. From research study result, the au-
thors make an assessment and recommendations for the application of fluidized
bed powder coating to propeller blade surfaces in shipbuilding.

Keywords: Fluidized bed powder coating, Surface, Metal, Corrosion, Propel-

lers, Shipbuilding.

Introduction

The amount of used metal in the next half century on the world is estimated to be
about three times more than today. Together with the increasing of industrial pollu-
tion, the rate of corrosion and damage to metal material are also multiply increasing.
With typical climatic zones, erosion and destruction of metal material will occur more
severely, with faster rates in underdeveloped areas, particularly in Asia, Africa, and
Central East.
Therefore, people need to find technologies for metal coating that is highly re-
sistant to the corrosion and does not significantly affect to the structure or fabrication
of machineries. Especially for machinery parts that direct and frequent contact with
the environment, parts of high complexity, and the coating does not cause adverse
effects on the environment.
Current, some methods of coating metal surface have been applied in practice for a
long time and have a good effect such as metalized, paint coating, surface passivation,
blackening, fluidized bed powder coating, etc. In which, the fluidized bed powder

© Springer Nature Singapore Pte Ltd. 2020 355

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_54
356 T. P. Anh and H. P. T. Thanh

coating method has been studied by many scientists and already applied various in
industrial products.
This paper mentions to an application of fluidized bed powder coating method on
shipbuilding technology. The application is coating for a propeller surface.

Fluidized power coating technology

Powder coating technology for metal products has been developed since the 1950s of
last century. At the early applications, powder coating was used to coat pipes for cor-
rosion protection and electric motor parts for insulation. These coatings were applied
using a fluidized-bed process in which heated parts was dipped into a vat containing
power suspended in air. In this process, once the particles of power contact and ad-
here to the heated metal parts, they begin to soften and flow into a smooth. Most of
the coating applications for fluidized-bed in this period were vinyl and epoxy powers.
In 1970s and 1980s decades, as a result from interesting in power grew, the
industry responded with technological improvements with many new resin systems.
Power coatings had formulated in a virtually limitless range of colors, glosses and
textures. The two major types of power coatings with popular using are thermoplastic
power coating and thermosetting power coating.
In nineties, Leong et al. developed experimental and numerical studies about correla-
tion among film properties and substrate materials and geometries in CHDFB (con-
ventional hot dipping) coating process [1]. M. Barletta at al. has developed a method
of matte finishes in electrostatic and conventional hot dipping fluidized bed coating
process [2]. In that study, a cylindrical fluidized bed, 250mm as diameter and 200mm
as static bed height, of epoxy-based powders, working in fast regime, up to 15m3/h of
flow volume rate, was used for coating works. The research found that the surface
roughness and waviness were found to be strictly dependent on curing temperature.
Increasing curing temperature means lower minimum viscosity values of the resin
onto the coated substrates during the first moment of the curing process. B. Piero at
al. has studied a novel concept of dense-fluidized bed coating of objects where the
effectiveness of coating is promoted by the intentional and controlled establishment of
shear flow around the object [3]. In that study, the fluidized powder is sheared by the
controlled oscillatory motion of the object with respect to the fluidized bed.
Today, fluidized powder coating technology for metal products is popular applica-
tion in industry.

Application of fluidized power coating for ship propellers

3.1 Fluidized power coating

In this study, a three-blade ship propeller has been used for applied fluidized power
coating method. Propeller is a device that creates propulsive force to help a ship mov-
ing on water surface. The propeller for applying fluidized power coating method is
200 mm in diameter. It material is copper. The detail parameters of the propeller are
shown in Table 1.
Application of Fluidized Power Coating for Propellers 357

Table 1 Parameters of ship propeller for fluidized power coating

Diameter D (mm) Blades Z Blade area ratio (%) Pitch P (mm) Length (mm)

260 3 40 208 60

Fig. 1 shows the pictures of the ship propeller before and after applied fluidized
power coating method. The propeller is cover with a plastic layer of 60 m. It could
be observed that the surfaces in both propellers, before and after made fluidized pow-
er coating, are smooth.

Fig. 1. Pictures of ship propellers, before and after fluidized power coating
3.2 Evaluation performance of propellers via CFD

To evaluate the performance of the propellers before and after fluidized power coat-
ing, CFD (Computational Fluid Dynamics) is used in this research. CFD method ena-
bles engineers to make better, faster decisions across the widest range of flu-
ids simulations [4]. The CFD had conducted with follow process.
Governing Equations: The system of two phases flow surrounding propellers is
governed by Navier - Stokes equations, taking into account k-ε turbulence model [5],
[6].

u u u 1 p 1  τ xy τ xz
 (1)
u v w       Fx
x y z ρ x ρ  y z 
v v v 1 p 1  τ yx τ yz  (2)
u v w       Fy
x y z ρ y ρ  x z 
w w w 1 p 1  τ zx τ zy  (3)
u v w       Fz
x y z ρ z ρ  x y 
358 T. P. Anh and H. P. T. Thanh

Domain and Boundary Conditions: Computational domain for CFD simulation in

this study is showing in Fig.2. It is a double tube domain. The inside domain is a tube
with 280 mm diameter and 100 mm in length. The outside domain is a tube with 800
mm diameter and 2,800 mm in length. For boundary conditions in this study, propel-
ler is rotation with different speeds from 100 rpm to 500 rpm, inlet is water velocity,
outlet is gauge pressure of 0 Pa, temperature is set at 300oK, density of water is 998.2
kg/m3 and viscosity of water is set at 1.003x10-3 kg/(ms).

Fig. 2. Domain and boundary conditions in CFD

Pressure characteristic: After carrying CFD, pressure characteristic in simulation
domain is shown in Fig.3. In this figure, Fig.3a shows the pressure contour and Fig.3b
shows the pressure transform of the water flow. The pressure of water flow is discon-
tinuous at before and after the blade.

(a) (b)
Fig. 3. Pressure characteristic of propeller in CFD
Velocity characteristic: Velocity characteristic in simulation domain is shown in
Fig.4. In this figure, Fig.4a shows the velocity contour and Fig.4b shows the velocity
transform of the water flow. The velocity of water flow around the propeller area has
a rapid increasing and creating a peak at the propeller blade disk.
Propeller characteristic: After CFD simulation, propeller characteristics of the orig-
inal propeller before applied fluidized power coating method is built as in Fig.5. In
Application of Fluidized Power Coating for Propellers 359

this figure, three charts of thrust curve (KT), torque curve (10KQ) and performance
curve () are drew based on different values of advance coefficient (J).

(a) (b)
Fig. 4. Velocity characteristic of propeller in CFD

Fig. 5. Propeller characteristic

Discussions

In this study, the authors have conducted experiments for checking the performance
of the propeller before and after applied fluidized power coating method. For conduct-
ing experiments, the propellers are put in free water and driven for rotation by a three
phase electrical motor. Speeds of the motor are controlled by an inverter. Performance
of the propellers have evaluated via their thrusts. After carried experiments, the charts
the thrust curves are built and shown in Fig. 6.
During the limitation of the equipment for conducting experiments, authors have
carried experiments at low speed of propeller rotation. The rotational speeds of pro-
pellers in the experiments are from 100 rpm to 500 rpm and no inlet velocity of the
water flow.
In Fig.6, the continuous curve is thrust chart in CFD simulation of the original
propeller, the triangle marks are the values of the thrust in experiments of the original
propeller with its rotational speeds respectively, the rhombus marks are the values of
the thrust in experiments of the original propeller with its rotational speeds respective-
ly.
360 T. P. Anh and H. P. T. Thanh

Fig. 6. Thrust of propellers in CFD and experiments

Conclusions

The experimental result shows that the thrusts of the propellers are not so different
between the original propeller and the fluidized power coating propeller. This means
the fluidized power coating method could be applied to ship propellers.
Although the fluidized power coating method can be applied to ship propellers to
prevent their surface corrosion, people need to study and test more about the strength
and longevity of the plastic coated layer when ship propellers working in sea envi-
ronment before it to be widely applicable in shipbuilding industry.

References
1. Leong, K.C., Lu, G.Q., Rudolph, V.A., A comparative study of the fluidized bed coating
of cylindrical metal surfaces with various thermoplastic polymer powders, Journal of Ma-
terials Processing Technology, ISSN: 0924-0136, Vol. 89–90, 354-360 (1999).
2. Barletta, M., Bolelli, G., Guarino, S., Lusvarghi, L., Development of matte finishes in elec-
trostatic (EFB) and conventional hot dipping (CHDFB) fluidized bed coating process, Pro-
gress in Organic Coatings, ISSN: 0300-9440, Vol. 59, 53-67 (2007).
3. Piero, B., Francesco, P., Piero, S., Shear-assisted fluidized bed powder-coating, Proceed-
ing of The 13th International Conference on Fluidization - New Paradigm in Fluidization
Engineering, 1-8 (2010).
4. ANSYS FLUENT Tutorial Guide, Release 14.5, ANSYS Inc., Southpointe 275 Technolo-
gy Drive Canonsburg, PA 15317, (2012).
5. Tuan, A. T., A study on hovercraft resistance using numerical modeling, Applied Mechan-
ics and Materials, ISSN: 1662-7482, Vol. 842, 186-190, (2016)
DOI:10.4028/www.scientific.net/AMM.842.186
6. Tuan, P. A., Quang, V. D., Estimation of Car Air Resistance by CFD Method, Journal of
Mechanical Engineering, ISSN:0866-7136, Vol. 36, 235-244, (2014), DOI:
https://doi.org/10.15625/0866-7136/36/3/4176
 Design of a Drinking Water Disinfection Systems using
Ultraviolet Irradiation and Electrolysis Cell

Tuan Phan Anh and Tan Nguyen Minh

Hanoi University of Science and Technology, Vietnam

tuan.phananh@hust.edu.vn

Abstract. In this paper, the authors develop a concept design of a safety drink-
ing water disinfection system powered by wind energy. The idea for carrying
this study is to develop a safety Drinking Water Disinfection System (DWDS)
for rural/island/mountain and removed areas in Vietnam as such as in ASEAN
developing countries where the national electric grids may be lacked. The
drinking water disinfection will use electricity from a small wind turbine. The
study will focus to develop two main parts: development an energy-efficient
UV treatment technology as main disinfection step with a small-scale electroly-
sis cell and development a small wind turbines for supporting the electric to the
UV treatment and electrolysis cell. A prototype product of the safety drinking
water disinfection powered by wind energy was created.

Keywords: Drinking water, Disinfection, UV treatment, Electrolysis cell, Wind

turbine.

1 Introduction

Although Vietnam has improved its water supply situation in the past few decades,
many rural parts of the country who are often the poorest communities, have not seen
significant improvement. It is reported that only 39% of the rural population has ac-
cess to safe water and sanitation [1]. The rural population has moved from using sur-
face water from shallow dug wells to groundwater pumped from private tube wells.
It could be said that safety drinking water is very important for our live. In the to-
tal picture of water needed in Vietnam, there are about 20% of its population has not
yet access to clean and safety water for their daily life. There are currently about 17.2
million Vietnamese people, equivalent to around 21.5% of the total population, who
are using direct water from drilled wells that has not been done water treatment.
Lots of places in rural areas in Vietnam are lack of safety drinking water and some
of these places are also lack of national electric grids. This paper presents a develop-
ment of a safety drinking water disinfection system for rural/island/mountain and
removed areas in Vietnam powered by wind energy. Vietnam is an ASEAN country
that wind energy could be used for electrical generation. It is expected that wind pow-
er will be strong development in near future in Vietnam [2].

© Springer Nature Singapore Pte Ltd. 2020 361

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_55
362 T. P. Anh and T. N. Minh

2 Drinking Water Disinfection Systems

2.1 The ultraviolet disinfection in the DWDS
The ultraviolet (UV) treatment process is an extremely quick physical process. Ultra-
violet light mutates and degrades DNA (or deoxyribonucleic acid) of organism. DNA
is the part of the cell that gives an organism its instructions on how to function and
reproduce. UV disinfection technology has been investigated by several authors.
Bergmann did a research study on drinking water disinfection by direct electrolysis –
state of the art [3]. Song et al. studied an application of ultraviolet light-emitting di-
odes (UV-LEDs) for water disinfection [4].
In the UV disinfection process, water is purified as it runs through a stainless
steel pipe chamber that contains a special UV - producing lamp in center. As the wa-
ter flows past the lamp, the microbes in the water receive a lethal dose of UV. The
water is then safe to drink. However, different organisms require different levels of
UV energy to disrupt their DNA. This energy level is known as a UV dose. In our
design of the DWDS, the UV lamp has maximum power of 25W. The lamp uses
24VDC source. To identify the level of energy for supporting to the UV lamp, a sensi-
tive sensor for checking organiems density and water impurity is setting up at the
stainless steel pipe chamber. An UV control unit that uses the sensitive sensor data for
analysis and control the power to the lamp is putted in a plastic box. Fig. 1 shows the
UV disinfection concept and device.

Fig.1. UV disinfection device

2.2 The electrochemical water disinfection in the DWDS

An electrochemical water disinfection system is a device that eradicates microorgan-
isms by using an electric current passed through the water under treatment by means
of suitable electrodes. Fig. 2 shows that the electrochemical water disinfection con-
cept and device.
In the electrochemical water disinfection process, at the phase boundary between
the electrodes and the water, the electric current leads to the electrochemical produc-
tion of disinfecting species from the water itself or from species dissolved in the wa-
ter. In our design of the DWDS, the electric current leads to the electrochemical pro-
duction of disinfecting by chloride that oxidised to free chlorine. The electrochemical
Design of a Drinking Water Disinfection Systems … 363

water disinfection device uses 24 VDC (volts DC) source. The electrodes are ex-
changed every a delay set time. The delay set time for changing the electrodes is de-
pending on the density of natural chlorine of the raw input water. Therefore, it is
needed to test water impurity before setting the delay time for exchanging the elec-
trodes.
For disinfection, free chlorine will be created by the following chemical reaction
Cl 2  H 2 O  H   Cl   HOCl (1)

Fig.2. Electrochemical water disinfection device

2.3 Electrical Support for DWDS

The power for running the UV disinfection and the electrochemical water disinfection
is generated from wind energy by a small vertical wind turbine. This wind turbine is
designed for generating of 300W at the wind speed of 12 m/s. Electric that converted
from wind energy by the wind turbine will be charged and store in a battery bank of
24V/100Ah.
The wind turbine for the DWDS works as a vertical wind turbine. The advantage
of the vertical wind turbines is that they can get the wind for generating electric from
any direction of wind flows. This simplifies their design and eliminates the problem
imposed by gyroscopic forces on the rotor of a convectional machine as the turbine
tracks the wind. The vertical axis of rotation also permits mounting the generators and
drive train at ground level for the big power generators. The disadvantages of vertical
wind turbines is that it is quite difficult to control power output by changing direction
of the wind flows to their blades as horizontal wind turbines.

3 Energy converting by wind turbines

The air flows of wind go through blades as a disk could be demonstrated as in Fig. 3.
The entrance and exit pressures of the air flows is p∞. The pressures p∞ are pressures
at locations before and after the air flows go through the blade disk that do not have
364 T. P. Anh and T. N. Minh

any effect from the disk. The interaction area of the blade disk is A. The speed of the
air flows go through the blade disk is v.
When the air flows go through the blade disk, their pressures at front and back of
the disk surfaces will be changed to differently. The pressure will increase from p∞ to
p+ in the left of the blade dick and then change from p+ to p- in the right of the blade
disk. After that, the air pressure increases from p- to p∞ again. Difference from the
pressure characteristics, air flows velocity are only reducing after blow through the
blade disk.

Fig.3. Air flow through a blade disk

The energy converting from air flows of wind as the power of the turbine could be
calculated as:
dE 1 2 dm 1
P  v  ρAv 3 (2)
dt 2 dt 2
where, E is kinetic energy; t is time; m is mass and  is air density.

4 Designed wind turbine for the DWDS

For supporting electricity to the drinking water disinfection system, in this design we
use a vertical wind turbine. The turbine blade area for converting electricity of a verti-
cal wind turbine could be demonstrated as in Fig. 4. Then, from Eq. 2, the converting
power of a vertical wind turbine could be estimated as:

P  CpρHDv3 (3)

where, Cp is power coefficient of the wind turbine. This coefficient is depended on

type of wind turbine and manufacturing technology.
Design of a Drinking Water Disinfection Systems … 365

In our design, the Cp coefficient of the wind turbine is around 0.3. Output power
chart of the wind turbine generator is shown in Fig. 5. The output power chart shows
that the wind turbine will work well with wind velocity from 6 – 14 m/s. When the air
velocity is stronger than 14 m/s, the output power of the wind turbine could not in-
crease because the limitation of the maximum output power of the electrical genera-
tor. When the wind velocity is too strong, the wind turbine will start its brake system
to protect itself.

Fig.4. Vertical wind turbine dimension Fig.5. Output wind turbine power chart

5 Discussions
In this study, a real safety drinking water disinfection powered by wind energy has
been created as a prototype product. It is a bi-disinfection system, electrochemical and
UV disinfections. The concept and diagram of the main blocks of the safety drinking
water disinfection system powered by wind energy are shown in Fig. 6.

Fig.6. Block diagram of the safety DWDS powered by wind energy

The DWDS takes row water from underground sources. A pump will deliver the row
water to the electrochemical disinfection. After solution by electrochemical disinfec-
tion, water will be delivered to the UV disinfection. After those disinfected solutions,
row water becomes safety drinking water. The safety drinking water is stored in a
pressured tank. It is a five liter capacity tank and located inside a stainless steel cup-
board with drinking taps.
366 T. P. Anh and T. N. Minh

The prototype version of the DWDS is installed at a primary school. The DWDS
is including three main blocks. The first block is the wind turbine and its towel. The
second block is drinking cupboard. The last block is a steel cupboard that stores all
remain devices of the DWDS inside, including batteries, electrochemical disinfection,
UV disinfection, pump, electronic control circuits. In this block, there are two screens
for monitoring the disinfection solution and indicate electric output power of wind
turbine.
The designed maximum capacity of the DWDS is 120 l/h. However, since the ca-
pacity of water disinfection of the DWDS could be set by a controllable valve, for the
good quality of water disinfection, the capacity of the DWSD should be set at a range
from 80 to 100 l/h. The DWDS is now operating for testing the stability of working
condition. The tests are conducted for monitoring the pH, microorganisms and chlo-
ride of the DWDS’s water product.

6 Conclusion
In this study, a concept of safety drinking water disinfection supply system for such
rural/island/mountain and removed areas, in consideration of energy demand with
cost efficiency has been designed. The drinking water disinfection system is designed
with low energy demand and low costs for suitable life standard of people who lives
in rural/island/mountain and removed in Vietnam.
A prototype product version of the safety drinking water disinfection system was
made. The maximum capacity of the drinking water disinfection system is 120 l/h.
However, for the good quality of water disinfection, the capacity of the DWSD should
be set at a range from 80 to 100 l/h. The drinking water disinfection system has been
operated for testing real working condition at a primary school in rural in middle of
Vietnam. It is found from this research study that the drinking water disinfection
powered by wind energy is very useful and suitable for development in ru-
ral/island/mountain and removed areas in Vietnam as such as in ASEAN countries.

Acknowledgement
This research study is funded by the grant numbered B2018-BKA-06SP from Minis-
try of Education and Training (MOET).

References
1. World Bank Group, Exposure to Floods, Climate Change, and Poverty in Vietnam, 2016.
2. Phan A. T., Nguyen D., Pham T. T. H., 2018. “Research and development of wind power
in Vietnam”. VSOE 2018, Lecturer Notes in Civil Engineering, ISSN: 2366-2557, Spring-
er, pp. 270-275, Vol. 18, DOI: https://doi.org/10.1007/978-981-13-2306-5, 2018.
3. Bergmann, H., 2008. “Drinking water disinfection by direct electrolysis—state of the art”.
proceeding of the CHISA 2008 and 8th European symposium on electrochemical engineer-
ing, Prague, pp. 24-28
4.Song, K., Mohseni, M. and Taghipour, F., 2016. “Application of ultraviolet light-emitting
diodes (UV-LEDs) for water disinfection: A review”. Water research, 94, pp.341-349.
 The effect of mineral admixture on the properties of the
binder towards using in making pervious concrete
Nguyen Van Dong1 , Pham Huu Hanh1, Nguyen Van Tuan1, Phan Quang Minh1,
Nguyen Viet Phuong1
1
National University of Civil Engineering, Ha Noi, Vietnam
dongnv@nuce.edu.vn

Abstract: Pervious concrete has not been much studied and applied in Vietnam, the
outstanding feature of this concrete is the open pore system for water flowing through. The two
most important indicators for evaluating the quality of pervious concretes are the permeability
coefficient and compressive strength but these are two opposite functions. Research into the
properties of this type of concrete, such as workability, strength, etc ... is one of the most
effective measures that will improve the quality of the binders. The purpose of this article is to
improve the characteristic of binder by separately using silicafume (SF) and fly ash (FA) with a
content of 10%; 20% and 30% and by using the combination compound of 10% SF with 10-
30% FA. The two typical properties of binders are: the viscosity of the binders through the
flowing time of Marsh cone, the instantaneous viscosity determined by the SV-10 viscosimeter
and the cstrength of the binders.

Keywords: pervious concrete, mineral admixture, viscosity, strength

1. Introduction
According to the American Concrete Institute (ACI), pervious concrete (PC),
which is a kind of concrete with very low slump or no slump, using an uncontinuous
grading of aggregate, includes portland cement, coarse aggregate, a small amount of
fine aggregates or not sung fine aggregates, water and additives. After hardening,
concrete has a system of open pores of size from 2 mm to 8 mm, from which the
water easily flows through. The porosity of concrete varies 15-35%, compressive
strength 2.8-28MPa. The permeability coefficient of this concrete depends on
porosity, aggregate size and usually about 1.35-12.17 mm/s [1-3].
Pervious concrete is made up of 2 main components, which are load-bearing
aggregates and form a pore system; the binder binds aggregate particles together and
creates strength for concrete. The characteristics of binder paste greatly affect the
properties of pervious concrete. Therefore, the research on binder paste is very
important in improvement of pervious concrete quality.
Binder used in pervious concrete is made of cement, mineral additives, chemical
additives and water. In fact, there have been a number of studies about using different
mineral additives such as SF, FA, GBFS, RHA [4-8], ... for increasing strength,
improving workability or reducing amount of cement. However, these studies often
assess the effect of additives on the properties of pervious concrete, not on binders of
this kind of concrete. Meanwhile, binder’s viscosity greatly affects the workability
and the quantity of the binder paste, the strength of binder directly affects the strength
of pervious concrete. Therefore, in this paper, the research on the effect of mineral
additives on the viscosity of the strength of binder will be presented.
2. Materials
Cement in this research is a PC40, has a density of 3.1 g/cm3; fly ash from Pha Lai

© Springer Nature Singapore Pte Ltd. 2020 367

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_56
368 N. V. Dong et al.

thermal power plant with a specific weight of 2.12 g/cm3; silica fume Elkem has a
density of 2.21 g/cm3; The superplasticizer ACE388 with a density of 1.21 g/cm3 and
water according to TCVN 4506 : 2012 are used in this research. The composition of
cement, fly ash and silica fume is shown in Table 1.
Table 1 - Chemical composition of Portland cement, Fly ash and Silica fume
Oxide Portland cement (C) Fly ash (FA) Silica fume (SF)
SiO2 22.58 58.38 93.45
Al2O3 5.72 25.12 0.92
Fe2O3 3.45 7.01 0.52
CaO 61.23 0.84 1.57
MgO 0.65 0.70
K2O 3.28
Na2O 0.30
SO3 0.14
L.O.I, (%) 1.71 3.89 4.20
B.E.T, (m2/g) 0.345 0.298 1.85
3. Methodology
3.1. Determination of flow ability with Marsh cones
Marsh cone according to American standard ASTM C939 [9] is used for flow test
of binder pastes.
3.2. Determination of viscosity
The viscosity of binder pastes was determined by using a viscometer SV-10 A&D.
4. Results and discussion
In order to study the effect of additives on the viscosity of binder paste and the
strength of binder, this research uses 2 types of mineral additives such as silicafume
(SF) and fly ash (FA) to increase the strength and replace partially an amount of
cement with the content is shown in Table 2.
Table 2 – Composition of binder
Name of composition Ratio W/B FA, % SF, %
ĐC 0 0
10FA 10 0
20FA 20 0
30FA 30 0
10SF 0 10
0.2; 0.22; 0.24
20SF 0 20
30SF 0 30
10SF10FA 10 10
10SF20FA 20 10
10SF30FA 30 10
4.1. Effect on viscosity of binder pastes.
- Effect of SF content and W/B ratio to viscosity of binder pastes
The viscosity of binder pastes determined by Cone Marsh and SV-10 is shown in
Figure 1 below. The results in figure 1 showed that the viscosity of the binder paste at
all W/B ratios was lowest when using 10% SF. When the SF content was increased to
20% and 30%, the viscosity of binder pastes increased rapidly, especially at the ratio
The effect of mineral admixture on the properties … 369

W/B of 0.20 (the lowest value in 1000 5

Viscosity (W/B=0.20)
the experimental range). Similar Viscosity (W/B=0.22)
to the viscosity, the test results 800
Viscosity (W/B=0.24)
4

Viscosity of binder, mmPa.s

Flow time (W/B=0.20)
for the flow time of binder paste Flow time (W/B=0.22)

log(flow time,s)
by Marsh cone are shown in Flow time (W/B=0.24)
600 3
Figure 1, the experimental results
also show that when using 10%
SF in the mixture, the flow time 400 2
of binder pastes decreased when
compared to the control sample 200 1
(ĐC). However, when the
amount of SF increased by 20% 0 0
and 30%, the flow time also ĐC 10SF 20SF 30SF
increased, and at the lowest W/B Content of SF, %
ratio (W/B= 0.20) the flow time Figure 1. Effect of SF content on viscosity
of the binder paste increased.
- Effect of FA content and W/B ratio to viscosity of binder paste.
Experimental results of the
1000 5
effect of FA on viscosity and Viscosity (W/B=0.20)
flow time of binder paste are Viscosity (W/B=0.22)
Viscosity (W/B=0.24)
shown in Figure 2. Experimental 800 4
Viscosity of binder, mmPa.s

Flow time (W/B=0.20)

results show that when using FA Flow time (W/B=0.22)

log(flow time,s)
Flow time (W/B=0.24)
in binder paste, the viscosity and 600 3
flow time of binder paste will be
reduced. At the same time, when
the corresponding FA content is 400 2
increased, the time of flow
through the Marsh cone and the 200 1
instant viscosity of paste B will
decrease. However, when the FA
content is increased (in this study 0 0
ĐC 10FA 20FA 30FA
with FA=30% and W/B
Content of FA, %
ratio=0.24), it was observed that
there is a phenomenon of Figure 2. Effect of FA content on viscosity
segregation in the mixture but
when combined with SF this phenomenon does not occur any more. Thus, the
combination of SF and FA has a mutual effect, limiting the disadvantages of each
mineral additive mixture and improving the viscosity of binder paste. Based on the
experimental results, this research select and fix the SF content of 10% to evaluate the
effects of combination of SF with FA in binder paste.
- Effect of mineral additive mixture of FA and SF on viscosity of binder paste
Experimental results of viscosity and flow time by Marsh cone of binder paste
when using a combination of FA and 10% SF with different W/B ratios are shown in
Figure 3.
370 N. V. Dong et al.

Experimental results 1000 5

Viscosity (W/B=0.20)
showed that the Viscosity (W/B=0.22)
combination of SF and 800 Viscosity (W/B=0.24) 4

Viscosity of binder, mmPa.s

FA significantly Flow time (W/B=0.20)
improved the viscosity Flow time (W/B=0.22)
of binder paste. The 600 Flow time (W/B=0.24) 3

log(flow time,s)
viscosity of the paste
decreased by about 8%
when compared to the 400 2
sample using only 10%
SF and about 10 % when 200 1
compared to samples
using only 10% FA.
When FA content 0 0
increases, the viscosity ĐC 10SF10FA 10SF20FA 10SF30FA
of binder paste also Content of mineral additive
decreases, when the FA Figure 3. Effect of mineral additive mixture FA
content is 20% and 30%, and SF mixture on viscosity
the viscosity of binder paste B was decreased respectively of 22% and 35% when
compared to composition using only use 10% SF. When there is a combination of
10% SF and 30% FA, binder paste does not occur segregation phenomenon, thus the
disadvantage of using only FA additive could be overcomed.
Similarly, the experimental results showed that when there was a combination of
SF and FA, the flow time of paste B through Marsh was also significantly improved
when compared to the control sample and samples containing only SF. When fixing
10% SF and increasing FA content, the flow time of binder paste decreases, which
means that the flow rate increases and the viscosity of paste B decreases.
Based on the above results, it is found that when using a combination of two
additives, FA and SF, there is a mutual effect between these two types of mineral
additives, reducing the viscosity and flow time of binder paste when compared to
When using the additive wall separately, this effect is explained as follows: SF
mineral additive has a smaller density than cement so when SF replaced with the
same mass, the volume of fine particles increased, then this increased amount of paste
will improve the workability of binder. On the other hand, the SF with spherical shape
and an average particle diameter of about 0.15µm reduces internal friction between
the particles in the mixture, also known as the "ball bearing" effect, helps larger
particles easily slip onto each other.
When using FA with increasing content, the viscosity and flow time of binder
paste decreases gradually. This is explained by the fact that FA has a spherical shape,
the surface is relatively smooth, so the amount of water to be adsorbed on the surface
was reduced, thus increases the amount of free water in binder paste, reducing the
flow time and viscosity of the paste. The result of this effect can increase the amount
of FA used to replace cement with a large amount in concrete. However, according to
the research results, when the FA content used up to 30%, binder paste occurs the
segregation of water, this phenomenon will be limited when there is a combination
with 10% SF.
 The effect of mineral admixture on the properties … 371

4.2. Affect the compressive strength of B

- Influence of SF content and W/B ratio to strength of binder
The effect of SF content and W/B ratio on compressive strength of binder is
shown in Figure 4.
120 120
Compressive strength, MPa

Compressive strength, MPa

100 100
80 80
60 60
W/B=0.2
40 40 W/B=0.2
W/B=0.22
20 W/B=0.22
20 W/B=0.24 W/B=0.24
0 0
0 10 20 30 0 10 20 30
Content of SF, % Content of FA, %
Figure 1. Effect of SF content and W/B ratio Figure 2. Effect of FA content and W/B ratio
to intensity of B to intensity of B
Experimental results show that, when using additive SF, the strength of binder is
significantly improved, especially when using 10% SF content and W/B ratio = 0.20.
The strength was increased about 40% when compared to the control sample. The role
of SF in binder is related to puzolanic reaction (reaction of SiO2 with Ca(OH) 2 to
form C-S-H and fill the pores in binder matrix). But when SF is increased to 20%, the
rate of increase in strength become slowed down and when it increased to 30% SF, it
tended to decrease because SF acts as an inert additive in the mixture at this high
content.
- Influence of FA content and W/B ratio to strength of binder
The effect of FA content and W/B ratio on compressive strength of B is shown
in Figure 5. Experimental results showed that when using 10% FA with all W/B
ratios, the strength of binder increased insignificantly. It is because FA is not a very
good reactive additive. When it is used with low content, it disperses the water in
binder system; on the other hand they adsorb a certain amount of water, then re-
release water for the hydration process 120
of cement. But when its content is up
Compressive strength, MPa

to 30%, the strength decreases rapidly, 100

and the beneficial effects of FA cannot 80
compensate for a significant reduction
in the amount of cement. 60
- Effect of mineral additive
40
mixture and W/B ratio to W/B=0.2
compressive strength of binder 20 W/B=0.22
The effect of mineral additive W/B=0.24
0
mixture and W/B ratio on compressive Ĉ& 10SF10FA 10SF20FA 10SF30FA
strength of binder is shown in Figure
Content of mineral additive
6. Experimental results showed that,
when using 10% SF in combination
with 10% FA, the compressive Figure 6. Effect of mineral additive mixture,
strength of binder increased W/B ratio to the compressive strength of B
372 N. V. Dong et al.

insignificantly when compared to the sample using only 10% SF. If increasing FA
content to 20%, the strength of binder will decrease, especially to 30% FA, the
strength decreases rapidly, because at this time the cement content in the mixture is
only 60%, the ratio of SF/C=0.167, therefore, the role of SF will be reduced gradually
by reducing the content of Ca(OH)2 to form CSH mineral.
5. Conclusions
From the results of experiments and analysis on binders using 2 types of mineral
additives used in pervious concrete, some conclusions are drawn as follows:
- The reasonable content of SF is 10%. It not only improves the properties of
binder paste but also significantly increases the strength of binders (about 40%
increase compared to the control sample).
- When using only FA to replace cement in binders, the reasonable content of FA
is 20%, This value will not significantly reduce the strength and cause the
phenomenon of water separation in binder paste.
- When combining SF and FA, they will have a mutual effect such as not causing
the segregation phenomenon, maintaining good workability, and increasing the
strength of binders. The reasonable amount of these two additives is 10% SF with
20% FA.
The results of binder give more information for designing and choosing pervious
concrete such as: calculating the maximum amount of binder, calculating the porosity,
permeability coefficient, strength, ... of the pervious concrete.
References
1. Tennis, P.D., M.L. Leming, and D.J. Akers, Pervious concrete pavements. Portland Cement
Association, Skokie, Illinois, & National Ready Mixed Concrete Association, Silver
Spring, Maryland, 2004.
2. Yukari Aoki, Development of pervious concrete. A thesis submitted to fulfilment of the
requirements for the degree of Master of Engineering, University of Technology, Sydney
Faculty of Engineering and Information Technology 2009.
3. Ghafoori, N. and S. Dutta, Building and nonpavement applications of no-fines concrete. a,
Building and nonpavement applications of no-fines concrete, 1995: p.286-289.
4. V.G.P., Experimental investigation and theoretical modeling of silica fume activity in
concrete. Cement and Concrete Research. 29, p. 79-86.
5. Minju Jo, L.S., Marleisa Arocho, Juliana St John, Sangchul Hwang, Optimum mix design
of fly ash geopolymer paste and its use in pervious concrete for removal of fecal coliforms
and phosphorus in water. Construction and Building Materials (2015).
6. Natalia I. Vázquez-Rivera, Juliana N. St John, Omar I. Molina-Bas, Sangchul S. Hwang,
Optimization of pervious concrete containing fly ash and iron oxide nanoparticles and its
application for phosphorus removal. Construction and Building Materials 93 (2015) 22–28.
7. Tawatchai Tho-in, V.S., Prinya Chindaprasirt, Chai Jaturapitakkul, Pervious high-calcium
fly ash geopolymer concrete. Construction and Building Materials, 30 (2012) 366–371.
8. Saeid Hesami, S.A., Mahdi Nematzadeh, Effects of rice husk ash and fiber on mechanical
properties of pervious concrete pavement. Construction and Building Materials, 2014
(Faculty of Civil Engineering, Babol Noshirvani University of Technology, 47148-71167
Babol, Iran).
9. Standard Test Method for Flow of Grout for Preplaced-Aggregate Concrete Copyright ©
ASTM, 100 Barr Harbor Drive, West Conshohocken, PA 19428-2959, United States.
10. Nicolas Roussel, R.L.R., The Marsh cone: a test or a rheological apparatus? Cement and
Concrete Research. 35(5), p. 823–830.
 Horizontal response of base-isolated buildings supported
to high damping rubber bearings

Nguyen Anh Dung1, Le Trung Phong1 and Tran Minh2

1
Thuyloi university, 175 Tay Son–Dong Da–Ha Noi, Viet Nam
2
State Audit Office of Vietnam, No. 116 Nguyen Chanh–Cau Giay–Hanoi, Vietnam
dung.kcct@tlu.edu.vn

Abstract. Seismic isolation is one of the most efficient techniques to protect

structures against earthquakes. Rubber bearings are suitable for low-rise and
medium-rise buildings due to its durability and easy fabrication. This paper pre-
sents the horizontal response of a six-storey base-isolated building using high
damping rubber bearings (HDRBs) under two ground motions of earthquakes as
types I and II in JRA (2002) by finite element analysis. In this analysis, these
bearings are modelled by the bilinear hysteretic model which is indicated in
JRA and AASHTO. Comparison of horizontal response including base shear
force and roof level acceleration between the two cases: base-isolated building
and fixed-base building is carried out to evaluate the effectiveness of the use of
HDRBs on the protection of buildings from earthquakes. The numerical results
show that the peak value of roof floor acceleration of the fixed-base building is
two times higher than that of the base-isolated building, and the floor accelera-
tions depend on the peak values of ground acceleration. In addition, the step-by-
step design procedure for determining the size of HDRBs used for buildings is
also presented in this paper.

Keywords: High damping rubber bearings; seismic responses; earthquakes;

buildings.

1 Introduction

A technique called seismic isolation has emerged as a practical and economical alter-
native to the conventional approach. The main concept of seismic isolation is to de-
couple a structure from the horizontal components of the ground motion by inserting
structural elements (e.g. isolation bearings) between the substructure and the super-
structure. Isolation bearings, which have low horizontal stiffness, can shift the natural
periods of structures to reduce the acceleration response, while its damping capability
can diminish the response of displacement. Various types of isolators have been man-
ufactured with a similar objective to provide a natural period shift and additional en-
ergy dissipation to structures. Recently, high damping rubber bearings (HDRBs) have
been widely used in Japan due to their high flexibility and high damping capability. In
the manufacturing process of high damping rubber material, a large number of fillers
(about 30%) including carbon black, silica, oils and some other particles are added

© Springer Nature Singapore Pte Ltd. 2020 373

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_57
374 N. A. Dung et al.

during the vulcanization process [1, 2] in order to improve the desirable material
properties, such as the strength and damping capabilities. They present high damping
capability, and bearings defined as HDRBs should provide an equivalent viscous
damping of at least 10% [3]. Gu and Itoh [4] estimated that the change of the equiva-
lent horizontal stiffness of HDRBs is only about 10-25% after 100 years. The purpose
of this paper is to investigate the effectiveness of base-isolated building using HDRBs
compared to the fixed-base building under ground motions of earthquakes. Further,
the effect of peak value and the time duration of ground accelerations on the horizon-
tal response of the building is also investigated in this paper.

2 Basic principles of base isolation

Fig. 1. Acceleration response spectrum in Fig. 2. Displacement response spectrum

function of the damping (as defined by EC8 in function of the damping (as defined by
[5]) EC8 [5])

The relationships between seismic response of structures and their periods are pre-
sented in Fig. 1&2 [5]. Based on these relationships, the principle of seismic isolation
is developed. The basic principle of base isolation is that the response of a structure is
modified so that the ground below is capable of moving without transmitting minimal
or no motion to the structure above. For this purpose, base isolation is inserted be-
tween the base of a structure and its foundation. The isolation is characterized by a
high flexibility in large earthquakes. This flexibility must be such as to shift the fun-
damental response frequencies of the structure to values that are below the range
where the seismic motion of soil is amplified. In this way, the soil motion is filtered,
because most energy is withdrawn from the frequency range that characterizes the
vibration modes of the superstructure. In principle, a structure may be isolated in both
the horizontal and the vertical directions. However, the application of seismic isola-
tion is usually limited to horizontal directions. The reasons are that seismic loads are
largely less critical in the vertical direction, compared to the horizontal; also, possible
rocking due to three-directional isolation must be avoided [3]. The use HDRBs has
proven, even in recent earthquakes, to be a very efficient technique to protect struc-
tures from earthquakes. HDRBs with their high flexibility and high damping capabil-
ity can achieve two effects: (1) To shift the own period of the structures reducing in
Horizontal response of base-isolated buildings … 375

such a way the seismic response and (2); To dissipate energy reducing again the re-
sponse and the displacement in Fig. 1&2.

3 Numerical analysis of a base-isolated building supported to

HDRBs

3.1 Prototype building and high damping rubber bearings

The case study building is a six – story reinforced concrete frame structure. The plan
dimensions are 21.6m x 18m and total height is 24.5m. The thickness of the floor is
100mm. Cross column section is 450x450mm. The layout of structural elements at
each floor is symmetrical in both plan directions. The critical damping of superstruc-
ture is assumed to be 5% in all the modes. HDRBs are placed between the column-
ends and substructures. One rubber bearing is provided under each column. In this
study, these bearings are modelled by bilinear hysteretic model as shown in Fig. 3.
The model is presented in some specifications [7, 8].

Fig. 3: Bilinear model for the bearings

[7, 8]

The step-by-step design procedure for determining the size of HDRBs and total thick-
ness of rubber in bearings are proposed as below:
Step 1: To determine the ground acceleration a g by Eq (1)
ag   I agR (1)
Where: agR is the peak value of ground acceleration on soil type A; I is the im-
portance coefficient of the building.
Step 2: Evaluation of the horizontal elastic acceleration spectrum at the period of 1
T 
second (S1) by Eq: TC  T  TD : S1  Se T   ag .S..2,5.  C  (2)
T 
Step 3: To determine the region coefficient F at the period of 1 (s) in Table 11.4-2 of
ASCE/SEI 7-10.
Step 4: To calculate dynamic coefficients SM1 and SD1 by Eq (3) & (4)
2
SM 1  F S1 (3); S D1  S M 1 (4)
3
Step 5: To apply the parameters of BD or BM based on the effective viscosity coeffi-
cient D.
Step 6: To select the period of the building at the horizontal displacement design
value, Td by Eq: 3T f  Td  3  s  (5)
Tf is the natural period of the superstructure.
376 N. A. Dung et al.

Step 7: Evaluate the effective horizontal stiffness of the bearings, K eff by Eq (6)
where W is the total vertical force caused by the dead load and live loads.
2
W W  2 
Td  2 => K eff   (6)
K d ,min g g  Td 
Step 8: To evaluate the design horizontal displacement of the bearings, DD by Eq (7)
 1 S T
DD   2  D1 d (7)
 4  BD
D D
Step 9: To determine the total rubber thickness (tr) by Eq:   D  t r  D (8)
tr 
Where  in the design displacement value following design standards
Keff tr
Step 10: To calculate the total cross-sectional area of the bearings, A  (9)
G
Where G is the shear modulus of HDRBs.
A
Step 11: To calculate the cross-sectional area of one bearing, Ai by Ai  (10)
n
Where n is the number of columns.
Step 12: To determine the design parameters of the bilinear model
HDRBs in this analysis are the same bearings in [5], in where, the bilinear parameters
C1 , C2 , and  cr are determined from the 1st cycle of sinusoidal loading tests and they
are presented in Table 1.
Table 1. Parameters in Nguyen (2017) Table 2. Design bilinear parameters
C1 (MPa) C2 (MPa) cr (MPa) K1 (N/mm) K2 (N/mm) Qy (N)
17.29 1.136 1.215 21613 1420 48600

These design parameters in Table 1 are converted into the above new design dimen-
A A
sions of HDRBs by Eqs: K1  b C1; Q y  Ab cr ; K 2  b C2 (11)
h h

Where Ab is cross-section area of the bearing, h is the total thickness of rubber.

Finally, the dimension of bearing is 200x200mm, the total thickness of each bearing is
32mm, and details of design parameters of the bearings are given in Table 2.

3.2 Ground motions of earthquakes and the analysis method of base-isolated

building supported to HDRBs
Fig. 4 (a) and (b) present the ground acceleration histories, which are designed for
earthquakes level-2 type-I and II by [8].
The time history analysis of both base-isolated and fixed-base buildings is carried out
using the computer program SAP2000 v14.2.2. A full three-dimensional representa-
tion of the structure is used in time history analysis of both base-isolated and fixed-
base buildings. Each isolator is modelled as a bilinear hysteretic element. The follow-
ing assumptions are made for the analysis of base-isolated building:
Horizontal response of base-isolated buildings … 377

• The superstructure is elastic at all time, the non-linear behaviour is restricted in

isolators only.
• All frame substructures are connected at each floor level by a diaphragm, which is
infinitely rigid in its own plane.
• Each floor has three degrees of freedom (two translations and one rotation) at-
tached to the center of mass of each floor.
• The isolation devices are rigid in the vertical direction and each has negligible
torsion resistance.
• Effect of infilled walls on stiffness of the structure is negligible. These are consid-
ered in computation of mass only.

(a) (b)
Fig. 4. Typical ground acceleration histories: (a) type-I, (b) type-II

3.3 Seismic response of the building

(a) (b)
Fig. 5. Base shear force at Joint 10: (a) type-I earthquake (b) type-II earthquake

(a) (b)
Fig. 6. Roof floor acceleration by (a) Earthquake type I (b) Earthquake type II
Table 2: Effectiveness of base isolation with HDRB in Y-axis

Peak value of roof floor accele-

Base shear at Joint 10 (KN)
Earthquakes ration (m/s2)
Fixed-base Base-isolated Fixed-base Base-isolated
Type-I (Kanto, 1923) 1374.16 582.57 10.39 4.91
Type-II (Kobe, 1995) 1229.04 470.83 13.91 6.18
378 N. A. Dung et al.

The comparison of seismic response in Y-axis of the fixed-base building to that of

base-isolated building is given in Fig. 5&6 and Table 2. The continuous lines present
the response of the fixed-base building and the hide lines describe the response of the
base-isolated building in these figures.

4 Conclusions

A seismic analysis of the six-storey building is carried out to investigate the effective-
ness of the use of HDRBs to protect buildings from earthquakes. This paper also pro-
poses a design procedure to determine the size of HDRBs. This procedure can support
to engineers to design base-isolated buildings using HDRBs.
The peak values of roof floor acceleration of the fixed-base building are 2.14 times
and 2.25 times higher than those of the base-isolated building under earthquakes type-
I and type-II, respectively. Moreover, the base shear forces of the fixed-base building
are 2.36 times and 2.61 times higher than the forces of the base-isolated building un-
der the earthquakes type-I and type-II, respectively. The base shear forces of the base-
isolated building depend on not only the peak values of ground acceleration, but also
the time duration of ground acceleration of earthquakes. These results show that using
HDRBs in buildings can improve the seismic capability of the structure against earth-
quakes.

References
1. Kelly, J.M.: Earthquake resistant design with rubber. 2nd edition, Springer-Verlag Berlin
Heidelberg, New York (1997).
2. Yoshida, J., Abe, M., Fujino, Y.: Constitutive model of high-damping rubber materials.
Journal of engineering mechanics, 130(2), pp 129-141 (2004).
3. Marioni, A.: The use of high damping rubber bearings for the protection of the structures
from the seismic risk. Jornadas portugesas de engenharia de estruturas, Lisbon, Portugal
(1998).
4. Gu, H., Itoh, Y.: Ageing effects on high damping bridge rubber bearing. Proceeding of the
6th asia-pacific structural engineering and construction conference, Kuala Lumpur, Malay-
sia (2006).
5. EC8.: Eurocode 8: Design of structures for earthquake resistance Part 1: General rules,
seismic actions and rules for buildings (EN 1998-1:2004). European Committee for Stand-
ardization (2004).
6. Nguyen Anh Dung (2017). “Effect of modeling on seismic response prediction of a pier
with high damping rubber bearings under earthquakes”. The 10th national Conference on
Mechanics, Hanoi, 8-9/12/2017, pp 191-197 (2017) (In Vietnamese).
7. American association of state highways and transportation officials-AASHTO.: 3rd Edi-
tion. Washington DC: Guide Specification for Seismic Isolation Design (2010).
8. Japan road association-JRA.: Specifications for highway bridges. Part V: seismic design.
Tokyo: Maruzen (2002).
 A Study on Behavior of Reinforcement Concrete Beam
using the Recycled Concrete

Anh-Thang Le1, Thanh-Hung Nguyen1, Cong-Vu-Duc Phan2

1
Faculty of Construction, HCMC University of Technology and Education, Vietnam
2
Student, Faculty of High Quality, HCMC University of Technology and Education, Vietnam
thangla@hcmute.edu.vn

Abstract. The demand for concrete materials in construction becomes huge due
to the development of the economy and urbanization in Vietnam. Recently, re-
cycled aggregate concrete has been studied for partly replacing natural stone of
concrete in the worldwide. It could save cost for land filling, and con-serve the
national resources such as river sand and aggregate. Recycling concrete gives
the benefit for environment. However, it may affect to the strength and perfor-
mance of the construction containing the concrete waste. The paper is a discus-
sion based on the experiments of reinforcement concrete beams with and with-
out recycled concrete. Testing results are simulated by the ATENA program on
exploring the behavior of beams. The analysis and experiment results show that
replacing 20% of natural stone with re-cycled aggregate concrete does not af-
fect both the bearing capacity and performance of beams.

Keywords: recycled concrete beams; simulation of a beam; ATENA 3D;

theflexural behavior of the beam.

1 Introduction

Construction demolition has created a huge amount of waste concrete, which raises
disposal cost due to the limited capacity of land filling. This has caused many other
issues such as the management of waste, environmental pollution, exhaustion of natu-
ral resources. Therefore, many scientists in the world have focused on the use of recy-
cled aggregate concrete [1]. Waste from demolished building has been recycled
around the world. Especially in European Union countries, the waste from demolition
building could be up to 45% [2]. In Vietnam, the recycled concrete obtained from
demolished building is still limited due to lack of researching of this problem.
The early research on structural performance of the recycled aggregate concrete
(RCA) was published in Japan. KhaldounRahal [3] compared RCA and natural ag-
gregate concrete (NAC). The concrete samples were tested with a compressive
strength of 25 to 30Mpa. There were the conclusions that the compressive strength of
the block after 28 days, and the direct shear strength of RCA achieved an average of
90% those of NAC with the same mixing ratio. Sami W.Tabsh et al. [4] researched
that using RCA with a grade of 50 MPa that could give compressive strength and
concrete tensile strength equivalent to natural aggregate concrete.

© Springer Nature Singapore Pte Ltd. 2020 379

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_58
380 A.-T. Le et al.

Researchers had various conclusions related the percent of RCA replacement that
causes reducing strength, wider cracks opening, larger deflection in the construction
components. Ajdukiewicz et al. [1] used partial and full recycled aggregate for rein-
forcement concrete beams. All the beams had the rectangle section of 200x300 and
2600 mm long. They tested with two longitudinal reinforcement ratios of 0.90% and
1.60%. It was reported that the RCA beams had slightly (3.5% in average) lower mo-
ment capacity and higher deflection compared with the conventional concrete (CC)
beams. Mahdi Arezoumandi et al. [5] investigated of the flexural strength of full scale
reinforced concrete beams constructed with both RCA (100%) and CC. The RCA
beams showed comparable flexural capacity of the CC beams.
Bai and Sun [6] used RCA of eight to ten years-old with different replacement levels
of 50%, 70%, and 100%. They observed similar crack pattern, but deflection and
crack width increased with the increment of RCA replacement level. Knaack and
Kurama [7] tested on the 150x230 mm cross section and 2000 mm long beams. They
used RCA from late 1920s with both 50% and 100% replacement level. They reported
higher deflection for the RCA beams, but they concluded that the existing analytical
models and specification for CC beams can be used for the RCA beams. Based on the
above literatures, it is possible to use recycled concrete up to 100%. We always ex-
pected that 100% of RCA could be reused in new concrete structure because of envi-
ronmental benefit. However, specific deficiency of full-scale concrete beams needs to
be investigated in this article.
In this study, the 20% of RCA replacing natural stone are studied on both bending
strength of the beams and the beam performance. This ratio of RCA is selected be-
cause experimental results on the testing samples indicate that there do not signifi-
cantly differences in compressive and tensile strength between concrete with 20% of
RCA and conventional concrete using natural stone. The considered grade of concrete
is 25MPa, a popular grade of concrete using in the building in Vietnam recently. The
ATENA software is used for the steel rebar exploration.

2 Experiments

Concrete mixtures are conducted according to TCVN 4453:1995 [8]. The considering
grade of concrete is 25 MPa, which is tested based on the compressive strength of
concrete cubic samples of 150x150x150 after 28 days according to TCVN 3118:1993
[9]. The selected concrete grade is corresponding to the concrete of class C20/25 of
Eurocode 2.

2.1 RCA material

The tested cubic samples of 150x150x150, are milled to product the large particles of
RCA. Compressive strengths of all milled cubic samples are controlled above 30
MPa. Natural aggregates are gradually replaced by aggregates from waste concrete at
the rates of 20%, 40%, 60%, 80%, and 100% of the aggregate weight. Compressive
strengths are investigated and summarized in Figure 1a. The diagram shows the com-
 A Study on Behavior of Reinforcement Concrete Beam … 381

pressive strength of the recycled concrete specimen decrease as the ratio of RCA in-
creasing. In case of 20% RCA, the compressive strength reduced a little, about 1.7%.
The maximum reduction is 28.9% as the rate of RCA reaching 80%. As increasing the
rate of RCA up to 100%, experimental results show the compressive strength of con-
crete slightly improved. The compressive strength of recycled concrete with 100%
RCA higher that of 80% RCA is due to the homogeneous use of aggregate.
The average tensile strengths of concrete with and without 20% RCA was 1.81 and
1.93 Mpa, respectively. Thus, there were 6% reducing of tensile strength as 20% RCA
replacing for natural aggregate in the concrete mixture. The ratio of 20% RCA in
concrete beams was selected in this study.
160
300
(a) 140
Compressive Strength

250
120

Total load (kN)

200 RCA1
(daN/cm²)

100
150 80 RCA2
CC
100 60
40
50
20
0 0
0 20 40 60 80 100 0 5 10 15 20 25 30 35 40
RCA (%)
Beam deflection (mm)
Fig. 1. (a) Concrete aggregate after milling and the compressive strength of concrete speci-
mens according to the ratios of RCA, (b) Deflection of beams according to load levels.

2.2 Beam testing results

The tested reinforcement concrete beams with size of 200x300x3300mm, Figure 2a.
Two beams with 20% of RCA are named as RCA1 and RCA2. The beam with 100%
natural aggregate is named as CC.
28 27 26 25 24
(a)23 22 21 20 19 18 17 16 15 14 13 12 11 10 9 8
(b)
2 7

5 6
1 3
4

Fig. 2. (a) Diagram of the reinforcement concrete beam (sizes in mm), (b) steel rebar
numbering.

The ultimate loads of the beams of RCA1, RCA2, and CC are 136 KN, 138.8 KN, and
145.1 KN, respectively. Figure 1b shows the relationship between the displacement at
the mid-span of beams and total loading. The error between the ultimate load of the
beam with 100% natural aggregate and the lowest ultimate load of the beam with 20%
of RCA is 6.27%.

2.3 Crack patterns

382 A.-T. Le et al.

Figure 3 presents the failure modes of beams. RCA1 and RCA2 have a total of 12
cracks and 10 cracks, respectively, while the CC has a total of 9 cracks. The cracks
are almost located in the middle zone of beams. The crack paths are developed fol-
lowing the perpendicular path to the tension stress caused by bending. It indicates that
the small amount RCA could cause more cracks in the failure mode caused by bend-
ing. It could be said that the performance of a beam in long term could be affected by
the present of RCA.

Fig. 4. Stress distribution of reinforce-

Fig. 3. Crack patterns ments in the beam.

3 Model of the Beam

The finite-element analysis software, ATENA, has been selected for the modeling of
the testing beams. The layout of reinforcement inside the beam and the stress
distribution in the concrete beams after analyzing are captured in Figure 4.
The input parameters of both concrete and steel materials are summarized in Table
1. Those values are selected based on the concrete of class C20/25 (Eurocode 2).

Table 1. The input parameter of the models

Parameter Symbol RCA CC

Compressive strength of concrete (MPa) fc 19.55 21.25
Tensile strength of concrete (MPa) ft 1.941 2.052
Concrete elastic modulus (MPa) E 27070 28060
fy(Φ16) 300 300
Yield strength of steel (MPa)
fy (Φ14) 300 300
Elastic modulus of steel (GPa) E 200 200

3.1 Verification for the beam models

Figure 5 shows the comparison of the relationship between total load and deflection at
mid-span of the beams of RCA and CC. It could be observed that there is not a big
gap between the curves of experience and those of modeling. It indicates that the
models created in the environment of ATENA software could be used for the further
A Study on Behavior of Reinforcement Concrete Beam … 383

investigation.
160 160
140
(a) (b)
140
Total load (kN)

Total load (kN)

120 120
100 100
RCA1 CC
80 80
RCA2 Model of CC
60 60
Model of RCA
40 40
20 20
0 0
0 5 10 15 20 25 30 35 40 0 5 10 15 20 25 30 35 40
Beam deflection (mm) Beam deflection (mm)
Fig. 5. Comparison of experiment results and those obtained from modeling: (a) the curves of
RCA, and (b) the curves of CC.

3.2 Stress distribution in the reinforcement of the concrete beams

Table 2 summaries the distribution of stress in the beam models. The first column is
the numbering of longitudinal reinforcements, which are illustrated in Figure 2b. The
stresses showed in the table are the steel stress values obtained corresponding to the
ultimate loads. It could be observed that the steel bars in both the compressive zone
and tensile zone of the beam could be reached to the yield-point of steel. The differ-
ence of steel stresses in models of RCA, and CC could be up to 80%. It shows that the
stress distribution inside the longitudinal steel bars was affected by the using 20% of
RCA in the beam.

Table 2. The stress in the longitudinal steel bars

Steel stress in CC (MPa) Steel stress in RCA (MPa) Error (%)

No.
Max Min Max Min Max Min
1 300.0 -8.9 300.0 -8.8 0 1
2 132.1 -300.0 26.9 -300.0 80 0
3 300.0 -19.3 300.0 -11.9 0 39
4 300.0 -11.0 300.0 -10.4 0 6
5 300.0 -5.8 300.0 -8.8 0 53
6 300.0 -15.0 300.0 -11.3 0 25
7 97.3 -300.0 46.4 -300.0 52 0

4 Conclusions

Study on recycled concrete aggregate partly replacing the natural aggregate in rein-
forcement concrete components is the current trend in Vietnam. A study on the beams
containing 20% of RCA is performed in which the qualities of RCA are carefully
control. The following conclusions are made:
384 A.-T. Le et al.

- Both compressive and tensile strengths of concrete are reduced a little even
though the compressive strength of concrete producing RCA has the higher
compressive strength of the desire concrete. The experimental results show
the reduction of compressive strength is 1.7%, while the reduction of tensile
strength is 6%.
- The performance of reinforcement concrete beam with 20% of RCA could
also be affected. The experimental results show that the deflections increase
comparing to that of the conventional beam. Besides, the number of crack-
ing appeared in the beams is also increased. On the other words, RCA could
affect to the long-term performance of the reinforcement concrete compo-
nents.
- Based on the simulation models in the ATENA software, the distribution of
stress in the longitudinal steel bars of a beam changed in various models of
the beams with and without RCA. The maximum error of steel stresses
could be up to 80%. The longitudinal steel bars were affected by the using
20% of RCA in the beam.

Acknowledgments
The authors are grateful for the financial support as well as the experimental equip-
ment of the HCM University of Technology and Education, and the unconditional
help from the faculty of civil engineering so that we can complete this research.

References
1. Ajdukiewicz AB, Kliszczewicz AT.: Comparative tests of beams and columns made of re-
cycled aggregate concrete and natural aggregate concrete. J Adv Concr Technol 5(2), 259–
73 (2007).
2. European Commission (DG ENV), Service contract on canagement of construction and
demolition waste -SR1: Final report task 2. BIO intelligence service, Paris, 2011.
3. KhaldounRahal.: Mechanical properties of concrete with recycled coarse aggregate. Build-
ing and Environment 42(1), 407-415 (2007).
4. Sami W.Tabsh, Akmal S.Abdelfatah.: Influence of recycled concrete aggregates on
strength properties of concrete. Construction and Building Materials 23(2), 1163-1167
(2009).
5. Mahdi Arezoumandi, Adam Smith, Jeffery S. Volz, Kamal H. Khayat.: An experimental
study on flexural strength of reinforced concrete beams with 100% recycled concrete ag-
gregate. Engineering Structures 88, 154–162 (2015).
6. Bai WH, Sun BX.: Experimental study on flexural behavior of recycled coarse aggregate
concrete beam. Applied Mechanics and Materials 29, 543-548 (2010).
7. Knaack, Adam M., and Yahya C. Kurama.: Behavior of reinforced concrete beams with
recycled concrete coarse aggregates. Journal of Structural Engineering 141.3 (2014).
8. TCVN 4453:1995, Monolithic concrete and reinforced concrete structures - Codes for con-
struction, check and acceptance.
9. TCVN 3118:1993, Heavyweight concrete - Method for determination of compressive
strength.
 Geochemical modelling for prediction of chloride
diffusion in concrete exposed to seawater

Hoang Long Nguyen1, Van Quan Tran1, Long Khanh Nguyen1, Tuan Anh Pham1
Quoc Trinh Ngo1 and Van Loi Giap1
1
University of Transport Technology, No. 54 Trieu Khuc Street, Thanh Xuan District, Hanoi,
Vietnam
quantv@utt.edu.vn

Abstract. Steel corrosion is one of major problems that affect the durability of
reinforced concrete structures. Chloride concentration is the key parameter for
the steel corrosion risk of reinforced concrete exposed to seawater. Therefore, the
durability of reinforced concrete can be evaluated by the prediction of chloride
concentration into the reinforced concrete. Numerous numerical models have
been developed to predict the chloride concentration in concrete however these
numerical models have not yet fully simulated the nature of the chemo-physical
processes taking place between the concrete and seawater. In this study, the pre-
diction of chloride concentration is carried out by using the geochemical model
including chemo-physical process. The geochemical model can improve the ac-
curacy of the durability prediction of reinforced concrete. The accuracy of dura-
bility prediction is proved by the comparisons between modelled results and ex-
periments results found in the literature.

Keywords: Reinforced concrete, seawater, durability, chloride, geochemical

model.

1 Introduction

The main deterioration cause for reinforced concrete structures exposed to marine
environment is the steel corrosion induced by chloride penetration into concrete. Owing
to the expensive costs of repair and maintenance it is highly requested to develop reli-
able models intended to predict the chloride concentration. A number of chloride
transport models can be found in the literature, such as models based on Fick's second
law or multispecies ionic transport equations [1], [2] or physically and chemically cou-
pled model [3]. However, the model simulated the reinforced concrete submerged into
a synthetic solution such as: NaCl + NaOH solution. The seawater had not been con-
sidered in these models. Therefore, the durability of reinforced concrete exposed to
seawater could not be rigorously assessed. The geochemical model can take into ac-
count chemo-physical process: the ingress of seawater and the interactions between
concrete and seawater such as: interaction between hydrated cement and agents (Cl-,
SO42-, CO32-, Mg2+, Ca2+…), aqueous complexation leading to the mineral dissolu-
tion/precipitation, pH modification, leaching of concrete. Therefore, the accuracy of

© Springer Nature Singapore Pte Ltd. 2020 385

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_59
386 H. L. Nguyen et al.

durability prediction would be improved by using the geochemical model. First step,
the model approach: transport equation, principles of thermodynamic equilibrium, and
kinetic control will be presented. Secondly, input data of model will be also described
such as: mineralogical composition of hydrated cement paste, exposure conditions. Fi-
nally, some experiments data from the literature are compared with the modelled results
to prove the feasibility of geochemical model.

2 Modelling Approach

2.1 Transport equations

In a saturated environment without convection phenomena, ion transport is governed

by diffusion. Therefore, the principle of mass balance equation can be written in a gen-
eral form in one-dimension model [4], [5]:

𝛛𝐌𝐣 𝛛𝐉𝐣 (1)

=− + 𝐪𝐣
𝛛𝐭 𝛛𝐱

Mj, Jj and x denote the molar accumulation of species j in the aqueous phase (mol.m−3
of material), the flux of species j in solution (mol.m−2.s−1) and the distance of diffusion
(m), respectively. The qj source or sink of species (mol.m−3.s−1) is interpreted by the
removing or the releasing of species j.
In porous media, considering the ionic interactions are taken into
account in the source term, the flux of a species j, Jj, can be described by the first Fick’s
law:

𝛛𝐂𝐣 (2)
𝐉𝐣 = −𝐃𝐞
𝛛𝐱
De effective diffusion coefficient of species j (m2 ⋅s-1)
2.2 Thermodynamic equilibrium

The interaction between the ionic species and the mineral species leads to precipita-
tion/dissolution of minerals. The mineral saturation ratio Ω𝑚 can be expressed as:
𝐍𝐜
(3)
Ω𝐦 = 𝐊 −𝟏
𝐬,𝐦 ∏(𝛄𝐣 𝐂𝐣 )𝛎𝐦𝐣 𝐦 = 𝟏, … , 𝐍𝐩
𝐣=𝟏
Where m is the indice of the mineral species, K s,m is the equilibrium constant.Cj is
the molal concentration of primary species in the solution (mol. kg −1 ), i.e. the species
that are supposed to move in the solution/ Other ions (often more complex) are taken
into account when ionic complexations are calculated (see equation 4). νmj is the stoi-
chiometric coefficient of the primary species, γj is the activity coefficient of the ion j ,
Nc , Np are the number of corresponding primary species and mineral species. The state
of equilibrium (or disequilibrium) of mineral species in the solution is controlled by the
mineral saturation index ISm , as follows:
Geochemical modelling for prediction of chloride … 387

𝐈𝐒𝐦 = 𝐥𝐨𝐠 𝛀𝐦 (4)

For a given mineral species, the solution is in equilibrium with the mineral species
if ISm = 0. The solution is under-saturated and the mineral species can still dissolve if
ISm < 0. Finally, the solution is super-saturated and the mineral species may be pre-
cipitated if ISm > 0. Aqueous complexes are formed by interactions between primary
species in the solution. These reactions are assumed to be at local equilibrium. By using
the mass action law, the concentration of aqueous complexes can be expressed as a
function of the concentration of primary species, as follows:
𝐍𝐜
(5)
𝐂𝐢 = 𝐊 −𝟏 −𝟏
𝐜,𝐢 𝛄𝐢 ∏(𝛄𝐣 𝐂𝐣 )𝛎𝐢𝐣 𝐢 = 𝟏, … , 𝐍𝐱
𝐣=𝟏

where 𝐶𝑖 is the molal concentration of the aqueous complexes i (mol. kg −1 ). 𝛾𝑖 , 𝛾𝑗

are the activity coefficients. 𝐾𝑐,𝑖 is the equilibrium constant of aqueous complexation.
𝑁𝑥 is the number of aqueous complexes considered in the solution.

Thermodynamic database includes these parameters, the thermodynamic database

THERMODDEM of Blanc et al [6] is applied in this paper.

2.3 Kinetic control

A requirement of control rate chemical reaction under kinetic
constraints is necessary. The kinetic laws of mineral dissolution/precipitation is
used in this study, given by Lasaga et al. [7] .The rate expression is functions of
the primary species:
𝛈
𝐫𝐧 = 𝐟(𝐂𝟏 , 𝐂𝟐 , … , 𝐂𝐍𝐜 ) = ±𝐤 𝐧 𝐀𝐦𝐬,𝐧 |𝟏 − 𝛀𝛉𝐧 | n=1,..,Nq (6)

Where positive values of r indicate dissolution and negative values precipitation,

k is the rate constant (mol.m-2.s-1). Ams is the specific reactive surface area
(m2.g-1), Ω is the mineral saturation ratio defined in (eq. 2). When the reaction
approaches equilibrium state, the Ω approaches 1, the chemical reaction rate
tends toward zero. The parameters θ, η can be determined from experiments, they
are usually taken equal to one due to lack of measurement. q is the number of
minerals under kinetic constraints. The kinetic parameter of cement hydrated was ex-
tracted from Baur et al. [8].

3 Required input data

Simulations are carried out basing on the experiment conditions of Henrik et al. [9] and
Tang [10] such as: mix design of concrete, effective chloride diffusion coefficient. Hen-
rik et al. [9] measured the total chloride concentration of Faroe bridges exposed to sea-
water after 11 years, the bridges are located in Denmark The Faroe bridges used sulfate
388 H. L. Nguyen et al.

resisting Portland cement SRPC and fly ash FA with ratio 77% SRPC + 23% FA. Tang
[10] used 100% SRPC to cast the concrete sample which was exposed to seawater for
10 years in Sweden. Mix design of the concretes and chemical composition of Denmark
SRPC + FA and Sweden SRPC are extracted from the Henrik et al. [9] and Tang [10],
respectively. Further detail can be described in the works. Based on the mix design and
the chemical composition, mineralogical composition of hydrated cement is shown in
the table 1, thanks to the hydration model GEMS software [11]. The effective chloride
diffusion coefficient is also presented in this table.

Table 1. Mineralogical composition of hydrated cement and effective chloride diffusion coeffi-
cient

Mineral Volume fraction

Denmark SRPC+FA Sweden SRPC
Jennite 0.1411 0.1221
Tobermorite 0.0150 0.0130
Ettringite 0 0
Monosulfoalumi-
0.0272 0.0410
nate
Portlandite 0.0307 0.0602
Hydrotalcite 0.0290 0.0062
Straetlingite 0.0095 0
Effective chloride
diffusion coeffi- 1.54 × 10−13 1.60 × 10−12
cient (m2.s-1)

4 Results and discussions

Using the geochemical code Phreeqc software [12] of United State geological survey
USGS to take into account all the parameter mentioned above. The numerical and ex-
periment of total chloride concentration over depth are presented in Fig.1 and Fig 2.
The geochemical model simulated successfully two different types of cement in two
exposures duration: 10 years and 11 years. Especially, the model simulated correctly
trend of chloride profile in the first ten millimeter. In fact, using only the second law’s
Fick cannot correctly predict the chloride concentration in this zone [9], [10]. Moreo-
ver, the inaccuracy induces difficult interaction simulation of the first zone concrete
exposed to seawater. The results show that geochemical model can relatively forecast
total chloride profile over depth in two cases. Furthermore, pH and free chloride con-
centration (cf. Fig.3) which are two indicator of concrete durability [13], are also cal-
culated by the geochemical model. Via the results, the geochemical model proves its
simulation capability of concrete/solution interaction. Furthermore, mix design
70%SRPC+30%FA concrete prevents chloride diffusion better than 100% SRPC
Geochemical modelling for prediction of chloride … 389

concrete even if 70%SRPC+30%FA concrete exposed longer than 100% SRPC. That
shows partial replacement of fly ash FA improve resistance of concrete to chloride in-
gress as corrosion resistance of reinforced concrete.

1.00
(g/100g of concrete)

0.80
Numerical Experiment
Total chloride
concentration

0.60
0.40
0.20
0.00
0 10 20 30 40 50 60
Depth (mm)
Fig. 1. Total chloride concentration profiles of Faroe concrete bridge exposed to seawater after
11 years: comparison between numerical result and experiment result [9]

1.00
Numerical Experiment
(g/100g of concrete)

0.80
Total chloride
concentration

0.60
0.40
0.20
0.00
0 10 20 30 40 50 60
Depth (mm)
Fig. 2. Total chloride concentration profiles of Sweden SRPC concrete exposed to seawater after
10 years: comparison between numerical result and experiment result [10]

Free chloride concentration

14 0.15 (g/100g of concrete)
13
12 0.10
pH-SRPC+FA_11 years exposure
11
pH-Sweden SRPC_10 years exposure
pH

10 0.05
9
8 0.00
0 10 20 30 40 50
Depth (mm)
Fig. 3. Numerical pH, free chloride concentration of Sweden SRPC concrete and SRPC+FA ex-
posed to seawater after 10 years, 11 years respectively.

5 Conclusions

In this paper, the simulation capability of geochemical model is briefly proved. The
model can predict correctly chloride profile of concrete exposed to seawater for long
time. Moreover, the different indicators of concrete durability such as pH, free chloride
concentration can be forecasted by the geochemical model. It seems that concrete
70%SRPC+30%FA has better resistance of concrete to chloride ingress than
100%SRPC. However, the results of geochemical model are only verified by total
390 H. L. Nguyen et al.

chloride profile in a moment. Therefore, to improve the feasibility of geochemical

model for different case, further work needs to be done such as: comparison between
numerical and experimental pH, free chloride profile, Ca2+ quantity, SO42- quantity,
Mg2+ quantity,etc …in different time points.

Acknowledgments

This research is funded by Vietnam National Foundation for Science and Technol-
ogy Development (NAFOSTED) under grant number 107.99-2018.337.

References
1. E. Samson and J. Marchand, “Modeling the transport of ions in unsaturated cement-based
materials,” Comput. Struct., vol. 85, no. 23–24, pp. 1740–1756, (2007).
2. V. Baroghel-Bouny, M. Thiéry, and X. Wang, “Modelling of isothermal coupled moisture–
ion transport in cementitious materials,” Cem. Concr. Res., vol. 41, no. 8, pp. 828–841,
(2011).
3. E. Samson and J. Marchand, “Modeling the effect of temperature on ionic transport in ce-
mentitious materials,” Cem. Concr. Res., vol. 37, no. 3, pp. 455–468, (2007).
4. T. Xu, N. Spycher, and E. Sonnenthal, “TOUGHREACT User’s Guide: A Simulation Pro-
gram for Non-isothermal Multiphase Reactive Transport in Variably Saturated Geologic
Media, version 2.0,” Lawrence Berkeley …, no. October, (2012).
5. D. L. Parkhurst and C. A. J. Appelo, “Description of input and examples for PHREEQC
Version 3 — A computer program for speciation, batch-reaction, one-dimensional transport,
and inverse geochemical calculations,” U.S. Geol. Surv. Tech. Methods, B. 6, chapter A43,
(2013).
6. P. Blanc et al., “Thermoddem: A geochemical database focused on low temperature wa-
ter/rock interactions and waste materials,” Appl. Geochemistry, (2012).
7. A. C. Lasaga, J. M. Soler, J. Ganor, T. E. Burch, and K. L. Nagy, “Chemical weathering rate
laws and global geochemical cycles,” Geochim. Cosmochim. Acta, vol. 58, no. 10, pp.
2361–2386, (1994).
8. I. Baur, P. Keller, D. Mavrocordatos, B. Wehrli, and C. A. Johnson, “Dissolution-precipita-
tion behaviour of ettringite, monosulfate, and calcium silicate hydrate,” Cem. Concr. Res.,
vol. 34, no. 2, pp. 341–348, (2004).
9. B. Henrik and E. Sørensen, “Chloride ingress in old Danish bridges,” (2014).
10. T. Luping, “Cement and Concrete Research Engineering expression of the ClinConc model
for prediction of free and total chloride ingress in submerged marine concrete,” vol. 38, pp.
1092–1097, (2008).
11. D. A. Kulik et al., “GEM-Selektor geochemical modeling package: revised algorithm and
GEMS3K numerical kernel for coupled simulation codes,” Comput. Geosci., vol. 17, no. 1,
pp. 1–24, (2013).
12. D. L. Parkhurst and C. A. J. Appelo, “User’s Guide To PHREEQC (version 2) — a Com-
puter Program for Speciation, and Inverse Geochemical Calculations,” U.S. Geol. Surv. Wa-
ter-Resources Investig. Rep., (1999).
13. U. Angst, B. Elsener, C. K. Larsen, and Ø. Vennesland, “Critical chloride content in rein-
forced concrete - A review,” Cem. Concr. Res., vol. 39, no. 12, pp. 1122–1138, (2009).
 Application of Asphalt Concrete using Limestone
with Cement & Admixture

Nguyen Ba Hoang1, Pham Van Hung2, Tran Viet Khanh 1, Nguyen Viet Huy2

1
Ho Chi Minh City University of Transport
2
Institute of Transport Science and Technology
hoang_ba2001@yahoo.com

Abstract. Vietnam is one of the countries in the world that will likely suffer from
climate change predictions. The natural state of severe seasonal weather pat-
terns in Vietnam, when combined with heavy trucks, makes the current conven-
tional types of asphalt concrete are no longer able to bear the load. This results in
common defects, such as rutting resistance and fatigue cracking of the current
asphalt concrete roads surfaces. This article aims to introduce a type of high
strength asphalt concrete using limestone filler, combined with cement and an
adhesion additive that will help to solve the mentioned problem. The place of the
pilot project was on a ten kilometers road (Km 1613+00 to Km 1623+00) on the
National Highway number 1, which suffers from the heaviest traffic loads. The
asphalt concrete was designed with a Marshall- Bailey Modification Method. After
four years of use, the road was recently tested and remains in an operational and
stable condition. There are many solutions proposed to apply the solution of
high-intensity asphalt concrete using limestone filler combined with cement and
admix. Wetfix®BE increased adhesive is the best solution due to the product’s
cost-effectiveness, the availability of materials: like sand and stone. The asphalt
concrete is high in strength, heat resistant, rutting resistance and fatigue crack-
ing.

Keywords: Asphalt concrete, limestone, admix

1 Introduction
Over the past few years, Vietnam has implementing and completing many road
bridge projects, such as the Phap Van - Cau Gie expressway and the Hanoi - Hai
Phong expressway. However, after a period of exploitation, on some sections of na-
tional highways with noticeable high traffic loadings, wheel rutting and fatigue
cracked on the asphalt surface has occurred.
Vietnam Institute of Transport Science and Technology inspected on a number of
specific projects, including the National HighwayNo.1 (NH1) and National Highway
No.5 (NH5) projects, finding:
- Sink track appearing on a project using Asphalt Concrete (AC) that was
constructed according to Vietnamese code 22TCN 249: 1998 and TCVN 8819: 2011.
© Springer Nature Singapore Pte Ltd. 2020 391
C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_60
392 N. B. Hoang et al.

- Sink track appearing on a project with the top layer using AC10 (22TCN249-98),
AC15 (22TCN249-98), and AC19 (according to TCVN8819: 2011)
- Wheel rutting usually occurs on lane trucks, with larger levels typically
concentrated on slow-moving vehicles near intersections, industrial parks, mining
areas and ports. In fact, many places that have high flows, heavy loads, large vertical
slopes or crossroads, etc. or a combination of the aforementioned, AC is not sufficient
to withstand the load, leading to early damage.
Many studies and proposals are mentioned to reduce and overcome the
phenomenon of wheel rutting including: reviews the proposal to change, add and
develop technical standards / regulations for AC design work including road structure
and construction; strengthening of quality management at all stages: surveys, designs,
construction, acceptance. Research has also been undertaken on the use of new
materials. In order to improve the quality of the road surface, many researchers have
been carrying out research (both in materials and technology) to improve the quality
of the concrete, such as the use of high-performance polymers, good quality
aggregate, appropriate aggregate material; use of more additives to enhance the
quality, stability; Asphalt Cement has been studied in many countries, but the result is
not satisfactory due to the limitation of brittleness and cracking ([1], [2]). The Asphalt
surface is easy to crack due to fatigue. If the AC is not cracked due to fatigue, the
intensity of AC and the resistance to wheel rutting is low. Asphalt Cement, presented
in this paper, differs from other research projects. We haven’t simply put only cement
into the asphalt, but we have attempted to put cement together with limestone powder,
combined with a sticky additive between stone and bitumen.
The purpose is to produce anti-crispy for AC and creating a type of AC with
Marshall high intensity from 12 ÷ 20kN. Depending on the type the Marshall
flexibility is still between 3 ÷ 4mm. It met fully the two mechanical properties of AC,
being good against wheel-rutting and fatigue resistance, ensuring the road surface
does not crack.
2. Experiment
2.1. Design method of AC
Nowadays, the most common AC design method in Vietnam is based on the
Marshall Design (MD) method and a more modern AC design method, the
Super_pave method. MD method is too long and many inadequacies have been
exposed due to social reality changing over 40-50 years of its application. The main
content of the MD method is only to find the optimum bitumen content, together with
the materials and aggregates that make up the AC, have been used to its maximum
extent to create a new AC material that can meet the increasing demands of social
reality, the MD method has not yet been scientifically grounded.
Meanwhile, the Super pave design method is a modern design method, but it must
train human resources in order to understand it and investing in the refitting of all
laboratories across Vietnam is intensive. Besides that, there is a need to undertake
surveys and background monitoring of the different regions in Vietnam before we
could put this design method into practice. In summary, the application of the Super
pave design method is expensive and it would take a long time from investment
before it could be put into practice.
Application of Asphalt Concrete using Limestone … 393

Thus, the AC design method we used was the Marshall Bailey Modified method
that is a combination of two basic methods, the MD and Bailey methods making it
more realistic.
Marshall Bailey Modified design method consists of 3 stages as follows:
1. Orientation Design
2. Optimum Design according to Bailey Method
3. Design with normal Marshall Method

Oriented pilot design stage: To select materials suitable for the available areas, the
material handling capability can meet the requirements of producing high-strength
asphalt concrete, or control the price of asphalt concrete, oriented mechanical norms
of asphalt concrete to serve the design of high strength asphalt concrete.
The optimal design stage of asphalt concrete after Bailey method: used for the
optimal design of roads according to load bearing capacity, the resistance to
wheelbase, the resistance to asphalt concrete, according to oriented sections that have
been selected to meet the required design force calculation and also to accord with the
actual exploitation of the works.
Marshall General Design Phase: Four steps in accordance with Vietnamese code
TCVN 8820: 2011 to determine the optimum bitumen content, the actual AC recipe at
the station, the technology of concrete surface treatment and the daily quality control
of the production process of asphalt concrete pavement. The content of the Marshall-
Bailey Modification design methodology will be discussed in detail in a separate
article.
We put the mix of adhesive poor adhesion, limestone minerals of normal active
nature, semi hard solid 60/70 cement and additive sticking to stone with asphalt to
create high intensity, heat and load resistance. It will save large vehicles with good
fatigue resistance and anti-wheel-rutting. So far, AC on the road has been used for
more than three years, but it's still in very good condition in KM1613 ÷ KM1623,
Binh Thuan province. In this article, we present the results of the AC 19 using local
stone materials of low quality, the adhesion of the stone of grade 2 with the standard
bitumen.

2.2 Some experimental results

Based on the criteria of using available local materials to ensure the quality and cost
of AC, the most effective price of conventional AC are produced in the market.
AC mixing station is also used by AC mixing plant which is popularly used in the
market.
Some results of materials testing as Table 1.

Table 1: Some experimental results

No. Item Index
I Stone (10 ÷ 20mm)
1 Degradation degree in the Los Angeles experiment (%) 21.38
2 Soft agregate percentage(%) 1.63
3 Percentage of thin stone (%) 7.58
4 Percentage of dust (%) 0.96
394 N. B. Hoang et al.

No. Item Index

5 Percentage of Clay(%) 0.00
6 Eigen Mass(g/cm3) 2.649
II Small Stone(5 ÷ 10mm)
1 Degradation degree in the Los Angeles experiment (%) 21.21
2 Soft agregate percentage(%) 1.88
3 Percentage of thin stone (%) 7.42
4 Percentage of dust (%) 1.06
5 Percentage of Clay(%) 0.00
6 Eigen Mass(g/cm3) 2.644
III Supper Small Stone(0-:-5mm)
1 Degradation degree in the Los Angeles experiment (%) 20.82
2 Soft agregate percentage(%) 1.53
3 Percentage of thin stone (%) 7.60
4 Percentage of dust (%) 1.34
5 Percentage of Clay(%) 0.00
6 Eigen Mass(g/cm3) 2.644
IV Sand
1 Grade modul 2.63
2 Sand index 97.63
3 Percentage of Dust(%) 0.9
4 Percentage of Clay (%) 0
5 Angle degree of sand 44.4
V Filler
1 Eigen Mass (g/cm3) 2.817
2 Iinfiltration(%) 0.827
3 Humidity (%) 0.617
4 Eigen mass of Filler and bitumen mix(g/cm3) 2.299
5 Bitumen in Filler 59.57
VI Bitumen
1 Penetration index 25oC(0.1mm) 67.2
2 Elongation 25oC(cm) 145.5
3 Pyrometric cone equivalent (oC) 47.8
4 Temperature of firer(oC) 339
5 Penetration index 163oC / Penetration index 25oC (%) 89.62
6 Lost percentage at 163oC within 5h (%) 0.18
7 Eigen mass 25oC 1.036
8 Agglutinate with stone Grade 4
VII Agglutinate of local stone with bitumen
1 Bitumen 60/70 Petrolimex Grade 2
2 Bitumen 60/70 Petrolimex treatment 0.3% Grade 2÷3
3 Bitumen 60/70 Petrolimex treatment 0.5% Grade 3÷4
4 Bitumen 40/50 Petrolimex Grade 2
5 Bitumen 60/70 COLAS Grade 2
6 Bitumen 60/70 COLAS treatment Grade 4
Application of Asphalt Concrete using Limestone … 395

2.3. Several results of AC in the optimum Bailey design stage

The results of AC in the optimum Bailey design stage is shown in Table 2. In 5
samples, Sample BTNC 19-4 (0.3% Wetfix B and high quality mineral powder + 2%
cement) and Sample BTNC 19-5 (0.3% Wetfix B and high quality mineral powder +
3% cement) are to consideration. Other items are equal but remain Marshall stability
percentage of BTNC 19-4 and BTNC 19-5 are higher. The Marshall stability index of
BTNC 19-5 is higher than BTNC 19-4, so the sample BTNC 19-5 is better.

Table 2 : Several results of AC in the optimum Bailey design stage

Item (1) (2) (3) (4) (5)
Mass per 1m3 2.304 2.290 2.291 2.345 2.343
Marshall stability index 19.03 18.02 17.81 13.51 16.67
Marshall plastic index 3.62 2.85 3.00 3.23 3.67
Remain Marshall stability percentage
79 81 82 91 90
(%)
Voids percentage of aggregate , Va 16.83 17.32 17.29 15.35 15.44
Voids in the mineral aggregate, VMA 5.24 5.77 5.73 5.15 3.62
Grad
Level of sticking material with stones Grad 2 Grad 2 Grad 3 Grad3
2
Voids filled with asphalt, VFA 68.44 65.19 66.93 68.02 76.33
Bitumen percentage (%) 5.21 5.21 5.21 5.21 5.21
Remarks: (1) Sample BTNC 19-1: Aggregate with much natural sand, Petrolimex 60/70 bitumen not yet
sticky adhesion treatment.
(2) Sample BTNC 19-2: Aggregate with grinding sand, bitumen COLAS 60/70 has sticky adhesion
treatment
(3) Sample BTNC 19-3: Standard Aggregate with natural sand, bitumen Petrolimex 40/50
(4) Sample BTNC 19-4: Aggregate with much natural sand, Petrolimex 60/70 asphalt has a sticky
adhesion of 0.3% Wetfix B and high quality mineral powder + 2% cement
(5) Sample BTNC 19-5: Aggregate with much natural sand, Petrolimex 60/70 asphalt has a sticky
adhesion of 0.3% Wetfix B and high quality mineral powder + 3% cement

Fig 1. Result of experiment wheel rutting

396 N. B. Hoang et al.

The results of the test of rutting on the wheel tracking in Figure 1 show that: The
sample was tested at a temperature of 50OC. As we see in the about graph the deep of
rutting increased with the number of loading cycles, collecting data at both the left
and right sides of the wheel, at the time of 15,000 loading cycles, the result is an aver-
age of rutting of 4.1mm. According to TCVN 8819: 2011, the allowable wheel rutting
is ≤12.5mm, according to ASTM D6433, the danger level of rutting is low at 6-
13mm. Thus, the results of the test of rutting for the sample are very positive. The
deep of rutting according to the width of wheel is shown in graph below for both the
left and right sides of the wheel (100mm each in the Horizontal profile).
3. Conclusion:
High strength AC materials have been developed using cemented limestone pow-
der and additive for stone with resin of Marshall Strength greater than 13.5 kN, Mar-
shall Plasticity from 3 ÷ 4mm, and the thickness of the wheel is 3 ÷ 7mm.
AC mixing stations are normally used in general, no need to extend and upgrade.
Types of natural stone and sand are used with locally available materials that are
sticky, poorly adhesive to heavy roads, heavy vehicle traffic, and hot weather such as
in National Highway 1 through the province Binh Thuan (Km1613 ÷ Km1623). Min-
eral powders use limestone powder, PCB 30 cement and 60/70 resins. The Sticky
additive uses for stone with bitumen using wet fix type B.
The test results during the exploitation from the begining to the present time more
than 4 years showed that the road surface is very good, no signs of cracking due to
fatigue or wheel rutting, loss of flatness. All traffic in the exploitation process,
comparing with adjacent sections used by conventional types of AC has had to repair
and maintain a lot.
The production cost of this kind of high-yield natural rubber isn’t significantly
higher than that of the conventional method, due to the bitumen savings, the cement
price difference and low additive cost. This makes it very economical and is
maintenance-free with no incidents in hot weather under heavy loading and high
traffic.
References:
1. D.A. Anderson. et al, “Physical properties of Asphalt Cement and the development of per-
formance-related specifications,” Association of Asphalt Paving Technologists, vol. 60, pp.
437-475, Jun. 1998.
2. YANG. Jinbo. et al, “Study on the hardening mechanism of cement asphalt binder,” Science
China Technological Sciences, vol. 53, No. 5, pp. 14061412, May. 2010.
 Experimental Investigation of a Self-powered
Magnetorheological Damper for Seismic Mitigation

Daniel C1[ 0000-0002-4024-4742 ], Hemalatha G1, Sarala L1, Tensing D1,

Sundar Manoharan S2, Xian-Xu Bai3,

1
Karunya Institute of Technology and Sciences, Coimbatore
2 Sathyabama Institute of Science and Technology, Chennai
3
Hafei University of Technology, China
*danielckarunya@gmail.com

Abstract. The present work investigates the effectiveness of Magnetorheological

Damper (MR) damper coupled with the smart self-powered system, MR damper
acts as electromagnetic induction (EMI) device in controlling seismic vibration.
The proposed smart damping system with an EMI device is capable of converting
vibration energy into electrical energy. Thus, the EMI device attached with MR
damper can be used as an effective and alternative power source for the MR
damper, making it a self-powering system. The primary aim of the experimental
study is to identify the performance of the proposed smart damping system using
time history loading (El Centro earthquake). For experimentation, the MR
damper with EMI was designed and fabricated. To reduce sedimentation, nano
Fe3O4 was used in the preparation of MR fluid. The performances of the proposed
smart damping system are compared with the passive, semi-active and active con-
trol system in force and displacement to evaluate the effectiveness of the self-
powered smart damping system in reducing seismic vibration. The experimental
results show that the self-powered smart damping system produces more damp-
ing force and reduction in displacement. The maximum damping force obtained
is 0.67 kN. In an active system, a force was increased by 12.9% and displacement
was reduced by 13.4% when compared with the semi-active control system. The
results revealed that the proposed EMI can act as a sole power source for the
damping system.

Keywords: Self-Power, Nano Fe3O4, Magnetec oil, MR Damper

1 Introduction

In recent days there has been a momentous enhancement in the application of mag-
netorheological (MR) fluid mainly because of its rheological properties under exerted
magnetic fields [1]. Its characteristics has paved the way to the construction of various
MRF-based devices. Some of the devices are MR damper, MR valve, MR brake, and
MR clutch. Potential application such as long range controllable damping force, quick
reaction time, and low power they are widely being used in MRF-based devices [2].
MR dampers are being used in automotive industry for the ,manufacturing of off-road
vehicles [3, 4]. In addition, they are also being employed in naval gun controlling [5],

© Springer Nature Singapore Pte Ltd. 2020 397

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_61
398 C. Daniel et al.

field of landing gear [6], prosthetic knees [7], washing machines [8], high speed train
suspension [9, 10], particularly In building, for seismic vibration control [11, 12], and
so forth. The kinetic energy that is formed by the vibration is being ruined by heating
and not being harvested as a power that can be reused. This can be because of the fact
that there is no energy harvesting device. This is a major setback. Recently, energy
harvesting MR dampers have received much attention owing to its capability of recov-
ering kinetic energy that normally transformed by traditional MR dampers currently,
many researchers focussed on the principles and designs of self-powered MR dampers,
which can be classified into two categories. The first category is to convert the linear
damper vibration into oscillatory rotation and kinetic energy was harvested with the
help of rotational permanent magnetic DC or AC generators to harvest kinetic energy.
These mechanical mechanisms include rack and pinion, ball screw, and hydraulic trans-
mission. [13-14] proposed an electrorheological (ER) shock absorber integrated with a
generator by employing a rack and pinion gear mechanism, which converted a linear
motion of the piston to a rotary motion, thus activating a generator to produce electrical
energy to self-power the excitation coil in the piston head [15-19]. The guild layer and
shield layer were adopted to minimize magnetic interaction effect between permanent
generator and MR damper. However, the extra shield and guild layer increased the
complex of the whole structure while the component itself provided litter contribution
to the dynamic control or improvement of generating effect. In this paper, a new MR
damper with energy harvesting ability was proposed based on the EMI principle.MR
fluid was synthesised using nano Fe3O4 as smart material and magnetec oil as carrier
fluid. The nano iron particle was used to reduce sedimentation. Passive, semi-active
and active control system for time history loading is studied for MR Damper. The prop-
erties of proposed MR damper were experimentally investigated, and power regenera-
tive characteristics were also discussed.

2 MR fluid
A magnetorheological fluid (MR fluid) belongs to the smart fluid category, concist
of a smart material usually a magnetic particle and a carrier fluid. In the presence of
magnetic field, the MR fluid increases in apparent viscosity and becomes a viscoelastic
solid the chain-like structure arises when the fluid is subjected to the external magnetic
field, Furthermore, the yield stress of the MR fluid is altered and controlled precisely
by modifying the intensity of magnetic field [20]. In the present work, MR fluid was
prepared by mixing 60% weight percentage of nano Fe 3O4 of average particle size 12
nm calculated by Scherrer equation, with 40% magnetec oil. The nano Fe 3O4 & mag-
netec oil was mixed and stirred in the overhead stirrer for one complete day as depicts
in fig. 1. Thus, the MR fluid obtained was used for further work.
Experimental Investigation of a Self-powered … 399

Fig. 1 Preparation of MR fluid

3 MR Damper
FEMM software was used to model the shear mode MR Damper. The configuration
of the MR damper was designed and fabricated using mild steel material. In the piston
head, 260 turns coil was wounded with a standard wire gauge (SWG) of 21. The annular
gap is maintained as 1mm. The EMI system is placed above the cylinder cap. Ferrite
magnet grade N42 is used for the study to create power and coil with has SWG 23,
thickness of 0.610 mm with 1277 number of turns, which has an air gap of 2 mm was
used in EMI system as shown in fig. 2.

Fig. 2 schematic representation of self-power MR Damper

4 Experimental method
Experimental investigation was studied in MTS Universal testing machine (UTM),
strength of material lab, Karunya Institute of Technology and Sciences. The MR
damper is fixed with the proper fixtures. The experimental setup for the study depicts
in fig. 4.a). The experiment is carried out without supplying current to MR damper for
passive system. The current of 2.5A is given from DC power supply to the coil con-
nected with the piston head for semi-active control. However, EMI generates 29.6 mV
of power was transformed to the piston through proper connection where active control
400 C. Daniel et al.

system studied. Fig. 4 b) shows current vs time for El Centro earthquake data. The
1940 El Centro earthquake occurred on May 18 at Southern California ground acceler-
ation data is the time history data for all experimental study.

Fig. 4 a) Experimental setup of MR Damper in MTS UTM b) Current vs. time for El Centro
ground acceleration data

5 Result and Discussion

The time history data used for the study which contains 31.18 seconds time for pas-
sive, semi-active and active study. With self-power MR Damper 12.9% force increased
when compared to semi-active MR damping system which has a maximum force of
0.67 kN as depicts in fig.5 a). With self-power MR damper 13.4% displacement was
reduced when compared to semi-active MR damping system as shown in fig. 5 b). Table
1 shows the maximum force, maximum displacement values for passive, semi-active
and active control system.

Fig. 5 a) Force vs. time b) Displacement vs. time

Experimental Investigation of a Self-powered … 401

Table 1 Maximum force and displacement of control system

Control system Force, kN Displacement, mm

Passive 0.35 4.47

Semi - active 0.59 2.97
Active 0.67 2.57

6 Conclusion
To conclude, a MR damper with self-powered device was designed, fabricated, and
tested. The cylinder cap of MR damper shares the EMI part. As a result, the magnetic
field interference was reduced. Experimental tests were carried out to address the per-
formances of the proposed MR damper. With self-power MR Damper 12.9% force in-
creased when compared to semi-active MR damping system which has a maximum
force of 0.67 kN. With self-power MR damper 13.4% displacement reduced when com-
pared to semi-active MR damping system. The experimental result revealed the self-
powered smart damping unit performs more than the passive control cases in reducing
the seismic vibration.

Acknowledgment
The authors gratefully thank Karunya Institute of Technology and Sciences, Coim-
batore, Tamilnadu, India for their constant support. We also extend our acknowledge-
ment to the Department of Science and Technology for the financial support to carry
out the research work (Grant No: DST/TSG/STS/2015/30-G).

Reference
1. Li, W. H., Du, H., Chen, G., Yeo, S. H., & Guo, N. (2003). Nonlinear viscoelastic properties
of MR fluids under large-amplitude-oscillatory-shear. Rheologica Acta, 42(3), 280-286.4
2. Zhu, X., Jing, X., & Cheng, L. (2012). Magnetorheological fluid dampers: A review on
structure design and analysis. Journal of Intelligent Material Systems and Structures, 23(8),
839-873.
3. El Majdoub, K., Ghani, D., Giri, F., & Chaoui, F. Z. (2015). Adaptive semi-active suspen-
sion of quarter-vehicle with magnetorheological damper. Journal of Dynamic Systems,
Measurement, and Control, 137(2), 021010.
4. Shin, Y. J., You, W. H., Hur, H. M., Park, J. H., & Lee, G. S. (2014). Improvement of ride
quality of railway vehicle by semiactive secondary suspension system on roller rig using
magnetorheological damper. Advances in Mechanical Engineering, 6, 298382.
5. Singh, H. J., & Wereley, N. M. (2014). Optimal control of gun recoil in direct fire using
magnetorheological absorbers. Smart materials and Structures, 23(5), 055009.
402 C. Daniel et al.

6. Powell, L. A., Hu, W., & Wereley, N. M. (2013). Magnetorheological fluid composites syn-
thesized for helicopter landing gear applications. Journal of intelligent material systems and
structures, 24(9), 1043-1048.
7. Gudmundsson, K. H., Jonsdottir, F., Thorsteinsson, F., & Gutfleisch, O. (2011). An experi-
mental investigation of unimodal and bimodal magnetorheological fluids with an applica-
tion in prosthetic devices. Journal of Intelligent Material Systems and Structures, 22(6),
539-549.
8. Nguyen, Q. H., Choi, S. B., & Woo, J. K. (2014). Optimal design of magnetorheological
fluid-based dampers for front-loaded washing machines. Proceedings of the Institution of
Mechanical Engineers, Part C: Journal of Mechanical Engineering Science, 228(2), 294-
306.
9. Li, Z., Ni, Y. Q., Dai, H., & Ye, S. (2013). Viscoelastic plastic continuous physical model
of a magnetorheological damper applied in the high speed train. Science China Technolog-
ical Sciences, 56(10), 2433-2446.
10. Sun, S., Deng, H., Li, W., Du, H., Ni, Y. Q., Zhang, J., & Yang, J. (2013). Improving the
critical speeds of high-speed trains using magnetorheological technology. Smart Materials
and Structures, 22(11), 115012.
11. Yang, G., Spencer Jr, B. F., Jung, H. J., & Carlson, J. D. (2004). Dynamic modeling of large-
scale magnetorheological damper systems for civil engineering applications. Journal of En-
gineering Mechanics, 130(9), 1107-1114.
12. Amini, F., & Ghaderi, P. (2012). Optimal locations for MR dampers in civil structures using
improved ant colony algorithm. Optimal control applications and methods, 33(2), 232-248.
13. Solomon, S. M., Dharmaraj, T., Loganathan, S., Immanuel, S., Gladston, H., & Cruze, D.
(2018). Experimental Investigation on Magnetorheological Damper for RCC Frames Sub-
jected to Cyclic Loading. Advances in Civil Engineering Materials, 7(3).
14. Choi, S. B., Seong, M. S., & Kim, K. S. (2009). Vibration control of an electrorheological
fluid-based suspension system with an energy regenerative mechanism. Proceedings of the
Institution of Mechanical Engineers, Part D: Journal of Automobile Engineering, 223(4),
459-469.
15. Li, Z., Zuo, L., Luhrs, G., Lin, L., & Qin, Y. X. (2013). Electromagnetic energy-harvesting
shock absorbers: design, modeling, and road tests. IEEE Transactions on Vehicular Tech-
nology, 62(3), 1065-1074.
16. Li, Z., Zuo, L., Kuang, J., & Luhrs, G. (2012). Energy-harvesting shock absorber with a
mechanical motion rectifier. Smart Materials and Structures, 22(2), 025008.
17. Yu, M., Peng, Y., Wang, S., Fu, J., & Choi, S. B. (2014). A new energy-harvesting device
system for wireless sensors, adaptable to on-site monitoring of MR damper motion. Smart
Materials and Structures, 23(7), 077002.
18. Zhang, Y., Huang, K., Yu, F., Gu, Y., & Li, D. (2007, December). Experimental verification
of energy-regenerative feasibility for an automotive electrical suspension system. In Vehic-
ular Electronics and Safety, 2007. ICVES. IEEE International Conference on (pp. 1-5).
IEEE.
19. Xinchun, G., Yonghu, H., Yi, R., Hui, L., & Jinping, O. (2015). A novel self-powered MR
damper: theoretical and experimental analysis. Smart Materials and Structures, 24(10),
105033.
20. Daniel, C., Hemalatha, G., Magdalene, A., Tensing, D., & Manoharan, S. S. (2017, May).
Magnetorheological Damper for Performance Enhancement Against Seismic Forces. In In-
ternational Congress and Exhibition" Sustainable Civil Infrastructures: Innovative Infra-
structure Geotechnology" (pp. 104-117). Springer, Cham.
 Effect of Ground Blast Furnace Slag in Replacement of
Cement in Ternary Binder on Performance of Sand Concrete

Thanh Sang Nguyen1 and Long Hai Chu2

1
Faculty of Civil Engineering, University of Transport and Communications, Vietnam
nguyenthanhsang@utc.edu.vn
2
Faculty of International Education, University of Transport and Communications, Vietnam
longhaichuutc@gmail.com

Abstract. With plentiful material resources including dune sand from Central
Coastal provinces and fly ash – a by-product of burning coal at electronic power
plants, sand concrete is widely used in Vietnam in respond to both technical and
economical requirements. In this research, with the proportion of 18% cement,
37% crushed sand, 33% dune sand, and 12% fly ash, the length change in alka-
li-silica reaction and strength development tests were conducted in order to
evaluate the effect of ground (granulated) blast furnace slag (GBFS) in re-
placement of cement in 0%, 20%, 30%, 40%, and 50%, respectively on perfor-
mance of sand concrete. Besides, the relevant microstructural changes due to
GBFS proportion and 02 different curing conditions were observed to support
the assessment on potential alkali-silica reactions by conducting scanning elec-
tron microscopy-energy dispersive X-ray spectroscopy (SEM-EDX) and X-ray
diffraction (XRD). As a result of these tests, it is confirmed that 30% GBFS in
replacement of cement is suitable content in this sand concrete.

Keywords: GBFS (Ground Blast Furnace Slag), Potential Alkali Silica Reac-
tion (ASR), XRD, SEM/EDX, Strength, Sand Concrete.

1 Introduction

With the position of a developing country, over the past years, Vietnam has paid
great attention to the investment in the development of infrastructure and urban sys-
tems. Therefore, the demand for construction materials, especially cement concrete, is
huge. However, large aggregates (crushed stone, gravel) - the main component in
traditional cement concrete - are gradually depleted. Moreover, river sand is gradually
becoming unstable at high prices because of the tight policies for extraction manage-
ment due to negative impacts on river ecosystems, navigation and flood control.
Meanwhile, with the benefit of more than 3200km coastline, sand dunes in the coastal
area in Vietnam have large reserves and easy to exploit. In addition, industrial by-
products such as fly ash, ground blast furnace slag (GBFS), etc. in huge quantities
should be used reasonably instead of making costly landfill. Thus, sand concrete is a
suitable solution for Vietnam's construction.

© Springer Nature Singapore Pte Ltd. 2020 403

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_62
404 T. S. Nguyen and L. H. Chu

The two main differences between sea-sand and traditional raw materials are on the
presence and the type of impurities. Meanwhile, the combination of cement, fly ash,
and GBFS as a ternary binder causes the complex chemical reactions and lead to
presence of uncontrolled substances. Thus, it is extremely necessary to evaluate the
quality of concrete before applying widely in practice, especially with this sand con-
crete having potential threats to the life of concrete as well as construction works in a
tropical monsoon climate country like Vietnam. The alkali-silica reaction (ASR) is
one of the most important criteria to assessment the quality and durability of concrete.
It is an expansive reaction between certain forms of silica in aggregates and potassium
and sodium alkalis in cement paste. Indications of the presence of alkali-aggregate
reactivity may be a network of cracks, closed or spalling joints, or movement of por-
tions of a structure. Alkali-silica reaction can be controlled through proper aggregate
selection and/or the use of supplementary cementitious materials (such as fly ash or
slag) or blended cements proven by testing to control the reaction [2].
In Vietnam condition, there is limited research about the effect of combined bind-
ers on quality of concrete, especially in potential ASR. Therefore, this research focus-
es on evaluate the effect of GBFS content in replacement of cement in ternary binder
(including cement, fly ash, and GBFS) on potential ASR of sand concrete by length
change methods and strength development as well as explanations in microstructure
by using XRD analysis and SEM/EDX observation.

2 Materials and Methods

2.1 Materials

Table 1. Main physical properties and chemical composition of PC 40, Fly ash, and GBFS

SCM
PC 40 Fly Ash (FA) GBFS
Parameters
SiO2 (%) 21.49 53.88 36.64
Al2O3 (%) 5.40 21.82 12.91
CaO (%) 63.56 4.27 34.74
MgO (%) 1.40 1.45 7.78
SO3 (%) 1.65 0.20 1.45
Na2O (%) 0.15 0.67 0.52
LOI (%) 1.20 6.27 -
Mean particle size (μm) 14.6791 29.42781 19.00592

Table 2. Properties of dune sand, and crushed sand

Properties Dune Sand Crushed Sand

Fineness modulus 1.80 2.80
Absorption (%) 1.24 2.46
Bulk dry specific gravity (g/cm3) 2.49 2.54
Effect of Ground Blast Furnace Slag in Replacement … 405

Fig. 1. Grading curve of SCMs (cement PC 40, fly ash, and GBFS)

Portland cement, GBFS (ground granulated blast furnace slag), fly ash, Phu Ly
crushed sand, and Quang Binh dune sand were used in this study.

2.2 Mixture proportions

In this research, sand concrete mixtures were designed based on the absolute vol-
ume of the constituent materials in which the paste volume was computed from the
void content in the sand mixture. The designed compressive strength of this sand
concrete was fixed at 60 MPa.

3 Results and Discussion

3.1 Length changes

In Fig. 2, the lengths of all specimens tend to increase over time. These changes in
length, although reached the maximum value about 0.08% at 35 days with specimens
soaked in 1M NaOH solution at 800C and – 0.027% at 35 days with specimens cured
at 380C with high humidity, are lower than the permissible limit of 0.2% at 16 days
according to ASTM C1260 and 0.05% at 90 days according to ASTM C227.

3.2 Compressive strength

(Fig.3) Specimens with 30% GBFS in replacement of cement were less influenced
by curing conditions, and also have maximum compressive strength in 28 days.
406 T. S. Nguyen and L. H. Chu

Fig. 2. Length change results of specimens in 02 different curing environments according to

different GBFS content

Fig. 3. Compressive strength results of specimens in 02 environments

Effect of Ground Blast Furnace Slag in Replacement … 407

3.3 XRD analysis

The XRD results of sand concrete specimens according to different GBFS content in
replacement of cement and different curing conditions are shown in Fig. 4.

Fig. 4. XRD results according to different GBFS content in 1M NaOH solution at 800C

3.4 SEM/EDX analysis

The SEM/EDX results of sand concrete specimens according to different GBFS con-
tent in replacement of cement and different curing conditions are shown in Fig. 5.

3.5 Discussion

The positive effect of GBFS on compressive strength will be due to the high poz-
zolanity of GBFS resulting from the high silica content. GBFS and fly ash reacts in-
tensively with the water and the calcium hydroxide generated from the hydration of
cement to produce additional C–S–H. The additional C–S–H itself is the main
strength-contributing compound, and also fills in the capillary pores to improve the
microstructure of the paste matrix and transition zone in concrete resulting in en-
hancement of compressive strength. Therefore, the specimens with 30% GBFS in
replacement of cement have almost good results in other tests as analysed above. In
Fig.4, the presence of portlandite mostly shows that if having a huge of alkali concen-
tration, the length change is small because of the combination of portlandite with
alkali for making a non-swelling gel C-N(K)-S-H [2].
408 T. S. Nguyen and L. H. Chu

Fig. 5. SEM/EDX results according to 0 and 30% GBFS content in 1M NaOH solution

This is a suitable explanation for the higher length change in specimens cured at
800C in 1M NaOH solution. Erlin [1] noted that the presence of ettringite or other
phases in bands at aggregate boundaries shows only that the paste has expanded.

4 Conclusions
Some important conclusions have been given as below:
 30% GBFS in replacement of cement is suitable content to prevent the po-
tential alkali-silica reaction in this sand concrete;
 Alkali-silica reactions were not detected by method of Length Change of
mortars. However, the minerals/crystals influenced in length change and
strength development have found by XRD and SEM/EDX such as: ettring-
ite, calcium silicate hydrate C-S-H, alkali gel Na2SiO3.2H2O, were found.

References
1. B.Erlin: Ettringite – whatever you may think it is. International Cement Microscopy Asso-
ciation, Duncansville, USA, p.380-381 (1996).
2. Geoffrey E. Blight, Mark G Alexander: Alkali-Aggregate Reaction and Structural Damage
to Concrete. University of Leeds (2011).
3. Thanh Sang Nguyen et al.: Study on high performance fine-grained concrete containing
rice husk ash. International Journal of Concrete Structures and Materials, p.301–307
(2014).
 Experimental Study on Effect of Ground Granulated
Blast Furnace Slag of Strength and Durability of Sand
Concrete

Nguyen Thanh Sang1 and Nguyen Tan Khoa2

1
University of Transport and Communications, Hanoi, Vietnam
2
Dong A University, Da Nang, Vietnam

Abstract. Concrete used for building structures in the marine environment

requires a high level of durability, and as such an effective method to improve
the level is the use of ground granulated blast furnace slag (GGBFS). The paper
presents an experimental study on the effect of the replacement cement by 20%,
30%, 40%, 50% GGBFS on workability, mechanical properties and durability
of sand concrete (SC). The results show SC containing 20% GGBFS had the
highest compressive strength, splitting tensile strength, elastic modulus and
abrasion resistance. The lowest chloride penetration corresponded to SC
containing 30% GGBFS.

Keywords: Sand concrete, GGBFS, chloride penetration, fly ash, durability.

1 Introduction

Sustainable development by using industrial waste and local materials is the

current trend of the concrete industry in Vietnam and in the world. In Vietnam,
millions of tons of fly ash, bottom ash and slag has been discharged annually from
thermal power plants, iron and steel refining and this has created serious
environmental pollution[1]. In the central coastal provinces of Vietnam, there is huge
dune sand but it has not been used in traditional concrete because of the small
fineness module. Meanwhile, these materials can fabricate SC and apply for various
construction fields.

Sand concrete (SC) is a type of fine-grained concrete produced by mixing fine

aggregates, cement, mineral additives, water and chemical additives[2, 3]. In sand
concrete, the type of aggregate used is mainly sand, so SC is popularly used in some
regions and countries with a lot of scarce coarse aggregates such as France, Russia,
Morocco, Algeria, Germany, etc [4]. Like normal concrete, it is entirely possible to
fabricate sand concretes with strength up to 70 MPa [5-7]. However, in order to be
able to be used indifferent environmental conditions, especially harsh environment
such as marine environment, durability requirements of SC need to be met.

© Springer Nature Singapore Pte Ltd. 2020 409

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_63
 410 N. T. Sang and N. T. Khoa

The paper presented an experimental study on fabrication of SC with high strength

and durability using local fine sands and industrial waste products. In particular, the
study focused on investigating the influence of GGBFS replacing a part of cements to
workability, mechanical properties and durability of sand concrete.

2 Experiment Setup

2.1 Material and Mixture proportions

The study used cement PC40 which conforms to TVCN 2682: 2009. Fly ash (FA)
conforms to TCVN 10302: 2014. GGBFS conforms to TCVN 11586: 2016. The
chemical composition and physical properties of cement, FA and GGBFS are
presented in Table 1. Fine sand (FS) - Local sand has its properties and grading
shown in Table 2. MasterGlenium ACE 8509 conforms to ASTM C494 type F.

Table 1.Chemical composition and physical properties of cement, fly ash and GGBFS

Chemical Mean particle Density

CaO SiO2 Fe2O3 Al2O3 MgO K2O Na2O SO3 LOI
analyses (%) size (µm) (g/cm3)
Cement PC40 63.2 21.9 3.3 5.72 1.1 0.3 0.12 1.9 0.19 16.12 3.13
Fly ash 4.27 53.9 6.7 21.8 1.45 3.4 0.67 0.2 6.27 26.87 2.2
GGBFS 34.7 36.6 0.03 12.91 7.78 1.26 0.52 1.45 0.01 20.73 2.87

Table 2. Sieve analysis and properties of fine sand

Sieve size Fineness Density Absorption

9.5 4.75 2.36 1.18 0.6 0.3 0.15 0.075
(mm) modulus (g/cm3) (%)
Passing (%) 100 100 100 100 100 12.5 1.0 0 1.85 2.71 2.24

Table 3. Mixture proportions of sand concrete investigated

Mixture proportion CSC 20%SL 30%SL 40%SL 50%SL

Cement PC40 (kg/m3) 410 320 280 240 205
GGBFS (kg/m3) 0 90 130 170 205
Fly ash (kg/m3) 510 510 510 510 510
Fine sand (kg/m3) 1190 1190 1190 1190 1190
Water (lit/m3) 185 185 185 185 185
SP (%) 0.8 0.8 0.8 0.8 0.8
The selection of aggregate mix for SC was based on the principle of optimizing the
compacted volume of dry mix [3, 8]. The most optimal dry mix of compacted volume
had the ratio of FA: FS determined to be 30:70. The proportions of SC mixture were
designed on the basis of the absolute volume of constituent materials [4]. GGBFS was
incorporated with the replacement level 0%, 20%, 30%, 40%, 50% by cement weight.
One control mixture without GGBFS was prepared for comparison purpose. Mixture
proportions of sand concretes are shown in Table 3
Experimental Study on Effect of Ground Granulated … 411

2.2 Preparation of Test Specimens

Slump flow test was conducted to evaluate the workability of concrete mixtures
[9]. The samples were molded and cured to test concrete properties according to
ASTM C192. The compressive strength, splitting tensile strength, elastic modulus
was determined according to ASTM C39, ASTM C496, and ASTM C469,
respectively. Chloride penetration testing was executed according to ASTM C1202 at
28 days. Abrasion test was executed at 28 days according to TCVN 3114: 1993.

3 Results and Discussion

3.1 Workability
The slump flow of SC mixtures is presented in Figure 1. When the proportion of
GGBFS increased, the slump flow quickly rose to the highest value at 20% SL, then
decreased slightly and almost unchanged at higher proportions of GGBFS. Due to
better dispersibility than cement particles, smooth surface and imperviousness of
GGBFS during mixing, the friction in concrete mixtures using GGBFS was lower
than that of mixtures with only cement [10, 11]. However, fresh concrete mixtures
will have the best workability when they have the most reasonable proportion of fine
powder. With this proportion, the finer particles just fit into the hollows of larger
particles and the particles easily move and roll each other, which leads to the least
friction and the best slump flow. With replacement rates of 30%, 40%, 50% GGBFS,
the slump flow of fresh SC mixtures reduced slightly and almost unchanged.

3.2 Compressive strength, splitting tensile strength and elastic modulus.

The compressive strength of SC mixtures at 3, 7 and 14 days is shown in Figure 2.
At the age of 3 days, the compressive strength of SC using GGBFS was lower than
that of the control sand concrete (CSC). At 7 days, the compressive strength of 20%
SL and 30% SL was higher than CSC. At 28 days, the compressive strength of 20%
SL, 30% SL and 40% SL was higher than CSC, and that of 50% SL was approximate
to CSC. It can be seen that the compressive strength tended to increase at later age for
SC using GGBFS. The reason is that the chemical composition of GGBFS contains
more silica and less lime than Portland cement, so the product is hydrated using a
cement-slag mixture as a binder to produce more C-S-H products and less limestone
than cement. The resulting microstructure of the hydrated cement paste is denser.
However, the initial hydration of GGBFS is very slow because it depends on the
breakdown of glass by hydroxyl ions released during the hydration of Portland
cement[10]. At 7 and 28 days, the highest compressive strength was 56 MPa and 85
MPa corresponding to 20% SL. That is because at the proportion of 20% GGBFS, the
amount of cement is large enough to develop early strength in combination with the
filling effect and densification of the microstructure of GGBFS, so it’s the best ratio
of GGBFS for compressive strength. This result agreed with Mohd Nasir and Khatib
[12, 13].
412 N. T. Sang and N. T. Khoa

35 100 3 days 7 days 28 days

Compressive strength (MPa)

30
80
Slump flow (cm)

25
20 60

15 40
10
20
5
0 0
CSC 20%SL 30%SL 40%SL 50%SL CSC 20%SL 30%SL 40%SL 50%SL

Fig. 1. Slump flow of SC mixtures Fig. 2. Compressive strength of SC mixtures

6 50

Elastic modulus (GPa)

Splitting tensile strength

5 40
4
30
3
(MPa)

20
2
1 10

0 0
CSC 20%SL 30%SL 40%SL 50%SL CSC 20%SL 30%SL 40%SL 50%SL

Fig. 3. Splitting tensile strength of SC Fig. 4. Elastic modulus of SC mixtures

The splitting tensile strength of sand concrete mixtures ranges from 4.21 MPa to
5.38 MPa are shown in Figure 3. The splitting tensile strength of 20% SL was
approximate to the CSC while that of all other GGBFS ratios was lower than the
CSC. This result is caused by the fact that 20% SL has enough cement to develop
strength and facilitate early hydration of GGBFS [10]. However, with other ratios of
GGBFS, the slow hydration will affect splitting tensile strength.
Results of elastic modulus of SC mixtures range from 45.9 GPa to 47 GPa, which
are shown in Figure 4. The elastic modulus of SC mixtures using GGBFS was
approximately the same and slightly higher than the CSC. Thus, the use of GGBFS
replacing a part of cement does not negatively affect the elastic modulus of SC or
even produces better results.

3.3 Mechanical Abrasion Index and Chloride penetration

Mechanical abrasion index of SC mixtures at 28 days is shown in Figure 5. When
increasing the replacement rate of GGBFS, the results ranged 0.091 g/cm2 to 0.132
g/cm2, and the lowest value was 0.091 g/cm2 corresponding to 20% SL. The higher
abrasion index is, the lower mechanical abrasion resistance is. As reported in [10, 14],
compressive strength is one of the most important influence factors and is directly
proportional to the abrasion resistance, so the research results show rationality as the
abrasion index tends to be inversely proportional to compressive strength.
Chloride penetration of SC mixtures is expressed by charge passed as shown in
Figure 6. The charge passed of SC mixtures ranged from 247 coulombs to 500
Experimental Study on Effect of Ground Granulated … 413

coulombs, which were “very low” according to ASTM C1202. This reduction is
probably due to the combination of fly ash and GGBFS in SC. In particular, GGBFS
in concrete will create a large amount of C-S-H gel to reduce and distribute the pore
structure, increase the micro structural consistency of concrete [15, 16]. Fly ash with a
slower pozzolanic reaction consumed Ca(OH)2 generated from hydration products of
cement, forming CSH gel - reducing pore structure and improving long-term
durability of concrete [17].
The result shows that chloride penetration levels tended to decrease then increase
again when the ratio of GGBFS increased. The lowest charge passed was 247
coulombs, corresponding to 30% SL. The optimized GGBFS ratio for strength was
20% SL, but the optimized GGBFS ratio for chloride penetration was 30% SL.
The minimum of chloride penetration may be caused by the fact that the ratio of
30% GGBFS is just enough to promote the best role of GGBFS with filler and
participate in hydration reaction to reduce the pore structure of traditional cement.
Meanwhile, the amount of cement is also high enough to produce a large amount of
Ca(OH)2 for pozzolanic reaction of FA.
With a GGBFS ratio less than 30%, a less amount of GGBFS reduces its role in
improving concrete microstructure but increases pozzolanic reaction of FA. On the
contrary, with a GGBFS ratio more than 30%, a high amount of GGBFS and a lower
amount of cement makes the hydration process slower, the level of hydration of slag
at 28 days will not be much and the reaction of FA will be slower, which in turns
increases ability of chloride penetration of the concrete.
0.14 2000 2000
Charge pass (Coulomb)

0.12
Abrasion index (g/cm2)

1500 Low
0.1
0.08
1000 1000
0.06
Very
0.04 500 low
0.02
100
0 0
CSC 20%SL30%SL40%SL50%SL CSC 20%SL 30%SL 40%SL 50%SL
Fig. 5.Abrasion of SC mixtures Fig. 6.Chloride penetration of SC mixtures

4 Conclusions

From results obtained for the workability, the strength and durability properties of
SC mixtures in this study, the following conclusions can be drawn:
GGBFS improves workability of SC when replacing cement due to its smooth and
impervious surface. The type of 20% SL has the largest slump flow because its
powder composition creates the most reasonable proportion.
The strength properties of SC show optimal results at the replacement rate of re
20% GGBFS and the downward trend when the replacement rate of GGBFS increases
due to the late hydration characteristics of GGBFS.
414 N. T. Sang and N. T. Khoa

The abrasion of SC using GGBFS and FA is quite low; the lowest value
corresponds to 20% SL concrete; Other GGBFS replacement rates have greater
abrasion than CSC.
The obtained Chloride penetration of SC mixtures using GGBFS and FA was
"very low" according to ASTM C1202. The lowest Chloride penetration corresponds
to a GGBFS replacement rate of 30% GGBFS. This is a balance ratio between the
durability improvement mechanism of GGBFS and fly ash.

References
1. Hiếu, T.T., Nghiên cứu ứng dụng BTXM tro bay làm mặt đường ô tô ở Việt Nam, Khoa
Công Trình, Đại học Giao Thông Vận Tải. (2016).
2. NF-P18-500, A.S., Bétons de sables: French. (1995).
3. Bouziani, T., M. Bederina, and M. Hadjoudja: Effect of dune sand on the properties of
flowing sand-concrete (FSC). International Journal of Concrete Structures and Materials
6(1), 59-64. (2012).
4. Sablocrete: Synthèse du Projet National de Recherche et Développement, Bétons de sable,
caractéristiques et pratiques d’utilisation, Presses de l’école Nationale des Ponts et
Chaussées LCPC. (1994 ).
5. Sang, N.T., Nghiên cứu thành phần, tính chất cơ học và khả năng ứng dụng bê tông cát để
xây dựng đường ôtô ở Việt Nam, Đại học Giao thông Vận tải. (2010).
6. Dalila Benamara, L.Z., Bouzidi Mezghiche: High Performance sand concrete, formulation,
physico-machanical properties and durability. Benamara Project. (2015).
7. Le, H.T., S.T. Nguyen, and H.-M. Ludwig: A study on high performance fine-grained
concrete containing rice husk ash. International Journal of Concrete Structures and
Materials 8(4), 301-307. (2014).
8. Chauvin, J.J.a.G., G.: Les bétons de sable. Bulletin de liaison Laboratoires des Ponts et
Chaussées (LCPC) No. 157, 9-15. (1988).
9. EFNARC, The European guidelines for self-compacting concrete, Specification,
Production and use, The European Federation of Specialist Construction Chemicals and
Concrete Systems. (2002).
10. A.M, N., Properties of concrete. Fifth Edition. Pearson Education, London. (2011).
11. ACI233R-03, Slag Cement in Concrete and Mortar ACI Committee, American Concrete
Institute. (2011).
12. Khatib, J.M. and J.J. Hibbert: Selected engineering properties of concrete incorporating
slag and metakaolin. Construction and Building Materials 19(6), 460-472. (2005).
13. Mohd Nasir, N.A., Abd Aziz, F.N.A., Safiee, N.A: Hydration of the combinations of
ground granulated blast furnace slag cements. Australian Journal of Basic and Applied
Sciences 8(1), 5. (2014).
14. Horszczaruk, E.: Abrasion resistance of high-strength concrete in hydraulic structures.
Wear 259(1-6), 62-69. (2005).
15. Daube, J. and R. Bakker, Portland blast-furnace slag cement: a review, in Blended
Cements, ASTM International, (1986).
16. Basheer, P., et al.: Monitoring electrical resistance of concretes containing alternative
cementitious materials to assess their resistance to chloride penetration. Cement and
Concrete Composites 24(5), 437-449. (2002).
17. Rafat Siddique , M.I.K., Supplementary Cementing Materials. Verlag Berlin Heidelberg:
Springer. (2011).
 Mechanical properties of fly ash based geopolymer
concrete using only steel slag as aggregate

Dong Van Dao1 and Son Hoang Trinh1

1
University of Transport Technology, 54 Trieu Khuc, Thanh Xuan, Hanoi, Vietnam

Abstract. Geopolymer concrete (GPC) is a sustainable and eco-friendly con-

struction material in which industrial byproducts containing Si and Al such as
fly ash and slag are polymerized using an alkaline activating solution to create
hardened binder or inorganic polymer cement. This study evaluated GPC mix-
tures using fly ash as a geopolymer binder and granular slag materials in place
of natural coarse and fine aggregates. Three GPC mixtures were designed to
achieve specified compressive strengths (f'c) of 25, 30, 35 MPa by varying (1)
mass of fly ash between 482 to 507 kg/m3 of GPC, (2) the ratio (by mass) of al-
kali activating solution to fly ash (AAS/FA) between 0.4 and 0.48, and (3) the
ratio (by mass) of sodium silicate (Na2SiO3) to sodium hydroxide solution
(NaOH) (or SS/SH) between 2 and 3. The mixtures were prepared for testing in
hardened conditions. The results showed that GPC specimens (air-cured) yield-
ed good compressive strength (between 34.80 and 44.85 MPa), flexural strength
(ranging from 4.50 to 5.91 MPa), elastic modulus (between 30.99 and 35.50
GPa). These properties are likely a result of the high strength of slag aggregates
and a chemical reaction between steel slag aggregates and the geopolymer
binder, improving the interfacial transition zone and GPC microstructure. This
study suggests that slag aggregates can be used in place of natural aggregate
materials to improve GPC’s mechanical properties with a conventional curing
process for construction applications.

Keywords: Steel slag aggregate, fly ash based geopolymer concrete.

1 Introduction

To reduce greenhouse gas emissions generated from construction activities, innova-

tive materials have been developed over the last decades. One of such materials is
geopolymer binder [1]. Unlike the process of manufacturing conventional Portland
cement using virgin materials, such as limestone and coal, a geopolymer binder is
mainly made of by-products and can be considered a sustainable material. A
geopolymer binder can be mixed with aggregates to make geopolymer concrete
(GPC). Based on past studies [2, 3], GPC can perform as well as conventional Port-
land cement concrete (PCC) in both structural and nonstructural applications.
While GPC is a sustainable material that can use in many applications, its main disad-
vantage is slow hardening process occurring in ambient conditions. Previous studies
[2-4] suggested that GPC should be cured at a temperature as high as 60 oC to promote

© Springer Nature Singapore Pte Ltd. 2020 415

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_64
416 D. V. Dao and S. H. Trinh

polymerization and develop strength. The high-temperature curing process requires

more energy and limits the use of GPC in structures that are cast on site. Thus, re-
search has been conducted to improve the hardening process of GPC without the
high-temperature curing to increase its applications beyond precast structures.
Also, since slag is released abundantly from steel and iron plants as a waste in Vi-
etnam, use of slag is necessary to reduce waste and to replace virgin materials. In a
past study [5], a steel slag aggregate was used as coarse aggregate in GPC, and its
compressive strength was approximately 6% higher than that of the same mixture that
was made of natural coarse aggregate. Additional research is needed to use steel slag
materials locally available in Vietnam for both fine and coarse aggregates.
In this project, crushed granular steel slag was used as coarse and fine aggregates in
three fly ash-based GPC mixtures with designed specified compressive strengths (f'c)
of 25, 30, 35 MPa. The project evaluated mechanical properties of GPCs for use in
structural applications. In addition, the study also evaluated whether the slag aggre-
gates could help promote polymerization process of geopolymer and improve me-
chanical properties without using the high-temperature curing.

2 Materials and mixture proportions

2.1 Materials

Constituent materials used to prepare GPC mixtures are fly ash, alkali activating solu-
tion, water, fine and coarse aggregates. Fly ash used in this study was classified as
Class F in accordance with ASTM C618-03 [6]. This material was obtained from the
Pha Lai Thermal Power Plant in Vietnam. The chemical compositions of the fly ash
are presented in Table 1.

Table 1. Chemical compositions of fly ash (wt. %)

SiO2 Al2O3 Fe2O3 CaO MgO K2O Na2O TiO2 SO3 LOI*
51.74 24.53 5.59 0.81 1.95 4.42 0.11 0.76 0.31 8.98
*LOI: Loss in ignition

a) Coarse aggregate (4.75-19 mm) b) Fine aggregate (0.15-4.75 mm)

Figure 1. Aggregate Gradations
The alkali activating solution prepared by mixing a sodium hydroxide solution (molar
 Mechanical properties of fly ash based geopolymer … 417

mass of 12M) with sodium silicate solution. Steel slag was utilized to prepare aggre-
gates for this study. This material was collected from Thai Nguyen Iron and Steel
Industrial Factory in Vietnam. The slag was crushed and then fractionated into two
fractions. The coarse aggregate fraction included particles between 4.75 and 19 mm,
and the fine aggregate fraction included particles between 0.15 and 4.75 mm. The
gradations of the two fractions were determined in accordance with ASTM C33 [7]
and are plotted in Figure 1.

2.2 Mix proportions

Since the purpose of this study was to design GPCs that could be used for structural
applications, three mixtures were designed to achieve specified compressive strengths
(f'c) of 25, 30 and 35 MPa in accordance of ACI211-91 [8]. These mixtures were de-
signed with the constant amount of aggregates while varying the proportions of other
constituent materials as shown in Table 2.

Table 2. Mix proportions for three GPCs evaluated in this study

Mix 1 Mix 2 Mix 3
Description (f'c= 25 MPa) (f'c= 30 MPa) (f'c= 35 MPa)
kg/m3
Fly ash 482 492 507
Sodium hydroxide solution 57.7 56.5 52.1
Sodium silicate solution 173 165 154
Fine aggregate 716 716 716
Coarse aggregate 1671 1671 1671
Water 35.2 34.5 31.9

2.3 Mixing and preparing samples

Figure 2. Prepared samples

Constituent materials was mixed in accordance with ASTM C192 [9] using a
forced 60-liter concrete mixer with a dry mixing period of 3 minutes and a wet
mixing period of 5 minutes. The three mixtures were then used to cast specimens
for mechanical testing. Two sets of specimens were prepared, including 90 cylin-
ders (150x300 mm) and 90 beams (150x150x600 mm). After 48 hours, the speci-
mens were demolded and cured in the laboratory at 22±4°C with a humidity level
of 60±20% as shown in Figure 2.
 418 D. V. Dao and S. H. Trinh

3 Results and discussions

3.1 Mechanical properties

The specimens prepared for mechanical testing were tested after being cured for 28
days to determine the compressive strength, flexural strength and elastic modulus, as
shown in Figure 3. The tests were conducted in accordance with the respective ASTM
procedures [10-12]. Test results are presented in Table 3.

Figures 3. Compressive strength, flexural strength, elastic modulus tests

Table 3. Results for compressive strength, flexural strength and elastic modulus of GPCs
De- Standard Coefficient of Specified
Sam- Average
scriptio GPC mixture Deviation Variation value
ples (MPa)
n (MPa) COV (%) (MPa)
Com- Mix 1 30 34.80 3.23 9.271 27.29
pressive Mix 2 30 39.97 3.20 8.012 32.50
strength Mix 3 30 44.85 3.34 7.455 37.06
Elastic Mix 1 30 30989 1526 4.926 27433
modu- Mix 2 30 32547 1881 5.782 28162
lus Mix 3 30 33500 1016 3.035 31131
Mix 1 30 4.50 0.26 5.763 3.90
Flexural
Mix 2 30 5.09 0.32 6.265 4.35
strength
Mix 3 30 5.91 0.32 5.434 5.16

The results shown in Table 3 suggested that the three GPCs using the steel slag ag-
gregates can meet the requirements [9] for compressive strength, flexural strength and
elastic modulus after being cured in the conventional curing condition. The use of
steel slag may accelerate the polymerization process of the geopolymer binder.

A correlation between compressive strength and elastic modulus is shown in Figure

4a. As shown in Figure 4a, the elastic moduli of the three GPCs made of steel slag
aggregates were greater than those of the GPCs using natural aggregates [14] and
were close to the values calculated according to AASHTO 2012 [13]. The results
showed the important contribution of steel slag aggregates to improve the elastic
moduli of the three GPC mixtures that were cured at the conventional curing condi-
tion instead of at the evaluated temperature (i.e., 60 oC). Figure 4b shows a correlation
of the flexural strengths and the respective compressive strengths. The flexural
strengths of the three GPCs were greater than those of PCC [15] at the same compres-
sive strengths.
 Mechanical properties of fly ash based geopolymer … 419

Figure 4. Correlation between specified compressive strength and modulus of elasticity (a), and
flexural strength (b).

3.2 Interfacial Transition Zone

The interfacial transition zone (ITZ) was evaluated for the GPCs using a Scanner
Electron Microscope (SEM) in this study. The ITZ microstructure tested as shown in
Figure 5a, was compared with (1) that of another GPC using fly ash and natural basalt
aggregate (Figure 5b) and (2) that of a PCC using the same steel slag aggregate (Fig-
ure 5c). All of the specimens evaluated in this experiment were cured for six months
in the laboratory at a temperature of 22±4°C and a humidity of 60±20%.

Figure 5. SEM micrographs of ITZ for (a) GPC with steel slag aggregate; (b) GPC with natural
basalt aggregate; and (c) PCC with steel slag aggregate.

As shown in Figure 5c, there was a microcrack at the interface between the steel slag
aggregate particle and the cement paste, weakening the ITZ. While there was no
microcrack shown in Figure 5b, and a clear transition zone between the basalt aggre-
gate particle and the geopolymer matrix, suggesting no chemical reactions at the inter-
face of the basalt aggregate particle and the geopolymer matrix. Figure 5a shows no
microcrack nor a clear ITZ between the steel slag aggregate particle and the
geopolymer matrix, suggesting an improved microstructure of the ITZ and a possible
chemical reaction at the interface that can improve the bond between the steel slag
aggregate and the geopolymer matrix. The results suggested that the GPCs using the
 420 D. V. Dao and S. H. Trinh

steel slag aggregates can be cured in the conventional condition in place of the sug-
gested high temperature condition (i.e., at 60 oC).

4 Conclusions and Recommendations

In this study, three fly-ash based GPCs using the steel slag materials were evaluated
for structural applications. Based on the results of this study, the following conclusion
and recommendations can be drawn:
 Steel slag aggregates can be used as both coarse and fine aggregates (in place of
natural aggregate materials) for making fly-ash based GPCs.
 The GPCs using coarse and fine steel slag aggregates can meet the requirements
for compressive strength and other mechanical properties for use in structural applica-
tions after being cured in the conventional curing condition without the need for the
previously suggested high temperature (i.e., 60 oC) curing process.
 The SEM micrographs showed an improved microstructure at the ITZ for the GPC
using the steel slag aggregate. This helps explain the higher compressive strength,
flexural strength and elastic modulus obtained for the GPCs evaluated.

References
1. Davidovits J. Geopolymer (2011): Chemistry & Applications. Institute Geopolymer.
2. D. Hardjito and B. V. Rangan (2005), Development and Properties of Low Calcium Fly Ash
Based Geopolymer Concrete, Research Report GC1, Curtin University of Technology, Aus.
3. S.E. Wallah and B.V. Rangan (2006), Low calcium fly ash based geopolymer concrete: Long
term properties, Research report GC2, Curtin University of Technology, Australia.
4. Tran Viet Hung, Dao Van Dong and Nguyen Ngoc Long (2016), "Mix design for low calcium
of fly ash base Geopolymer concrete", The 7th International Conference of Asian Concrete
Federation “Sustainable Concrete for Now and the Future”, Hanoi, Vietnam;
5. O. M. Omar and et al (2015), effect of local steel slag as a coarse aggregate on properties of
fly-ash based geopolymer concrete, Int. Journal of Civil, Environmental, Structural, Construc-
tion and Architectural Engineering Vol:9, No:11.
6. ASTM C618-03 (2003), "Standard Specification for Coal Fly Ash and Raw or Calcined Natu-
ral Pozzolan for Use in Concrete".
7. ASTM C33 “Standard Specification for Concrete Aggregates”.
8. ACI211-91 “Standard Practice for Selecting Proportions for Normal, Heavyweight, and Mass
Concrete”.
9. ASTM C192 "Standard Practice for Making and Curing Concrete Test Specimens in the Lab".
10. ASTM C39 (2003), "Standard Test Method for Compressive Strength of Cylindrical Concrete
Specimens".
11. ASTM C78 (2002), "Standard Test Method for Flexural Strength of Concrete".
12. ASTM C469/C469M - 14, “Standard Test Method for Static Modulus of Elasticity and Pois-
son’s Ratio of Concrete in Compression”.
13. AASHTO 2012, "AASHTO LRFD Bridge Design Specifications”.
14. Hardjito et al (2005), The stress-strain behaviour of fly ash-based geopolymer concrete. De-
velopments in mechanics of structures and materials, A Balkema Publishers, The Netherlands.
15. ACI 318M (2011), “Building Code Requirements for Structural Concrete and Commentary”.
 Effect of fly ash on the mechanical properties and drying
shrinkage of the cement treated aggregate crushed stone

Ho Van Quan(1); Nguyen Van Tuoi(2); Chau V. Nguyen(3)

(1)
The University of Danang - University of Technology and Education, Danang, Vietnam;
(2)
Central Transport College No.V, Danang, Vietnam.
(3)
Institute of Research and Development, Duy Tan University, 03 Quang Trung, Danang,
Vietnam

Abstract. This paper presents the influence of fly ash on the mechanical prop-
erties and drying shrinkage of cement treated aggregate crushed stone (CTACS)
mixture. The fly ash can replace 20-40% the weight of cement for the mixture
of 4% CTACS, while it can supplement 3-9% the weight of aggregate crushed
stone for both mixtures of 3% and 4% CTACS. Furthermore, the experimental
findings indicated that the reduction in the compressive strength and the split-
ting tensile strength of CTACS was inversely proportional to the fly ash content
replaced partially by the cement. However, the increase in the compressive
strength and the splitting tensile strength of CTACS was directly proportional
to the fly ash content. It was observed from the test results that the drying
shrinkage of all mixtures was smaller than the control mixture without fly ash.
The drying shrinkage of mixtures reduced inversely to the proportion of the fly
ash content when 20-40% cement was replaced by the fly ash. When adding 3-
9% fly ash, the 6% fly ash gave the smallest drying shrinkage.

Keywords: Aggregate crushed stone; cement and fly ash treated aggregate
crushed stone; compressive strength; splitting tensile strength; drying shrinkage.

1 Introduction

Aggregate crushed stone (ACS), which is the most popular material, is often used in
bases of pavement structure in Vietnam. However, under disadvantage conditions
such as high category pavements having large traffic volumes or long-lasting wet
subgrade and pavement, the strength of ACS is usually deteriorated, thereby resulting
in the unstable pavement and the damage to surface course of flexible pavement such
as rutting, cracking, potholes, etc. In order to improve the strength and stability of the
ACS layer, the cement treated aggregate crushed stone (CTB) with the cement content
from 3 to 8% by weight of dry ACS is commonly used. The high stiffness of CTB
layer reduces the vertical deflection and tensile strain of the top layer materials. Be-
sides, the failures of pavement such as cracks and potholes are reduced because CTB
layer prevents moisture and stabilizes strength in wet conditions. In addition, subbase
and subgrade failures are decreased because CTB spread traffic loads over wide areas
and can span weak subgrade locations [1]. Although CTB showed many advantages,

© Springer Nature Singapore Pte Ltd. 2020 421

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_65
422 H. V. Quan et al.

it required a high construction cost. Moreover, this structure exhibited the high drying
shrinkage causing cracks, which can develop on the top layers and damage on the
surface of asphalt layer. These cracks lead to damage of the structure and reduce ser-
vice life of the pavement. Under the real condition, the CTB shows the large drying
shrinkage due to the water evaporation. Previous studies [2, 3, 4, 5, 6, 7] illustrated
that using a certain amount of fly ash (FA) can significantly reduce drying shrinkage
of CTB structure.
This study carried out experiments to determine the mechanical properties and dry-
ing shrinkage over time of cement and fly ash treated aggregate crushed stone
(CFATB) mixtures in the laboratory conditions up to 90 days. The results
demonstrated the change of drying shrinkage of CFATB with curing time as well as
the optimum value of fly ash.

2 Materials and methods

2.1 Materials

Aggregate crushed stone: Using Aggregate crushed stone type I, Dmax = 25 mm in

Danang. The particle size distribution, mechanical and physical properties of ACS are
in accordance with Vietnamese standard TCVN 8859:2011 [8] as shown in Table 1.
Cement: Using 40 MPa grade Kimdinh Portland blended cement conform to Viet-
namese standard TCVN 6260:2009 [9] and TCVN 8858:2011 [10].
Fly ash: Using fly ash (FA) of Nongson - Quangnam Thermal Power Plant. The
physical properties and chemical compositions of FA follow to ASTM C618 [11].

Table 1. The particle size, mechanical and physical properties of aggregate crushed stone, Dmax
= 25 mm

The particle size distributions

Sieve size (mm) 37.5 25.0 19.0 9.50 4.75 2.36 0.425 0.075
Passing Percentage (%) 100.0 86.9 72.6 56.1 41.3 31.1 16.6 5.7
Specification TCVN 8859:2011 (%) 100.0 79-90 67-83 49-64 34-54 25-40 12-24 2-12
The mechanical and physical properties
Order Specifications Test results Specification TCVN 8859:2011
1 Los Angeles (L.A) abrasion, % 27.9 ≤ 35
2 CBR index, % 135.0 ≥ 100
3 Liquid Limit, (%) 21.32 ≤ 25
4 Plastic index, (%) 4.58 ≤6
5 Elongation and flakiness index, (%) 12.4 ≤ 15

2.2 Test methods

Proctor compaction test.

In this study, the mixtures using 3% and 4% cement by weight in combination with
3%, 6% and 9% FA by weight of dry ACS were chosen. Additionally, 20%, 30% and
Effect of fly ash on the mechanical properties … 423

40% of the cement were replaced by the fly ash for CFATB having 4% cement by
weight.
The compaction test of CFATB mixtures was conducted according to Vietnam
standard 22TCN 333-2006 after removing oversize particles on 19.0 mm sieve [12].
The mould is 152 mm in diameter and 117 mm in height, respectively. The weight of
tamping rod is 4.5 kg and height of fall is 475 mm. Each testing sample was divided
into 5 layers and each layer was compacted with 56 times. The optimal moisture
content (W0) and the maximum dry density (γ max) of the CFATB mixtures are shown
in Table 2.

Table 2. Mixtures proportions for 100 kg of dry ACS

ACS W CM FA CM/ACS FA/ACS W0 max

Symbol W/CM W/B FA/CM
(kg) (lit) (kg) (kg) (%) (%) (%) (g/cm3)
3CM0FA 100 5.69 3.0 0.0 3.0 0.0 1.90 1.90 0.00 5.52 2.32
3CM3FA 100 6.59 3.0 3.0 3.0 3.0 2.20 1.10 1.00 6.22 2.36
3CM6FA 100 7.44 3.0 6.0 3.0 6.0 2.48 0.83 2.00 6.83 2.34
3CM9FA 100 8.33 3.0 9.0 3.0 9.0 2.78 0.69 3.00 7.44 2.28
4CM0FA 100 5.90 4.0 0.0 4.0 0.0 1.47 1.47 0.00 5.67 2.35
4CM3FA 100 6.79 4.0 3.0 4.0 3.0 1.70 0.97 0.75 6.35 2.40
4CM6FA 100 7.67 4.0 6.0 4.0 6.0 1.92 0.77 1.50 6.97 2.37
4CM9FA 100 8.55 4.0 9.0 4.0 9.0 2.14 0.66 2.25 7.57 2.30
4CM20FA 100 6.12 3.2 0.8 3.2 0.8 1.91 1.53 0.25 5.88 2.34
4CM30FA 100 6.31 2.8 1.2 2.8 1.2 2.25 1.58 0.43 6.07 2.33
4CM40FA 100 6.54 2.4 1.6 2.4 1.6 2.73 1.64 0.67 6.29 2.32
Note: W = Water; B (Binder) = CM + FA;

Experımental methods.
For the strength development, the preparation of the samples was conducted follow-
ing Table 2. After casting, all samples were covered by a moisture-proof evaporation
rag in 20 hours. Subsequently, all samples were demoulded and put into curing con-
tainer (covered with moisture sacks and sprayed with water to retain moisture) in
seven days. Finally, all specimens were immersed in the water at 250C until testing
days.
For drying shrinkage experiment, the samples were 150 mm in diameter and 300
mm in height. The samples were prepared following Table 2 and each sample was
divided into 12 layers. All samples were cured in moisture condition at 250C and
100% humidity in seven days. Afterwards, the value of drying shrinkage was meas-
ured.
The compressive strength and the splitting tensile strength of CFATB samples
were measured according to TCVN 8858 [10] and TCVN 8862 [13], respectively.
The measurement of drying shrinkage of CFATB samples was in accordance with
ASTM C490 [14] and ASTM C341 [15].

3 Results and discussion

424 H. V. Quan et al.

1.0 12.0
28 days 28 days

10.01
Splitting tensile strength, MPa
11.
11.0

Compressive strength, MPa

0.9

9.63
14 days 14 days

0.74

9.14
10.
10.0

0.70
0.8

8.85
0.67
7 days

8.52
0.66
7 days

0.63
9.0

8.06
0.7

0.60

7.50
7.30

7.17
0.52
8.0

0.51
0.6

6.87
0.48
0.47

6.50
0.44
0.5 7.0
0.36

5.47
0.4 6.0
0.3 5.0
0.2 4.0
0.1
0. 33.00
0 2.0
0.0 3.0 6.0 9.0 0.0 3.0 6.0 9.0
Addition of FA (%) Adition of FA (%)
a. Splitting tensile strength b. Compressive strength
Fig. 1. Influence of fly ash on strength characteristics of 3% cement treated ACS mixtures

1.2 28 days 14.0 28 days

Splitting tensile strength, MPa

1.09

12.50
12.18
1.0

Compressive strength, MPa

1.05

1.1 14 days 13.0 14 days

11.59
0.99

0.96

11.00
7 days
0.92

1.0 12.0 7 days

10.61
0.88

10.18
0.84

0.82

0.9 11.0

9.94
0.79

0.76

9.41
0.74

9.37
0.71

0.8
0.71

10.0

9.04
8.99
0.69

0.68

8.68
0.66

8.62

8.56
0.65

8.30
0.7 9.0

7.96

7.39
0.53

7.12
0.51

0.6 8.0
0.49

0.47

6.42

6.15
0.5 7.0

5.84
0.4 6.0
0.3 5.0
0.2 4.0
0.1 3.0
0 2.0
0 20 30 40 3.0 6.0 9.0 0 20 30 40 3.0 6.0 9.0

The FA replaces CM (%) Addition of FA (%) The FA replaces CM (%) Addition of FA (%)
a. Splitting tensile strength b. Compressive strength
Fig. 2. Influence of fly ash on strength characteristics of 4% cement treated ACS mixtures

0.090
0.085 1
5
0.080
0.075
0
Drying shrinkage, %

0.070
0 6
0.065
0 11
0.060
0 7
4
0.055
0 9
0.050
0 8
3
0.045
0 2
10
0.040
0
0.035
0 3CM0FA (1)
0.030
0 3CM3FA (2)
3CM6FA (3)
0.025 3CM9FA (4)
4CM0FA (5)
0.020 4CM20FA (6)
0.015 4CM30FA (7)
4CM40FA (8)
0.010 4CM3FA (9)
0.005 4CM6FA (10)
4CM9FA (11)
0.000
2 4 6 10 14 18 22 26 30 36 42 50 60 70 80 90
Time (days)
Fig. 3. The drying shrinkage of CFATB mixtures over time
Effect of fly ash on the mechanical properties … 425

The experimental results of this study are presented from Fig. 1 to Fig. 3. The
values of experimental results were the average value deriving from three samples.
The following comments can be drawn based on the experimental observations:
- The strength development of CFATB reduces when increasing the FA
replacement from 20% to 40%. It can be explained that the high amount of FA leads
to the reduction of cement content as well as the increase in the water to cement ratio
and water to binder ratio, as seen in Table 2. The slow hydration reaction of FA
contributes to the low strength development of CFATB.
- Regarding the different levels of FA addition by weight, the strength
developments of CFATB are improved when increasing FA contents. This
observation is attributed to the reduction of the water to binder ratio with a constant
cement content, as demonstrated in Table 2. Furthermore, the addition of FA
contributes to the hydration process and forms more the hydration products such as C-
S-H gel and C-A-S-H gel, which improves the strength of CFATB samples.
- As shown in Fig. 3, the drying shrinkage significantly increases in the curing time
from 2 to 30 days and slightly increases in the later age days. Especially, the drying
shrinkage of CFATB specimens seems not to develop in the curing time from 50 to 90
days. This phenomenon can be explained by the fact that in the early ages, the fast
cement hydration reaction leads to a significant development of the drying shrinkage.
Additionally, the water evaporation process in these samples contributes to the high
drying shrinkage in the early ages. In the later ages (after 30 days), the combination of
the completed hydration process and balance of humidity in the samples causes a
slight change of drying shrinkage.
- The drying shrinkage reduces when the partial replacement of cement by FA
increases from 20% to 40%. This result agrees with Cho Et. al [7].
- The drying shrinkage of CFATB specimens reduces when adding FA from 3% to
6%, while the higher addition of FA (6-9%) causes the increase of drying shrinkage.
The addition of FA with a smaller amount (3-6%) fills into the voids of the aggregates
and the microstructure of the samples becomes denser. The more FA is added, the
higher porosity and the higher humidity these specimens become. The high humidity
accelerates the evaporation process of CFATB samples and remarkably increases the
drying shrinkage of these mixtures.

4 Conclusions

Some conclusions can be drawn as follows:

1. The strength development of CFATB significantly reduces with the increase of
partial FA replacement to cement (20-40%) and remarkably increases with FA
addition from 3% to 9%.
2. The drying shrinkage of CFATB significantly increases in the early age of
curing. These values slightly increase in the age of 30 to 50 days and seem not to
develop in later age from 50 to 90 days. The increase in the FA replacement to cement
causes the decrease of drying shrinkage of these mixtures. The addition of FA
426 H. V. Quan et al.

decreases the drying shrinkage values. The lowest drying shrinkage values at 90 days
of curing were 0.048% and 0.043% of 3CM6FA and 4CM6FA mixtures, respectively.
In this study, the addition of 6% FA in (3 and 4)% cement treated ACS mixtures
was the optimal value according to laboratory works with the maximum size of
aggregates of 19 mm. However, there are many different conditions between
construction site and laboratory works such as over sized aggregates, temperature,
humidity, shrinkage surface area, friction coefficient with the underlying foundation,
etc and the development of strength over time. Therefore, it is necessary to observe
these optimized mixtures (3CM6FA and 4CM6FA) and reference mixtures (3CM0FA
and 4CM0FA) on construction sites to give the exact conclusions.

References
1. Adaska, W.S., and D.R. Luhr. Control of Reflective Cracking in Cement Stabilized Pave-
ments, 5 th International RILEM Conference, Limoges, France, pp.309-316, May, 2004.
2. El-Rahim, A., and George, K.P., Optimum Cracking for Improved Performance of Ce-
ment-Treated Bases, Proceedings, Ninth World Conference on Transport Research, Seoul,
Korea, July 2001.
3. George, K.P., Soil Stabilization Field Trial, Department of Civil Engineering, The Univer-
sity of Mississippi University, Mississippi, April, 2001.
4. Kota, P.B.V.S, Scullion, T. and Little, D.N., Investigation of Performance of Heavily Sta-
bilized Bases in Houston, Texas, District, Transportation Research Record 1486, Washing-
ton, D.C., 1995.
5. Sastry, D., Improving Soil Cement Characteristics with Fly Ash, Master’s Project, De-
partment of Civil Engineering, The University of Mississippi, University, MS. 1998.
6. George, P.K., “Minimizing Cracking in Cement-Treated materials for Improved Perfor-
mance”, Research and Development Bulletin RD123, Portland Cement Association, Skok-
ie, IL, 2002.
7. Yoon-Ho Cho, Kang-Won Lee, Sung-Woo Ryu, Development of Cement Treated Base
Material for Reducing Shrinkage Cracks. Transportation Research Record: Journal of the
Transportation Research Board, Vol. 1952, pp. 134-143, Jan, 2006.
8. TCVN 8859, Graded Aggregate Bases and Subbases Pavement - Specification for Con-
struction and Acceptance, Vietnam, 2011.
9. TCVN 6260, Portland blended cement – Specifications, Vietnam, 2009.
10. TCVN 8858, Cement Treated Aggregate Bases for Road Pavement - Specification for
Construction and Acceptance, Vietnam, 2011.
11. ASTM C618, Standard Specification for Coal Fly Ash and Raw or Calcined Natural Poz-
zolan for Use in Concrete, 2008.
12. 22TCN 333, Standard test method for compaction of soil and aggregate crushed stone in
the laboratory, Vietnam, 2006.
13. TCVN 8862, Standard test method for splitting tensile strength of aggregate material
bonded by adhesive binders, Vietnam, 2011.
14. ASTM C490, Standard Practice for Use of Apparatus for the Determination of Length
Change of Hardened Cement Paste, Mortar, and Concrete, 2007.
15. ASTM C341, Standard Practice for Length Change of Cast, Drilled, or Sawed Specimens
of Hydraulic-Cement Mortar and Concrete, 2010.
 Strength of Granulated Blast Furnace Slag during
Hydration Reaction Process

Tran Thanh Nhan1()[0000-0001-6029-3077], Hiroshi Matsuda2, Tran Xuan Thach3,

Nguyen Dai Vien4 and Ho Trung Thanh1
1 University of Sciences, Hue University, Hue city, Vietnam
2 Yamaguchi University, Ube city, Yamaguchi, Japan
3 Ha Tinh Department of Construction, Ha Tinh city, Vietnam
4 Thua Thien Hue Department of Construction, Hue city, Vietnam

ttnhan@hueuni.edu.vn

Abstract. Granulated Blast Furnace Slag (GBFS) is a by-product of iron pro-

duction process in which the molten slag from the blast furnace is cooled down
rapidly by the highly pressurized water. In many counntries, GBFS has been
widely used as a geo-material in civil works, meanwhile in Vietnam, due to the
lack of technical standards and development, iron and steel slags including
GBFS have not been used in construction. In this paper, firstly, typical proper-
ties and chemical content of Formosa GBFS (produced by Formosa Steel Plant
in Ha Tinh province, Vietnam) were observed and it was shown that physico-
mechanical properties and chemical components of Formosa GBFS are similar
to those of GBFS produced in Japan which have been widely used in the port
and habor works as alternative sands. Secondly, the hydraulic property and its
effects on the unconfined compressive strength were then clarified for different
GBFS samples and based on which, an estimation method of unconfined com-
pressive strength was then developed.

Keywords: Granulated blast furnace slag, Hydraulic property, Unconfined

compressive strength.

1 Potential Usage of GBFS as a Geo-Material in Construction

Blast Furnace Slag (BFS) forms when iron ore is melted together with auxiliary mate-
rials including lime stone and coke in blast furnace and the amount of BFS generation
is roughly 300 kg per ton of pig iron. BFS is divided into Air-Cooled Blast Furnace
Slag (ABFS) by cooling the molten slag in open pits or yards and Granulated Blast
Furnace Slag (GBFS) by quenching the molten slag under high pressurized water. In
Vietnam, iron and steel industry has developed rapidly and therefore the generation of
iron and steel slag increases considerably and the annual production is predicted to
exceed 12 million tons in 2020 [1].
In order to clarify the applicability of GBFS produced in Vietnam as an alternative
geo-material to the natural sands, typical geotechnical properties and chemical

© Springer Nature Singapore Pte Ltd. 2020 427

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_66
428 T. T. Nhan et al.

components of Formosa GBFS are compared with those of Japan products as shown
in Table 1 (GBFS-A, GBFS-B and GBFS-C in Table 1 are produced by different blast
furnaces in Japan). The observed data suggests that geotechnical properties of GBFS
are more advantageous than those of natural sand and that, the chemical components
of Formosa GBFS are similar to those of Japanese products which have been
confirmed to satisfy the environmental requirements [2]. In addition, GBFS shows the
similar particle size to the natural sand as shown in Fig. 1(a) and so, GBFS is adopted
as a fine aggregate for concrete and motar by Vietnamese standard [3]. The
compaction curves of Formosa GBFS are shown in Fig. 1(b) which includes those of
GBFS produced in Japan and the natural seabed sand. The dry densities of GBFS
show less affected by the change of water content compared with the seabed sand and
this means the easier compaction works at the construction site.

Table 1. Properties and chemical components of GBFS produced in Japan and Vietnam.

Physical properties and chemical content (%)

GBFS
s k10-4 IL
samples e emin CaO SiO2 Al2O3 MgO T-S FeO MnO
(g/cm3) max (m/s) (%)
GBFS-A 2.730 1.378 0.807 7.91 0.537 43.45 34.02 14.36 5.95 0.7 0.56 0.12
GBFS-B 2.680 1.571 0.967 9.22 0.092 42.65 34.23 14.54 6.46 0.9 0.43 0.27
GBFS-C 2.755 1.402 1.044 8.26 0.085 40.06 37.28 15.04 5.48 0.9 0.49 0.25
Formosa GBFS 2.790 1.398 0.881 28.6 0.140 41.09 34.72 13.15 7.37 0.6 0.40 -

(a) (b)
100 2.0
GBFS-A Seabed sand in Japan (Matsuda [2])
Percent finer by weight (%)

GBFS-B
Dry density, d (g/cm3 )

GBFS-C
Formosa GBFS
Formosa GBFS
Upper and lower
boundaries of
TCVN 7570:2006

GBFS produced in Japan (Matsuda [2])

0 1.2
0.01 0.1 1 10 0 Moisture content, w (%) 30
Grain size, d (mm)

Fig. 1. (a) Grain size distribution curves and (b) typical compaction curves of GBFS produced
in Japan and Vietnam.

2 Hydraulic Property of GBFS and Hydration Reaction Test

Since GBFS is produced by rapidly quenching the molten slag, mineral structure of
the particle is under unstable condition resulting in its latent hydraulic property by
which, GBFS solidifies in natural wet conditions without any additives. Therefore,
Strength of Granulated Blast Furnace Slag … 429

such a hydraulic property and its effect on physico-chemical properties of GBFS and
the environment should be clarified when promoting effective application of GBFS as
a geotechnical material.

2.1 Hydration Reaction Test

In this study, in order to evaluate the solidification of hydrated GBFS, the hydrate was
measured by the hydration reaction test proposed by Kondo and Ohsawa [4]. After
unconfined compression test, wet sample of hydrated GBFS was dried for hydration
reaction test. Because of the limitted page of this paper, procedure of the hydration
reaction test should be referred to Matsuda and Nhan [5]. By this method, the hydra-
tion reaction ratio (R, %) can be obtained by Eqs. (1) and (2).
mh (1)
Ri 
 IL 
md 1     msa
 100 
R = 100 - Ri (2)
where Ri (%) is the un-hydrated ratio; md and mh (g) are the mass of GBFS sample
before immersed into compound solution and after heating, respectively; IL (%) is the
ignition loss of un-hydrated GBFS; msa (g) is the mass of salinity in test sample (to
minimize the effect of the dissolution of salinity when curing solution is seawater).

2.2 Test Specimen and Condition

Materials used for observation of the hydraulic property are unhydrated samples of
GBFS-A, GBFS-B and GBFS-C (Table 1). In order to prepare test specimen, the mass
of dried sample at Dr = 80% was determined for each GBFS and then mixed with
curing solution which is seawater. After keeping for one day, the solution-mixed
GBFS was de-aired in the vacuum cell before poured into plastic mold of 50mm in
diameter and 100mm in height by the water pluviation method. In order to accelerate
the hydration reaction of GBFS, specimens were cured under the constant temperature
of T = 80ºC and the curing duration was set from t = 0 day to t = 168 days. After
curing, unconfined compression test and hydration reaction test were carried out.

3 Results and Discussions

3.1 Changes in the Hydration Reaction Ratio of GBFS with Curing Duration
In Fig. 2, changes in the hydration reation ratio (R) and its increment (R = R - R
where R is the hydration reaction ratio of unhydrated GBFS) with the curing duration
(t) are shown and a linearly increasing tendency of R and R with t can be confirmed
for all GBFS samples. The gradient of such a tendency suggests that the rate of the
hydration reaction is the highest on GBFS-A and the lowest on GBFS-C. Since the
unique testing preparation and condition were applied to the specimen and the grain
size of test materials is similar to each other (Fig. 1a), the differences of the hydration
430 T. T. Nhan et al.

reaction rate in Fig. 2 should be explained based on further considerations on produc-

tion process and the chemical content of GBFS and such problems are out of the
scope of this research.

8.0 8.0
GBFS-A (a) (b)

Increment of hydration reaction ratio

GBFS-A
Hydration reaction ratio R (%)

GBFS-B GBFS-B
GBFS-C GBFS-C

R (%)
4.0 4.0

0.0 0.0
0 100 200 0 100 200
Curing duration t (days)

Fig. 2. Changes in (a) the hydration reaction ratio (R) and (b) its increment (R) with the curing
duration (t) for all GBFS samples.

3.2 Effects of the Hydration Reaction on the Unconfined Compressive

Strength of GBFS
The changes in the unconfined compressive strength (qu) with the curing duration (t)
are shown by symbols in Fig. 3(a) and it is seen that qu increases with t. When com-
paring the tendencies in Fig. 2(b) with those in Fig. 3(a), it is indicated that GBFS
sample with higher hydration reaction rate (i.e. higher value of R) shows higher
unconfined compressive strength.

400 300
GBFS-A (a) GBFS-A, B and C (b)
Unconfined compressive strength

GBFS-B
GBFS-C

200
qu (kN/m2 )

200

100

qu = 47.585R + 22.305
R2 = 0.8339
0 0
0 100 200 0.0 3.0 6.0
Curing duration t (days) Increment of hydration reaction ratio R (%)

Fig. 3. (a) Changes in qu with t and (b) relations between qu and R for all GBFS samples.

The curing condition adopted in this study is to accelerate the hydration reaction on
GBFS and consequently to accelerate the increase in its strength compared with those
under ordinary in-situ wet conditions. Observed results in Figs. 2(b) and 3(a), howev-
Strength of Granulated Blast Furnace Slag … 431

er, show that the hydration reaction occurs continuously and the strength of GBFS
continues to increase. Matsuda [2] confirmed that the hydration reaction of GBFS
continues to occur even after 11 years of solidification in natural wet condition as a
full-scale embankment and this is an important characteristic for the application of
GBFS as an alternative material to natural sand in geotechnical engineering. In Fig.
3(b), relations between qu and R are shown by symbols for all specimens. In this
study, specimen was prepared at Dr = 80% and at this relative density, unhydrated
GBFS can maintain the specimen shape by its initial strength. Therefore, relation
which follows a solid line in Fig. 3(b) suggests that the shear strength of GBFS is
constituted of two components as expressed by Eq. (3). The first component, denoted
by qu, is considered as “primary strength” by the internal friction of unhydrated
GBFS particles. The second component, denoted by quR, is considered as “secondary
strength” due to hydrate-made bonding between hydrated GBFS particles.

qu = qu + quR (3)

Also in Fig. 3(b), quR proportionally increases with R and the relation can be ex-
pressed by Eq. (4) as follows:

quR = A  R (4)

where A is the experimental constant.

3.3 Prediction of the Unconfined Compressive Strength of GBFS

Relationships between R and the root square of curing duration (t, days) are shown
in Fig. 4(a) for all GBFS samples. Symbols in this figure are the observed results and
solid lines conrrespond to linear relations which are expressed by Eq. (5) as follows:

R = C  t (5)

where C is the experimental constant depending on the hydration reaction rate of

GBFS.

(a) (b) (c)

8.0
GBFS-A GBFS-B GBFS-C
400 GBFS Observed Calculated
400
GBFS-A
A
R = 0.3501t B GBFS-B
Observed qu (kN/m2 )

R2 = 0.8857 C GBFS-C
qu (kN/m2 )

R = 0.3023t
R (%)

R2 = 0.9041
R = 0.1569t
R2 = 0.7237

0.0 0 0
0 t (days) 15 0 t (days) 15 0 Calculated qu (kN/m2) 400

Fig. 4. (a) Relations of R versus t, (b) and (c) comparisons between observed and estimated
results of qu for all test specimens.

Consequently, the unconfined compressive strength of GBFS can be expressed by

Eq. (6) as follows:
432 T. T. Nhan et al.

qu = qu + A  C  t (6)

Relations between qu and t are shown in Fig. 4(b) for all test specimens. Symbols
in this figure show the experimental results and solid lines correspond to the calculat-
ed ones by using Eq. (6). Further comparisons between the observed and calculated
results are shown in Fig. 4(c). Reasonable agreements in Figs. 4(b) and 4(c) indicate
the validity of the developed method for estimating the strength of GBFS produced
from different steel plants.

Conclusions

The main conclusions are as follows:

Typical properties and chemical components of Formosa GBFS are similar to those
of GBFS produced in Japan indicating the more advantages compared with natural
sand and the satisfaction of environment requirements for using in construction.
The unconfined compressive strength and the hydration reaction ratio of GBFS in-
crease in the process of hydration reaction.
A proportionally increasing tendency between the increment of hydration reaction
ratio and the curing duration was confirmed for different GBFS samples.
The unconfined compressive strength of GBFS is constituted of two components:
the primary strength is due to the friction between GBFS particles and the secondary
strength is generated by hydrate-made bonding between the particles.
The secondary component of the unconfined compressive strength of GBFS in-
creases linearly with R regardless of the origin of GBFS sample and curing duration.
The unconfined compressive strength of different GBFS samples can be estimated
by using the developed method.

References
1. Hung, L.V.: Current situation of the regulations and standards respect to slag application in
Vietnam. In: Conference on Application of Iron/Steel Slag in Construction Industry for
Sustainable Development, pp. 41-52. Hanoi (2016).
2. Matsuda, H.: Application of granulated blast furnace slag as a geotechnical material. In:
2nd International Conference on Engineering Geology in Respond to Climate Change and
Sustainable Development of Infrastructure, pp. 75-84. Publishing House for Science and
Technology, Hanoi (2015).
3. TCVN 7570:2006 - Aggregates for concrete and mortar - Specification. Ministry of Sci-
ence and Technology, Hanoi (2006).
4. Kondo, R., Ohsawa, S.: Studies on a method to determine the amount of granulated blast
furnace slag and the rate of hydration of slag in cements. Ceramics Association Journal 77,
39-46 (1969).
5. Matsuda, H., Nhan, T.T.: Shear strength and permeability of granulated blast furnace slag
mixed with marine sand from low to high mixing ratio. In: 2rd International Conference on
Geological and Geotechnical Engineering in Response to Climate Change and Sustainable
Development of Infrastructure, pp. 63-70. Vietnam National University Press, Hanoi
(2016).
 Use of coal ash of thermal power plant for highway
embankment construction

Tung Hoang1, Viet Phuong Nguyen1 and Hong Nam Thai1

1
National University of Civil Engineering, 55 Giai Phong, Ha Noi, Viet Nam

Abstract. The highway embankment construction allows to consume a huge

volume of coal ash up to hundreds of thousands of tons per km. However, this
is a material with very different physical and chemical properties from tradi-
tional materials, which requires different design, construction and acceptance
solutions.
This article first presents the results of the laboratory experiments to evalu-
ate and classify coal ash according to the AASHTO M145 and ASTM C618,
TCVN 9436-2012 standards. These results help to define the composition, con-
struction and acceptance solutions of the road bed made of ash in Vietnamese
conditions. The proposed solutions have been experimented at a road section
constructed in 2017 in Ha Nam with the large scale (100m of grade IV, 12m
wide, 2.2m high, 960 tons of coal ash) and conduced to analytical results.

Keywords: subtainable construction materials, coal ash, embankment, high-

way.

1 Introduction
In Vietnam, according to the survey results of the coal ash emissions by the Ministry
of Construction, volume of coal ash is about 12.8 million tons by 2015; will be 25.4
million tons by 2020 and over 30 million tons by 2030; of which about 75% is fly
ash.
To take use of the coal ash, there are many solutions having been applied namely use
of coal ash in production of non-fire brick and cement. These solutions are however
not very effective given the fact that the volume of ash consumed in these fields is
limited. Thus there should be a more practical solution introduced, which could be the
use of coal ash in the construction of highway embankments. Nevertheless, relevant
studies are very rare in Vietnam with no practical experience recorded.
On contrary, in India and the United States of America, coal ash has been used as a
replacement of traditional filling materials in the construction of highway embank-
ments. Their studies and standards [1][2]...[6] indicate that compared to traditional
materials, coal ash has its own characteristics related to the composition of aggrigates,
intensity, water stability and environmental conditions. These factors help determine
the scope of application and contents during the design and construction of the roads.
Hence below studies have been conducted:
- Assessment of physical and chemical indexes with a focus put on the composition
of aggrigates, intensity, water stability in a laboratory.

© Springer Nature Singapore Pte Ltd. 2020 433

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_67
434 T. Hoang et al.

- Basic definition of the scope of application and measures to construct embank-

ments lined with coal ash through experimental construction at a road section with
100m long (see details in the relevant part below);
- Follow-up, measurement and assessment of experimental results;
Object of study is the coal ash collected from Nghi Son thermal power plant
2x300MW (Thanh Hoa Province, Vietnam). See details in the succeeding part.

2 Physical and chemical indexes to be studied and experimented.

Physical and chemical indexes to be studied shall be in line with the orientation as
described in Part 1. It shall also satisfy the design, construction and acceptance re-
quirements of highway embankments filled with coal ash as stated in existing Viet-
namese standards [6] [7]. The experimental method and results are presented in Table
1.
Table 1. Index and test methods
0
N Index Test methods Value
1 Wet and dry density TCVN 4195-2012 [10] Wet density: 171 g/cm3, Dry
TCVN 4202-2012 [11] density: 1.54 g/cm3
2 Aggregates AASHTO T27 [12] Average: 45m, more than 10%
passed sieve 75m
3 Optimal Moisture 22TCN 333-2006 [13], 12,5%
AASHTO T99 [14]
4 Expansion index 22TCN 333-2006 [13] 7%
5 CBR 22TCN 333-2006 [13], 2.6 (95%MDD) to 3.8 (100%
AASHTO T193 [15] MDD)
6 Optimal moisture: c =0,02 MPa,
Cohesion and internal
TCVN 4199:1995 [16] =18000’
friction angle
Saturation: c=0, =0
7 E50=2.01 MPa, qu=0,16MPa,
E50, qu, Cu ASTM D2166 [17]
Cu=0.08Mpa
8 Plasticity Index and
AASHTO T89&90[18] -
liquid limit
Experimental results indicate:
- Water stability of coal ash is very limited. When humidity is high, coal ash will
easily turn into liquid. At saturation condition, its cohesion and internal friction angle
are almost at 0. Determination of plasticity and liquid limit is also a challenge.
- Expansion is relatively high, which requires high water resistance.
- According to [9], with physical and chemical indexes as shown in Table 2, coal ash is
similar to the earth A3 – A4 with features tending to be A4 (dirt earth). This material hardly
satisfies required humidity, easy to be swept out when it is dry, with limited water stability.

3 Recommendations for application of coal ash in construction

of highway embankments
3.1 Scope of application
According to existing Vietnamese standards [19], coal ash can be used for filling up
highway embankments because coal ash is similar to the earth A3 – A4 with features
Use of coal ash of thermal power plant for highway … 435

tending to be A4 (dirt earth). However, it cannot be used for inundation areas or the
ones with underground water available and active zone of road bed.
Based on (i) the value CBR, coal ash classification and physical and chemical fea-
tures as presented in Table 1, (ii) reference to [19], coal ash should be used for em-
bankments from 0.8m to 1.5m from its subgrade down for embankments at Grade III
and IV. It should be 1.5m for Grade I, II. With the ones that are not A1, coal ash can be
directly applied;
In any case, water protection measures are always required to these embankments.

3.2 Solutions for embankment composition

In order to ensure water resistance, coal ash embankments can be designed as shown in
Fig 1 and Fig 2. In both cases, the embankments can be filled with cohesive earth at
slopes. Geotextile or reverse filter can be added inside the earth layer for water drainage.
Above the coal ash layer, a selective filling earth layer can be laid with thickness of 0.3m –
0.5m.
Selected earth (0.3-0.5m)

Cohesive earth (1m)

Coal ash (0.5m-3m)
Geotextile
Coal ash (0.5m-3m)

Fig. 1. Typical cross-section of embankment with alternate layer of selected earth

As for the alternate layer solution (Solution 1) (Figure 1), the thickness of the coal
ash layer is between 0.5m and 3m. Selective filling earth layers with the thickness of
0.3m to 0.5m should be added in between coal ash layers. The other solution (Solu-
tion 2) (Figure 2) is for the layer with purely coal ash applied with no alternate layers.
Selected earth (0.3-0.5m)

Cohesive earth (1m)

Geotextile
Coal ash (1.5m-3m)

Fig. 2. Typical cross-section of embankment with core of fly ash

To both solutions, selective filling earth layers play a very crucial role as covering
layers to prevent surface water to infiltrate into coal ash and prevent dirt when coal
ash is dry. Besides, these layers serve as a ground for convenient travel of construc-
tion vehicles and machines, even in bad weather conditions such as rains or high hu-
midity.

4 Transport and Storage of Coal Ash

Coal ash is loose and fine in composition. When being taken out of production fur-
nace, with humidity of almost 0 (dry), it is very dusty. Transport by trucks as normal-
ly practiced may not only reduce the humidity of the material but also cause environ-
mental pollution severely. As such it should be transported by land (specialized tank
trucks) or by waterway (by barges, then liquidizing it and pumping to the site as normally
applied for river sand (very popularly applied in Vietnam, especially in Southern region).
436 T. Hoang et al.

To store coal ash, there could be geotextile bags used. These bags are tailor-made
compatibly to the dimensions of embankments (if directly constructed), or to desired
dimensions as shown in Fig 3a and Fig 3b. On the surface of each bag, bag opening
(to connect the pumping pipe with the tank truck) can be arranged as shown in Figure
3a. This chart also describes the steps of this process during construction in order
ensure coal ash to be evenly placed over the whole site.
The advantage of this solution is to minimize dust exposed to the enrironment dur-
ing the pumping process, storage and construction, allowing watering the geotextile
for increasing humidity.
Construction method:
The experimental construction which can be seen in Part 3 shows that beside storage
1
5000 5000 5000
2000
3000
3000
2000

coal ash in-door

D120 mm, L=500 mm coal ash supplying schema

1
coal ash in-door Geotextile
D120 mm, L=500 mm

1000 1-1 1000

a. Bag design and position of bag opening b. Bag opening
Fig. 3. Coal ash bag design

and transport solutions as presented above, coal ash can be placed similarly to other
traditional materials. As for compaction, due to the small and even size of aggregates
with limited cohesive feature, coal ash can be easirly compacted. It can even be com-
pacted with a crawler excavator.

5 Experimental Results
5.1 Introduction of Experimental Project
The experimental construction section is located on Section T4, Experimental Training
Unit of National University of Civil Engineering at Ha Nam province.
Use of coal ash of thermal power plant for highway … 437

Selected earth (0.3 m)

Cohesive earth (1m)

Coal ash (0.8m)
Geotextile
Coal ash (0.8m)

Fig 4. Composition of alternate layers

Highway embankments Grade 4 (according to Vietnamese standard (TCVN) 4054-05
[20]), has been constructed with two methods applied namely alternate layers (section 1)
and purely coal ash (Section 2), each 50m long. The cross section of Section 1 is presented
at Figure 4. Nghi Sơn coal ash has physical and chemical features as mentioned in
Table 1, fill earth and alternate layers are clay loam exploited at Ha Nam (see Table
2).
Table 2. Characteristic of cohesive soil and selected earth
0
N Index Note Unit Value
Aggregates 0,5 – 0,08 mm % 8,1
1 0,08 – 0,005mm 56,2
< 0,005mm 35,7
2 Natural moisture W % 27,6
3 Dry density max g/cm3 1,38
4 Liquid limit WL % 45,6
5 Plasticity limit WP % 22,5
6 Plasticity index IP % 23,1
7 Cohesion C MPa 0,024
9 Internal friction angle  degree 21O16’
Section 2 has the cross section similarly to the description at Fig 2 and Fig 4 with
only one unique coal ash of 1.9m thick lying under the selective earth layer of 0.3m
thick. Density for coal ash is required to be between K95-K98, of which earth layer is
at K98 and slope at K95. Dry coal ash can be transported to site by tank trucks, then
pump to geotextile bags ART12, and made wet at a regulated level, laid and finally
compacted. The construction finished in January 2018.
5.2 Assessment of experimental results
During one year since January 2018, the embankment has been frequently supervised
and measured to gain data on geometric dimensions (1 year of supervision), expan-
sion index (using a hard steel slab), and radio active Aeff.
a. Geometric dimensions
After construction, there was a phenomenon recorded namely the expansion of the
edges. This horizontal deformation has slowly reduced and totally disappeared after
06 months with the maximum value measured being 56mm. It should be noted that
there had been a lot of heavy rains during 2018; however, the coah ash embankments
have remained stable.
438 T. Hoang et al.

b. Expansion Index
The expension index measured at the section with totally coal ash used (Solution 2) is
E=37MPa, which is suitable for the embankment Grade IV or below. With Solution 1, the
expansion index is very positive with E = 63 MPa, much higher than that of Solution 1.
The supervision indicates that the solution 1 (with alternate layer of earh) also
helps improve the water resistance of the embankment. As such the stability of the
whole system including slope surface, filling layers which lie on the constructed em-
bankment is all better than that of the other section.
c. Radio active
Index Aeff is 397 Bq/kg on average as recorded. According to regulations of [21],
this index can be up to 740 Bq/kg for industrial projects, urban roads and residential
areas. Hence Aeff of coal ash from Nghi Sơn thermal power plant for construction of
embankments, especially highways outside urban roads is suitable.

6 Conclusion
This article shows that the coal ash taken from Nghi Son thermal power plant owns
physical and chemical features which are suitable for construction of roads using
alternate layers or purely coal ash layers, at the active zone of embankments. In all the
cases, there must be water protection measures for the embankment and its slopes.
Application of coal ash in construction of embankments is a new trend in treatment
of coal ash disposed from thermal power industry of Vietnam.

References
1. American Coal Ash Association: Fly Ash Facts for Highway Engineers, Report No FHWA-IF-03-019 (2003).
2. ASTM International: Standard Guide for Design and Construction of Coal Ash Structural
Fills, ASTM E2277-14 (2014).
3. David J. White, Dale Harrington, and Zach Thomas: Fly ash Soil Stabilization for Non-
Uniform Subgrade Soils, Volume I: Engineering Properties and Construction Guidelines,
Final Report TR-461, Iowa State University (2005).
4. Indian Roads Congress: Guidelines for use of Fly Ash in Road Embankments, Special
Publication 58, New Delhi (2001).
5. Dallas N. Little, Syam Nair: Recommended Practice for Stabilization of Subgrade Soils and Base
Materials, Texas Transportation Institute, Texas A&M University, College Station, Texas (2009).
6. Ministry of Science and Technology: Highway embankments and cuttings – Construction
and quality control, TCVN 9436:2012, Vietnam (2012).
7. Ministry of Science and Technology: Coal ash of thermal power plant using as backfill
material – General requirement, TCVN 12249:2018, Vietnam (2018).
8. Tung H et al.: Use of coal ash of thermal power plant in highway embankment construc-
tion. Research of Ministry of Construction, RD 110-16-TX (2019).
9. AASHTO: Standard Specification for Classification of Soils and Soil-Aggregate Mixtures
for Highway Construction Purposes, AASHTO 145-91 (2017).
10. Ministry of Science and Technology: Soil - Method of laboratory determination of specific
weight, TCVN 4195-2012, Vietnam (2012).
11. Ministry of Science and Technology: TCVN 4202-2012. Soil - Laboratory method for de-
termination of unit weight
12. AASHTO: Standard Method of Test for Sieve Analysis of Fine and Coarse Aggregates,
AASHTO T27 (2014).
Use of coal ash of thermal power plant for highway … 439

13. Ministry of Transport: Standard compaction test method, 22TCN 333-2006, Vietnam (2006).
14. AASHTO: Standard Method of Test for Moisture–Density Relations of Soils Using a 2.5-
kg (5.5-lb) Rammer and a 305-mm (12-in.) Drop, AASHTO T99 (2018).
15. AASHTO: Standard Method of Test for The California Bearing Ratio, AASHTO T193 (2013).
16. Ministry of Science and Technology: Soil - Laboratoy method of determination of shear
resistance in a shear box apparatus, TCVN 4199:1995, Vietnam (1995).
17. ASTM: Standard test method for unconfined compressive strength of cohesive soil. American
Society for Testing and Materials Standards, D2166-00, Pennsylvania, USA (2000).
18. AASHTO: Determining The Plastic Limit and Plasticity Index of Soils, AASHTO T89&90
(2014).
19. Ministry of Science and Technology: Highway embankments and cuttings – Construction
and quality control, TCVN 9436:2012, Vietnam (2012).
20. Ministry of Science and Technology: Highway specification for design, TCVN 4054:05, Vietnam
(2005).
21. Ministry of Science and Technology: Activity admixture - Fly ash for concrete, mortar and
cement, TCVN 10302:2014, Vietnam (2014).
 Impact of the porosity of coarse aggregates on the
structuration of the paste-aggregate interface:
Elementary model study

Tran Duc Long 1, 2, Cassagnabère Franck 1, Mouret Michel 1

1
Université de Toulouse, UPS, INSA; LMDC (Laboratoire Matériaux et Durabilité des Cons-
tructions), 135, Avenue de Rangueil, F-31 077 Toulouse Cedex 04, France
2
University of Technology and Education, the University of Danang, Vietnam

Abstract. In the context of sustainable development for concrete production,

local aggregates and recycled aggregates should be preferred but they can be
porous and of lower quality. As a result, the compressive strength and the dura-
bility of concrete are adversely affected. In a preliminary understanding and
with a view to accentuating phenomena occurring in the concrete, the Elemen-
tary Model (EM composed of paste and gravel, was studied. In this paper, the
effect of Water Porosity (WP), moisture state and volume of coarse aggregates,
as well as the nature of mineral admixture on the WP of EM was taken into ac-
count. Results indicated that volume and water porosity of aggregate are the
two most important factors affecting WP of EM. Besides, it was also shown that
aggregate porosity has the same trend of impact on the WP of EM while does
not have the same trend of impact concerning the effect of the cement paste-
aggregate interface on the WP of EM whatever the nature of mineral admixture,
water porosity, moisture state and volume of aggregate. Explanatory attempts
are proposed from the Scanning electron microscope (SEM) in combination
with Energy dispersive spectroscopy (EDS) to support for statistic and ranking
analysis.

Keywords: Paste-aggregate interface, Mineral admixture, Porosity.

1 Introduction

The quality of the concrete depends upon the quality of cement paste and aggre-
gates, as well as the bond between them. Related works showed factors that affect the
properties of concrete, namely: the nature properties and moisture state of aggregates
[1] [2], the incorporation of mineral admixture [1] [3], and the volume content of
aggregates [4] [5]. These impact factors play important role on the quality of concrete,
but what extent of these impacts are still questionable.
In this paper, at the scale of elementary model (EM), composed by only paste and
gravel, the impact of water porosity (WP), moisture state (Saturated Surface Dry
(SSD) and Oven Dry (OD) state), and volume of aggregate on WP of EM was quali-
fied. Five aggregates differing in physical and chemical nature were used. Since the

© Springer Nature Singapore Pte Ltd. 2020 441

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_68
442 T. D. Long et al.

replacement of cement by mineral admixture not only reduces the cement content (for
environmental and economic benefits [6]) but also improves the properties of concrete
[7], four kinds of mineral admixture were also considered. Statistical method [8] and
ranking method [9] were applied to analyze the experimental results. The analysis was
then completed by the means of SEM/EDS observations/characterization.

2 Experimental procedure

2.1 Materials:

Raw materials. Table 1 shows the main chemical and physical properties of the
cement and the four mineral admixtures used.

Table 1. Characteristics of cement and mineral admixtures.

Cement I Limestone
Metakaolin Slag Fly ash
52.5 N filler
Physical characteristics
Density (g/cm3) 3.09 2.50 2.70 2.90 2.00
Specific surface 4600 16000 5960 4450 2300
(cm2/g) (Blaine) (BET) (Blaine) (Blaine) (Blaine)
Chemical / Mineralogical composition (wt. %)
SiO2 20.1 / 37.4 52.5
93.16
Al2O3 5.2 / 10.8 22.5
Fe2O3 3.3 / / 0.5 8.5
CaO 64.1 0.36 / 43.7 3.5
CaCO3 / / 95 / /

The five types of coarse aggregate along with the physical and chemical properties
are presented in Table 2.

Table 2. Characteristics of aggregates.

Water Water
Density Mineralogy
Rank Designation porosity absorption
(kg/m3) (>90% by mass)
(vol. %) (wt. %)
1 Concrete Recycled 2580 20,03 5,20 SiO2 , CaCO3
2 Soft limestone 2450 12,45 4,20 CaCO3
3 Siliceous-calcareous 2580 1,20 1,05 SiO2 , CaCO3
4 Hardened limestone 2670 0,88 0,56 CaCO3
5 Siliceous-quartz 2720 0,65 0,51 SiO2

Elementary Model (EM). The coarse aggregate volume (measured after 24-h im-
mersion in water) was fixed to 3.30cm3 ± 0.05 (V1), 7.30cm3 ± 0.05 (V2) and 9.80 cm3
± 0.05 (V3), which corresponded to two, four and six selected gravels to be incorpo-
rated in the paste. The pre-conditioning to obtain two moisture states were three days
Impact of the porosity of coarse aggregates … 443

immersed in water for Saturated Surface Dry (SSD) state and three days at 105°C for
Oven Dry (OD) state.
Five compositions of binder (water/binder ratio = 0.35) were considered. The ref-
erence composition was only cement, while the four other ones consisted in the re-
placement of cement with 15% metakaolin, or 25% limestone filler, or 30% slag or
30% fly ash by mass according to the maximum replacement rates fixed by the Euro-
pean standard EN 206/CN [10].
All EM were produced in cylindrical container (5cm-diameter and 4-cm height),
and cured in water at 20°C for 90 days.

2.2 Tests
After the curing period, the WP of EM was measured according to French standard
NF-P18-459 [11] (Eq. (1)):
(1)
Mair, Mdry, Mwater : SSD weight, OD weight and hydrostatic weight of EM, respective-
ly.
The SEM/EDS was a useful method to determine the chemical length of the inter-
facial transition zone (ITZ) based on the variation of the Ca, Si and C/S atomic ratios
through 100-point scans from the aggregate phase to the bulk paste (fig. 1).

Fig. 1. ITZ chemical length

3 Analysis method

3.1 Statistical method

A multi linear model was performed using the statistical environment R [8] in or-
der to evaluate and compare the impact of the studied factors on the WP of EM. In the
present case, there were 8 variables (X) (Table 3, column 2, row 1 to 8) and the ex-
444 T. D. Long et al.

pected variable (Yi) is WP of EM. Our dataset was big enough (450 samples including
3 times repeatability) and satisfied all assumptions of the model (linear relationship
between variables (Xi) and expected variables (Yi), independence between variables
(Yi), normal distribution of the residuals). The coefficient of determination R2 was
used to explain for the changing of the expected variable.

3.2 Ranking method

Aggregates were first ranked by decreasing porosity as presented in Table 2 (column
1). Then, when considering WP of EM, the ranking method [9], based on Kendall’s coef-
ficient of concordance, was applied to answer to three questions:
1. Is EM WP ranked in the same order as aggregate WP, whatever the binder nature?
2. Is the effect of the paste-aggregate interface on EM WP ranked in the same order
as aggregate WP, whatever the binder nature?
3. Is the chemical length of ITZ ranked in the same order as the aggregate WP,
whatever the binder nature?
Since the difference between EM WP or between EM ITZ length was not always
significant, the Kendall’s coefficient was calculated by considering ties case (Eq. (2)):
(2)
S = Variance of the sums of ranks; k = Judges (five binder natures); n = Contestants (1: EM WP; 2:
ITZ effect; 3: ITZ chemical length); t =Tie corresponding to a set s of ranks for a given judge.

4 Results and discussion

4.1 Result from multi linear regression model

When porous aggregates (rank 1 and 2, table 2; 180 EM samples out of 450) and
non-porous aggregates (rank 3, 4 and 5, table 2; 270 EM samples out of 450) are ana-
lyzed separately, it is clear that when porous aggregates are used, the moisture state
and the aggregates WP are factors influencing EM WP. In addition, the incorporation
of mineral admixtures such as limestone filler or slag becomes also influential factors.
When considering the five aggregates together, all variables explain for 93.85% the
change in the EM WP. The two most important variables, i.e. aggregate volume and
aggregate WP, account for approximately 75% and 11% the change in the EM WP,
respectively. Among all the remaining variables, only the slag has statistical meaning
explaining for 5.58% the change in the expected variable EM WP.

4.2 Result from the ranking method

Questions 1 and 2 (section 3.2) – Results from EM WP measurements. Con-
cerning question 1 (Table 4, column 3), the Kendall’s coefficient of concordance (W)
in all cases is higher than the threshold at the 0.01 significance level (0.571). Hence,
EM WP is classified in the same order as aggregate WP, whatever the nature of min-
eral admixture, but also the moisture state and the volume of aggregates.
Impact of the porosity of coarse aggregates … 445

Table 3. Coefficient of determination R2 of model and variables

R2 (%) (single)
Non porous Porous Both kinds
Variables aggregate aggregate of aggregate
(270 samples) (180 samples) (450 samples)

X1 Cement 0.00 0.0 0.00

X2 Metakaolin 0.03 0.11 0.00
X3 Limestone 0.06 1.34 0.28
X4 Slag 5.04 8.84 5.58
X5 Fly ash 0.00 0.00 0.00
X6 Moisture state of aggregate 0.02 2.22 0.12
X7 WP of aggregate 0.34 2.34 11.02
X8 Volume of aggregate 87.97 76.92 74.89
2
Coefficient of determination R (%) (Multi) 95.99 93.70 93.85

Table 4. Values obtained for the Kendall's coefficient of concordance W

Moisture state Aggregate Question 1 Question 2 Question 3 Threshold at 0.01

of aggregate volume (Section 3.2) (Section 3.2) (Section 3.2) Significance level
V1 0.808 0.333
SSD V2 0.956 0.068 0.311
V3 0.961 0.302
0.571
V1 0.632 0.124
0.372
OD V2 0.755 0.193
V3 0.633 0.567
To answer to question 2, the difference in EM WP between experimental method
(Eq. (1) - ITZ taken into account) and analytical calculations (Eq. (3) –ITZ not con-
sidered) was calculated. The W values shown in Table 4 (column 4) are lower than
the 0.01 significance level (0.571). This enable us to conclude that the effect of the
interface on the EM WP is not classified in the same order as aggregate WP whatever
the nature of mineral admixture, moisture state and volume of aggregate.
(3)
WP, V - Water porosity and volume of aggregate and paste; VEM - Volume of EM
Question 3 - Results from SEM-EDS observation. The chemical length of the
ITZ (fig. 1) was determined by the means of SEM/EDS tests in order to enhance the
conclusions of statistical and ranking results pertaining to EM WP measurements. The
chemical length of ITZ was then analyzed by ranking method (Table 4, column 5).
The resulting W values are lower than the level of significance at 1% (0.571). This
indicates that the ITZ chemical length of EM is not ranked in the same order as the
446 T. D. Long et al.

aggregate WP whatever the nature of the mineral admixture and moisture state of
aggregate.

5 Conclusions

From EM (paste + gravel(s)) investigations, the following conclusions can be drawn:

1. The two most important variables effect on EM WP are aggregate volume and
aggregate WP, which account for approximately 75% and 11% the change of EM
WP, respectively. The impact of moisture state and incorporation of mineral admix-
ture is important when porous aggregates are used. Particularly, slag presents remark-
able influence on the change in the WP of EM, whatever the aggregate WP.
2. Aggregate WP ranking is preserved for EM WP whatever the other parameters
studied (mineral admixture nature, moisture state and volume of aggregates) while not
preserved when evaluating the effect of paste-aggregate interface on EM WP, due to
the mineral admixture nature.
3. The dependence of the paste-aggregate interface structuration on the mineral
admixture nature is supported by SEM/EDS characterization enabling the determina-
tion of the chemical length of ITZ.

Reference
1. Kosmatka, S. H., Kerkhoff, B., Panarese, W. C.: Design and control of concrete mixtures.
14th edn. Portland Cement Association, Illinois (2011).
2. Alexander, M., Mindess, S.: Aggregates in concrete. 1st edn. CRC Press, London (2005).
3. Gonen, T., Yazicioglu, S.: The influence of mineral admixtures on the short and long-term
performance of concrete. Building and Environment 42(8), 3080-3085 (2007).
4. Meddah, M. S., Zitouni, S., Belâabes, S.: Effect of content and particle size distribution of
coarse aggregate on the compressive strength of concrete. Construction and Building Ma-
terials 24(4), 505-512 (2010).
5. Pereira, C. G., Castro-Gomes, J., de Oliveira, L. P.: Influence of natural coarse aggregate
size, mineralogy and water content on the permeability of structural concrete. Construction
and Building Materials 23(2), 602-608 (2009).
6. Cassagnabère, F., Mouret, M., Escadeillas, G., Broilliard, P., Bertrand, A.: Metakaolin, a
solution for the precast industry to limit the clinker content in concrete: Mechanical as-
pects. Construction and Building Materials 24(7), 1109-1118 (2010).
7. Duan, P., Shui, Z., Chen, W., Shen, C.: Effects of metakaolin, silica fume and slag on pore
structure, interfacial transition zone and compressive strength of concrete. Construction
and Building Materials, 44, 1-6 (2013).
8. Gandrud, C.: Reproducible research with R and R studio. 2nd edn. Chapman and
Hall/CRC, New York (2015).
9. Sheskin, D. J.: Parametric and nonparametric statistical procedures. 4th edn. Chapman and
Hall/CRC, Boca Raton (2000).
10. NF EN 206/CN: Concrete – Specification, performance, production and conformity. Na-
tional addition to the standard NF EN 206 (2014).
11. NF P18-459: Concrete – Testing hardened concrete – Testing porosity and density (2010).
 Strength and engineering properties of cementless paste
produced by GGBFS and MgO

Chao-Lung Hwang1, Duy-Hai Vo 2, Khanh-Dung Tran Thi1 and Mitiku Damtie

1
Yehualaw
1
Department of Civil and Construction Engineering
National Taiwan University of Science and Technology, Taipei, Taiwan
2
Faculty of Civil Engineering, University of Technology and Education,
The University of Danang, No. 48 Cao Thang street, Danang city, Vietnam
mikehwang@mail.ntust.edu.tw; duyhai88@gmail.com;
khanhdungtran412@gmail.com; mtkdmt2007@gmail.com

Abstract. This study aims to investigate engineering properties of cementless pastes which
were produced by reactive magnesium oxide (MgO) and ground granulated blast furnace
slag (GGBFS). The mixtures were designed in various levels of MgO at 2.5%, 5%, 7.5%,
10%, 15% and 20% of total binder weight. The slump flow test, compressive strength test,
ultrasonic pulse velocity test (UPV) and thermal conductivity test were conducted to
examine the engineering properties of the pastes until 28 curing days. The results indicate
that the high proportion of MgO causes the decrease of flow-ability of fresh pastes.
Increasing MgO content significantly promotes the hydration process and improves the
compressive strength and hardened properties of pastes through the UPV and thermal
conductivity testing results.

Keywords: cementless paste, compressive strength, engineering properties, thermal

conductivity.

1 Introduction
The increase of cement demand to adapt the sharp improvement of construction industry
has been generating a vast volume of CO2 during the decarbonization in cement
production process [1, 2]. Numerous studies have been carried out to research proper
alternative construction materials. Alkali-activated materials namely alkali-activated slag
and geo-polymers with the presence of sodium hydroxide (NaOH) and sodium silicate
(Na2SiO3) are considered as the current attractive replacements. These materials have
been proved to be eco-friendly and to satisfy the specification requirements such as high
compressive strength in early stage, low permeability, good chemical resistance, and
exceptional fire resistance [3, 4]. However, the residual alkali after reactions can
contaminate surrounding environment [5, 6]. Therefore, substituting these solutions by an
equivalently effective but less harmful substance is needed. Magnesium oxide (MgO), an

© Springer Nature Singapore Pte Ltd. 2020 447

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_69
448 C.-L. Hwang et al.

expansion agent, is potential to activate ground granulated blast-furnace slag (GGBFS) [1,
7] to produce alkali-activated slag (AAS). During the mixing process, MgO reacts with
water to create an alkaline environment.
MgO + H2O  Mg2+ + 2OH- (1)
The alkaline environment breaks covalent bonds (Si-O-Si; Al-O-Si) in slag particles. The
hydration reaction of MgO-GGBFS can be summarized as followed:
CaO-MgO-Al2O3-SiO2 + MgO + H2O  C-S-H + M-S-H + M-A-H + C-A-S-H (2)
(slag) (Ht)
However, the current researches concerning this issue are still limited. So, this study
attempts to use MgO as an activator for GGBFS in different contents: from 2.5% to 20%
by weight of binder. Some fresh and hardened properties of the pastes were investigated
by flow-ability tests, compressive strength tests, UPV, and thermal conductivity.

2 Materials and mix design

2.1 Materials

Table 1. Physical properties and chemical compositions of raw materials.

Items GGBFS MgO

Specific gravity 2.98 2.91

Physical
properties Mean particle size (μm) 14.563 5.598
2
Specific surface area (m /g) 1.44 2.12

SiO2 33.39 8.34

Al2O3 14.39 -
Fe2O3 0.19 0.14
Chemical CaO 41.08 0.42
composition MgO 7.22 90.4
(wt.%)
SO3 0.11 0.75
Basicity 1.88 -

Sulfide sulfur 0.72 -

In this study, GGBFS and MgO were the source materials. Both materials were collected
from a local company in Taiwan. Table 1. indicated the physical properties and chemical
compositions of GGBFS and MgO. Fig.1a and Fig.1b respectively give information about
particle size distributions (PSD) and XRD analysis of the raw materials.
Strength and engineering properties of cementless … 449

(a) PSD of raw materials. (b) XRD pattern of raw materials.

Fig. 1. Particle size distributions and XRD analysis of raw materials

2.2 Samples preparation

Table 2. Mix proportion of GGBFS-MgO cementless paste of 1000g powder.

Items w/b Slag MgO Water Flow-ability (mm)

M2.5 975 25 350 230
M5 950 50 350 230
M7.5 0.35 925 75 350 230
M10 900 100 350 220
M15 850 150 350 200
M20 800 200 350 180

In this study, the mixtures were prepared with GGBFS and MgO. The MgO content was
set at 2.5%, 5%, 7.5%, 10%, 15% and 20% of the total binder weight. The water-to-
binder ratio was fixed at 0.35 for all of mixtures. Firstly, MgO was dissolved in water in 2
mins, and then GGBFS was added and mixed in 2 mins. Finally, the rest water was mixed
in 3 mins to achieve a homogeneous fresh paste. After mixing, flow-ability of the paste
was tested by flow table test. The 50x50x50mm cubic molds were prepared for
compressive strength test while the cylinder 50x100mm samples were used for UPV test
and thermal conductivity test. All of the samples were covered by a thin film to prevent
water evaporation and were cured in ambient condition with 50 ± 5% humidity and at
25±2oC. After 24 hours, the samples were demolded and cured in a water container at
25±2oC for compressive strength and UPV test, while other cylinder samples were cured
in a chamber with 50 ± 5% humidity and at 25±2 oC for thermal conductivity analysis. The
450 C.-L. Hwang et al.

compressive strength test, thermal conductivity and UPV complied to ASTM C109,
ASTM C518 and C597, respectively.

3 Results and discussion

3.1 Flow-ability
Table 2. presents the flow-ability values of GGBFS-MgO cementless pastes. It can be
seen that the flow-ability was stable when the amount of GGBFS replaced by MgO
steadily increased up to 7.5% and this figure gradually declined when the MgO content
gained 10% and more. The decrease of workability was associated with the increase of
MgO level. Compared to GGBFS particles, the specific surface area of MgO is much
higher, resulting in the greater water absorption and thus, leading to the reduction of flow
values [8].
3.2 Compressive strength
The compressive strength of the pastes is illustrated in Fig. 2. As expected, the pastes
gained more strength with curing time. Throughout 28 days, the pastes became denser and
produced more strength due to the improvement of the amount of hydration products.
Additionally, the more MgO content incorporating to the mixtures, the higher
compressive strength the pastes will gain. Particularly, the compressive strength achieved
the highest values in the mixture with 15% wt.% MgO content, considerably in the early
ages. This phenomenon caused by the coexistence of C-S-H gel and hydrotalcite (Ht)
formed in the hydration processes [7-9]. However, as the MgO content exceeded 15%, the
compressive strength of the pastes began dropping. When MgO/GGBFS ratio increased,
the amount of alumino-silicate content declined, which partly mitigated the volume of C-
S-H gel. Furthermore, the rise of Ht, which is believed to be more voluminous than C-S-H
[7], properly led to the volume instability of the pastes structure [8].

Fig.2. Effect of MgO content on compressive strength development of cementless pastes.

Strength and engineering properties of cementless … 451

3.3 UPV results

Fig. 3 indicates the development of UPV values with curing up to the 28 th day. The
improvement of UPV values likewise was similar to that of the compressive strength [10].
The UPV results improved with increasing MgO levels. The figures were the highest in
the paste with 15 wt.% MgO of total binder and then going down as the MgO content was
over 15%. The explanation for these phenomena was alike to those of the compressive
strength as discussed above. C-S-H and Ht filled pores in the hardened pastes and formed
denser pastes, resulting in the higher values of UPV during the curing time. However,
more than 15% MgO content reduced the UPV values. The growth of MgO mass was
equivalent to the reduction of GGBFS in the mixture. That led to the corresponding
decrease of aluminosilicate and subsequent hydration products [8].

3.4 Thermal conductivity

Fig. 4 informs thermal conductivity data of the pastes under the curing period. Thermal
conductivity evaluated the heat conductivity capability of the pastes [11]. The previous
studies revealed several factors on thermal conductivity such as type of materials,
proportioning, moisture content, density, porosity, and curing condition. The thermal
conductivity recorded on 28th day was far lower than that on 7th day. The thermal
conductivity value declined due to the loss of water by evaporation and hydration process.
Specifically, on the 28th day, the thermal conductivity was the lowest in 2.5%-MgO
mixture. The thermal conductivity rose with increasing MgO content, which can be
attributed to the highly conductive nature of MgO [12].

Fig.3. Effect of MgO content on UPV Fig.4. Effect of MgO content on thermal
values of cementless pastes. conductivity of cementless pastes.
452 C.-L. Hwang et al.

4 Conclusion
Based on the experimental works of the present study, the following conclusions can be
drawn:
1. Using the high amount of MgO (over 10%) in the binder causes negative effects on
workability of the cementless pastes.
2. The compressive strength and UPV results of the paste samples significantly improve
with increasing MgO contents. The optimal mixture is the incorporation of 15% MgO and
85% GGBFS. The higher proportion of MgO reduces the strength and UPV value in
comparison with the optimal mixture.
3. Thermal conductivity of the cementless pastes gradually increases with the growth of
MgO content.

References
[1] K. Gu, F. Jin, A. Al-Tabbaa, B. Shi, J. Liu, Mechanical and hydration properties of ground
granulated blastfurnace slag pastes activated with MgO–CaO mixtures, Construction and Building
Materials 69 (2014) 101-108.
[2] S.A. Miller, V.M. John, S.A. Pacca, A. Horvath, Carbon dioxide reduction potential in the global
cement industry by 2050, Cement and Concrete Research 114 (2018) 115-124.
[3] C. Carreño-Gallardo, A. Tejeda-Ochoa, O.I. Perez-Ordonez, J.E. Ledezma-Sillas, D. Lardizabal-
Gutierrez, C. Prieto-Gomez, J.A. Valenzuela-Grado, F.C. Robles Hernandez, J.M. Herrera-Ramirez,
In the CO2 emission remediation by means of alternative geopolymers as substitutes for cements,
Journal of Environmental Chemical Engineering 6(4) (2018) 4878-4884.
[4] J.L. Provis, Alkali-activated materials, Cement and Concrete Research 114 (2018) 40-48.
[5] S. Paulose, P. Reddy, J. Kv, Swell Potential Studies on Soils Contaminated with NaOH
Solutions, 2014.
[6] R.V.P. Chavali, S.K. Vindula, H.P.R. P, A. Babu, R.J. Pillai, Swelling behavior of kaolinitic
clays contaminated with alkali solutions: A micro-level study, Applied Clay Science 135 (2017)
575-582.
[7] F. Jin, K. Gu, A. Al-Tabbaa, Strength and hydration properties of reactive MgO-activated
ground granulated blastfurnace slag paste, Cement and Concrete Composites 57 (2015) 8-16.
[8] C.-L. Hwang, D.-H. Vo, V.-A. Tran, M.D. Yehualaw, Effect of high MgO content on the
performance of alkali-activated fine slag under water and air curing conditions, Construction and
Building Materials 186 (2018) 503-513.
[9] S.A. Bernal, R. San Nicolas, R.J. Myers, R. Mejía de Gutiérrez, F. Puertas, J.S.J. van Deventer,
J.L. Provis, MgO content of slag controls phase evolution and structural changes induced by
accelerated carbonation in alkali-activated binders, Cement and Concrete Research 57 (2014) 33-43.
[10] T. Yılmaz, B. Ercikdi, K. Karaman, G. Külekçi, Assessment of strength properties of cemented
paste backfill by ultrasonic pulse velocity test, Ultrasonics 54(5) (2014) 1386-1394.
[11] I. Asadi, P. Shafigh, Z.F.B. Abu Hassan, N.B. Mahyuddin, Thermal conductivity of concrete –
A review, Journal of Building Engineering 20 (2018) 81-93.
[12] H. Yuan, Y. Shi, Z. Xu, C. Lu, Y. Ni, X. Lan, Effect of nano-MgO on thermal and mechanical
properties of aluminate cement composite thermal energy storage materials, Ceramics International
40(3) (2014) 4811-4817.
 A non-paraffinic PCM modified textile reinforced concrete sand-
wich panel
Zakaria Ilyes Djamai1, Amir Si Larbi1, Ferdinando Salvatore1
1
ENISE, Laboratoire LTDS, 58 Rue Jean Parot 4023 Saint-Etienne, France.
Zakaria-ilyes.djamai@enise.fr

Abstract

There is growing interest in the use of phase change materials (PCMs) in the
building industry, particularly in cementitious materials.
In the present study, a modified mortar matrix with different amounts of PCM
(5wt %, 10wt%, 15wt %) have been prepared. The different PCM-mortar ma-
trixes have been mechanically characterized. It was found that PCM drastically
decreases the mechanical performance of mortar.

The 10% PCM-mortar matrix has been reinforced with 2 layers of AR glass
Fabric and has been used as a skin of textile reinforced concrete (TRC) foamed
sandwich panel.

The performance of the 10% PCM modified TRC sandwich panel has been
compared with a reference TRC sandwich panel (without PCM).It has been
found that the mechanical performance of the PCM modified sandwich panel
decreases comparing to the reference, however TRC ductile behaviour is con-
served during bending which is very encouraging in view of developing a new
sustainable TRC sandwich panels with high mechanical and thermal efficiency

Keywords: phase change materials, textile reinforced concrete, sandwich pan-

els, building envelope.

1 Introduction
Textile-reinforced concrete (TRC) [1,2] consists of a fine-grained concrete reinforced by a
high-strength textile made of alkali-resistant glass which allows it to combine the high com-
pressive strength of concrete structures with the tensile resistance and noncorroding behaviour
of the textile fibers. TRCs emerge in modern building construction such as skins of light-weight
sandwich panels [3, 4]. However, thin TRC skins considerably affect thermal inertia of sand-
wich panels which makes their use difficult in building envelope. For all these reasons a PCM
modified TRC sandwich panel has been proposed in the paper, which combine both thermal
inertia and mechanical performance of TRC sandwich panels.

2 Materials and experimental methods

2.1 Materials

© Springer Nature Singapore Pte Ltd. 2020 453

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_70
454 Z. I. Djamai et al.

2.1.1 Encapsulated PCM

In the present study, a commercially available PCM (INERTEK 23) was used. This
PCM is a non-paraffinic fatty ester, derived from organic vegetal wax; it is chemically
identified as methyl hexadecanoate. Raw PCM with particle sizes in the range of 5–
25μm was encapsulated in a polymeric shell.
The melting point of the PCM as determined by DSC (Figure1) at a scanning rate
of 0.05°C/min is in the range of 23–27°C, with a peak at 25°C. The latent heat of
phase change is 160 kJ/kg.

80
Specific heat capacity (kJ/kg.K)

60

40
PCM powder

20

0
18 20 22 24 26 28 30 32 34 36
Temperature T (scanning rate of 0.05°C/min)

Figure 1 Specific heat capacity of microencapsulated PCM

2.1.2 Mortar
Portland cement mortar with a maximum aggregate size of 1.6 mm and a density of
2400 kg/m3 was used as the cement-based mortar to investigate the influence of
PCMs on the properties of cementitious materials.
2.2 Experimental methods
2.2.1 Preparation procedure of PCM-mortar matrix
Microencapsulated PCM was added in various amounts (0, 5, 10, and 15wt%) to
the reference mortar. Addition of a minimum of 5wt% PCM is necessary for it to have
a meaningful effect on the thermal performance of the composite [5].
The highly hydrophilic nature of the encapsulated PCM (which is attributed to its
high specific surface area (SPCM ≈1100 m2 /kg) and to the chemical adsorption of wa-
ter by the hydrogen bonds of the hydroxyl (-OH) and imino (–NH-) groups on the
polymer shell used for encapsulation) necessitates an adjustment of the water-to-
binder ratio for each composition of the PCM-mortar matrix [6] to meet the workabil-
ity requirement. The amount of additional water required to achieve a quasi-constant
slump for each PCM-mortar matrix
water not absorbed by PCM
(in accordance with an equivalent ratio) was evaluated by
cement
performing a water absorption test of 100 g of the PCM powder according to EN
1097-6; the test results are presented in Table 1.
 A non-paraffinic PCM modified textile reinforced … 455

Dried PCM powder Water saturated PCM Absorption coefficient

100 g 154 g 54 %
Table 1 Water absorption capacity of PCM

The compositions of the PCM-mortar matrix are presented Table 2. The mixing
procedure was performed such that PCMs were exposed to the mixing environment
for the shortest time possible in order to minimize any possible damage to the micro-
capsules.

PCM-mortars Reference 5wt% 10wt% 15wt%

PCM (kg) 0 0.3 0.6 0.9

Cement (kg) 1.65 1.65 1.65 1.65

Fine aggregate(kg) 4.35 4.35 4.35 4.35

Water (L) 0.80 0.96 1.12 1.29

w 0.48 0.58 0.68 0.78

c

water not absorbed by PCM 0.48 0.48 0.48 0.48

cement

slump (mm) 182 185 178 179

Table 2 PCM-mortar mixture compositions

2.2.2 Mechanical characterization of PCM-mortar

Compression tests were performed on the 0,5,10 and 15wt% PCM-mortar samples
according to EN 196-1 at a temperature of 20°C and an age of 28 days (five replicates
of each PCM-mortar matrix).

2.2.3 Mechanical characterization of PCM-TRC sandwich panels

Two sandwich panels have been realized. The reference sandwich is made of two
TRC skin (20 mm thick) which consist of the reference mortar (without PCM) rein-
forced with 2 layers of AR glass fabric and a polyurethane foam (50 mm thick). The
10% PCM-TRC sandwich panel is made of two PCM-TRC skins (20 mm thick)
456 Z. I. Djamai et al.

which consist of the 10% PCM-mortar matrix reinforced with 2 layers of AR glass
fabric and a polyurethane foam (50 mm thick).TRC and PCM-TRC skins have been
realized by the contact molding technique (manual stratification). This technique con-
sists in spreading a layer of concrete matrix in the mold, and then the textile rein-
forcement is placed and impregnated by the matrix with a trowel so that the matrix
penetrates inside the mesh of the reinforcement. The procedure is repeated so that the
composite is reinforced with the desired number of textile reinforcement layers (Fig-
ure 2).

The two sandwich panels have been tested under quasi-static four-point bending solic-
itation (0.2 mm/min). The panels span is equal to 1100 mm and the distance between
the loading cylinders and the supports is equal to 360 mm

Figure 2 Casting of TRC skins of TRC sandwich panels

3 Results and discussion

3.1 Results of the mechanical characterization of PCM-mortar matrix
Figure 3 shows the strength evolution of the 28-day-cured PCM-mortar samples in
the compression tests. It can be concluded that the addition of PCM induces a drastic
decrease of PCM-mortar matrix strength. This degradation increases with PCM con-
tent in the mortar. For PCM rates < 15%, mortar remains available for structural ap-
plications.

80
Compressive strength of composites (MPa)

60

-34,1%

40
-58,6%
70,3

-70,5%
46,28
20

29,07
20,7

0
0 5 10 15
PCM content (wt %)

Figure 3 Compressive strength of PCM-mortar matrix versus PCM content

A non-paraffinic PCM modified textile reinforced … 457

3.2 Results of the mechanical characterization of TRC sandwich panels

The results of the four-point bending test of the TRC sandwich panel (with and
without PCM) in terms of Force transmitted vs deflection is presented Figure 4

14000

Reference TRC sandwich panel

12000
10% PCM-TRC sandwich panel
Applied Force (N)

10000

8000

6000

4000

2000

Foam shear failure

0
0 10 20 30 40 50

Bending deflection (mm)

Figure 4 Sandwich panels Force vs bending deflection bending behaviour

The first observation that can be made by analyzing the curves (force vs deflection)
is a 39% degradation of the bending strength of the 10% PCM-TRC panel compared
to the reference TRC panel.

By analyzing in more detail, it can be seen that the overall behavior of the panels
(in both configurations) can be subdivided into three phases:
1-A first phase governed by the linear behavior of the components of sandwich
panels (facing + core) and the perfect interaction between them.
2-A second phase where there is a sudden drop in rigidity of the panels in both
configurations (at F = 5540 N for the reference TRC panel and at F = 2880N for the
10% PCM-TRC panel). The 51% decrease in force at the onset of the first crack in the
10% PCM-TRC panel is due to the integration of the PCM into the 10% PCM-TRC
facing. The decrease in flexural rigidity of the panels after the first crack in both cases
is due to the cracking of the lower TRC facing and load transmission from the cracked
matrixes towards the textile. The higher slope in the curve (Force vs deflection) of the
reference TRC panel comparing to the 10% PCM-TRC panel is due to the higher
intensity of textile/matrix interaction in the case of the reference panel.
3-A third phase where there is another drop in the flexural stiffness (slope inflec-
tion) in the two configurations of sandwich panels. This is due to foam hardening and
ultimately to its rupture by shearing. The same shear failure mode of the core was
found in both sandwich panel configurations.
Despite the decrease of the mechanical performance of the 10% PCM-TRC sand-
wich panel comparing to the reference TRC sandwich panel, it should be emphasized
that concrete/textile load transfer leading to multicracking in the lower PCM-TRC
facing is conserved which confirms the strain hardening behaviour (Figure 5). This
458 Z. I. Djamai et al.

result is encouraging in the objective of developing new TRC sandwich panels with
high thermal and mechanical performance.

Figure-5 Crack pattern in the lower PCM-TRC skin

4 Conclusion
This research presented a mechanical characterization of PCM modified cement
mortar where PCM is used for thermal inertia enhancement. It has been found that
the addition of PCM induces the degradation of the mechanical performance of Port-
land cement matrix, however for PCM rates < 15%, concrete remains available for
structural applications.
The 10% PCM modified mortar has been used as a matrix of a TRC facing of a 10%
PCM-TRC sandwich panel. The mechanical characterization of the panels has shown
a decrease of the mechanical performance of the 10% PCM-TRC sandwich panel
comparing to the reference TRC sandwich panel (a decrease of 39% of the load bear-
ing capacity has been noticed), however concrete/textile load transfer leading to mul-
ticracking in the lower PCM-TRC facing is conserved which confirms the strain hard-
ening behaviour (Figure 5). This result is encouraging in the objective of developing
new TRC sandwich panels with high thermal and mechanical performance.

5 References
[1] Si Larbi A, Contamine R, Ferrier.E, Hamelin P, Shear strengthening of RC beams with
textile reinforced (TRC) plate, Constr. Build. Mater. 24 (2010) 1928–1936.
[2] Hegger.J Will.N, Brckermann.O; Voss.S Load bearing behaviour and simulation of textile
reinforced concrete Material and structures 39(8) 765-776
[3] Djamai.Z.I, Bahrar.M, Salvatore F, Si-Larbi A, El-Mankibi.M Textile reinforced concrete
multiscale modeling: Application to TRC sandwich panels in Finite element in design and
analysis 2017
[4] Shams.A Hegger.J Horstmann.M An analytical model for sandwich panels made of textile
reinforced concrete Construction and building materials 2014;64p451-459
[5] Castellón C, Castell A, Medrano M, Martorell I, Cabeza L Experimental Study of PCM
Inclusion in Different Building Envelopes Journal of Solar Energy Engineering 131, 2009
[6] Djamai Z, Salvatore F, Si Larbi A, Gaochuang C, El Mankibi M Multiphysics analysis of
effects of phase change materials in cement mortas , Cement concrete research 119 (2019). 51-
63
 Design method for optimizing geopolymer concrete
proportions utilising entirely steel slag aggregates

Dong Van Dao1 and Son Hoang Trinh 2

University of Transport Technology, 54 Trieu Khuc, Thanh Xuan, Hanoi, Vietnam

1
dongdv@utt.edu.vn; 2sonht@utt.edu.vn

Abstract. Fly ash and steel slag, which are byproducts from thermal power and
steel plants respectively, have been used to manufacture geopolymer concrete
(GPC). When combined, fly ash-based GPC using steel slag aggregate (GPCs)
is effective in solving environmental concerns caused by the excess of by-
products. However, compared to Portland cement concrete (PCC), GPC is a
newer material, and its mix design method to target a specified compressive
strength has not been defined. To help make the GPCs mix design process more
efficient, this paper discusses a statistical model that can be used to estimate the
compressive strength of GPCs mixtures with 100 percent steel slag aggregates.
The model was developed based on an experimental design in which the three
variables were varied as follows: (1) the sodium solution (NaOH) concentration
was varied between 10 and 14M, (2) the mass ratio of alkaline activation solu-
tion to fly ash (AAS/FA) was changed from 0.4 to 0.5, and (3) and the mass ra-
tio of sodium silicate (Na2SiO3) to sodium hydroxide solution (SS/SH) was var-
ied from 2 to 3. Results obtained in this study show that the model can reasona-
bly estimate the compressive strength of GPCs mixtures based on the three var-
iables, so it is recommended for use in GPCs mix design in the future.

Keywords: Steel slag aggregate, fly ash-based geopolymer concrete, design of

experiments, mix design.

1 Introduction

Geopolymer concrete (GPC) is a mixture of an inorganic polymer binder and aggre-

gates. The aggregate materials in GPC are often consisted of crushed stones and natu-
ral sand [1]. The use of industrial by-products, such as steel slag aggregates, to re-
place natural aggregates in GPC will save more natural resources and reduce solid
waste [2].
As for PCC, a good performing GPC mixture starts with a good mix design [1, 3, 4].
A method proposed by Lloyd and Rangan can be followed to design fly ash-based
GPC mixtures meeting mechanical properties required in the fresh and hardened stag-
es [5-6]. Ferdous et al. [7] proposed a flow chart with nine steps to determine GPC
mixture proportions to yield a 28-day compressive strength of 45 MPa. These meth-
ods have been used to design GPC mixtures with the natural aggregates, but they have

© Springer Nature Singapore Pte Ltd. 2020 459

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_71
460 D. V. Dao and S. H. Trinh

not been used to design GPC mixtures with steel slag aggregates, which have a higher
density, porosity and water absorption.
This paper discusses a statistical model that can be used to select proportions of a
GPCs mixture to yield a targeted 28-day specified compressive strength between 25
and 35 MPa. The model was developed based on the Response Surface Method
(RSM) [8] and verified using laboratory test results.

2 Response Surface Model to determine GPCs proportions

2.1 Response and input parameters

A RSM, as shown in Equation 1, was utilized in this study. The response in this equa-
tion is the 28-day compressive strength of a GPCs as a function of three main effects,
including X1 (molar concentration of NaOH solution), X2 (mass ratio of AAS/FA),
and X3 (mass ratio of SS/SH).

a) START

Determning the target function and

input parameters

b)
Experimental procedure

Regression
analysis
Analysis of experimental results
Model fit
assessment
Determining the proportions of
GPCs mixtures from the prediction
model

No

Verifying test for designed

mixtures

Yes

END

Fig 1. (a) Design method procedure for GPCs by RSM; (b) Center Composite Design model
Design method for optimizing geopolymer … 461

Y = f (X1, X2, X3) (1)

The Center Composite Design (CCD) model of Face-centered Design (Quadratic
programming) was used with the distance in the experiment being chosen as α = 1.
The procedure for developing the RSM is shown in Figure 1(a). The CCD model is
illustrated in Figure 1(b).

2.2 Laboratory testing procedure

Three main effects as in Equations 1 and three levels for each main effect were ap-
plied. A total of 20 combinations were needed for this experiment. With six replicates
planned for each testing combination, a total of 120 cylinder samples (150 mm by 300
mm) were tested in this study. These cylinders were prepared, cured and then tested to
determine the 28-day compressive strength.

2.3 Analysis of experimental results

2.3.1. Regression analysis and goodness of fit assessment
Results of the compression strength test with all outliers rejected according to ASTM
E178 [9] are shown in Table 1. These results were then analyzed based on the CCD
model using the Minitab software (Version 17) [8].

Table 1. Experimental results of compressive strength tests

Code variables Results of compressive strength tests Average
No compressive
X1 X2 X3 Y1 Y2 Y3 Y4 Y5 Y6 strength
1 10 0.40 2.0 40.17 39.55 39.84 38.57 38.71 38.84 39.28
2 14 0.40 2.0 49.43 50.82 50.78 50.28 50.74 52.63 50.78
3 10 0.50 2.0 29.73 30.95 30.92 31.28 31.74 30.9 30.92
4 14 0.50 2.0 43.17 44.74 42.62 42.39 41.62 42.98 42.92
5 10 0.40 3.0 37.73 36.84 36.76 36.75 38.39 36.19 37.11
6 14 0.40 3.0 48.59 46.17 46.66 48.73 46.37 48.2 47.45
7 10 0.50 3.0 25.48 27.02 26.97 27.73 26.37 26.75 26.72
8 14 0.50 3.0 36.28 37.06 36.51 37.53 37.63 38.97 37.33
9 10 0.45 2.5 39.53 38.39 38.69 40.28 38.74 39.09 39.12
10 14 0.45 2.5 49.39 50.84 50.12 49.74 51.37 49.26 50.12
11 12 0.40 2.5 46.28 48.39 46.79 46.58 47.53 46.19 46.96
12 12 0.50 2.5 37.17 38.37 37.62 38.52 38.26 39.56 38.25
13 12 0.45 2.0 40.37 39.73 39.51 39.27 40.6 42.26 40.29
14 12 0.45 3.0 35.33 36.62 35.84 36.15 35.93 36.79 36.11
15 12 0.45 2.5 42.47 43.06 44.28 42.88 43.89 44.11 43.45
16 12 0.45 2.5 44.52 44.03 42.28 44.25 43.36 42.88 43.55
17 12 0.45 2.5 43.28 42.62 44.21 43.42 42.87 44.56 43.49
18 12 0.45 2.5 42.12 43.22 44.35 42.89 43.55 42.19 43.05
19 12 0.45 2.5 43.52 42.35 44.51 42.97 43.31 45.12 43.63
20 12 0.45 2.5 42.26 44.26 43.47 43.67 43.28 44.88 43.64
462 D. V. Dao and S. H. Trinh

2.3.2. Response surfaces

Figure 2 includes response surfaces and contour lines for the 28-day compressive
strength.

Fig 2. Response Surface graphs and counter lines of X1, X2, X3 variables and compressive
strength relationships.
These plots can help evaluate the impact of the main effects and their interactions on
the model response (i.e., the 28-day compressive strength), as follows.
 As the concentration of NaOH solution (X1) increases, the 28-day compressive
strength increases;
 As the AAS/FA ratio (X2) increases, the 28-day compressive strength decreases;
Design method for optimizing geopolymer … 463

 As the SS/SH ratio (X3) increases from 2.0 to 2.5, the 28-day compressive
strength increases significantly, but the 28-day compressive strength decreases
remarkably as the SS/SH ratio increases from 2.5 to 3.0.
2.4 Determining GPC proportions from the prediction model
For GPCs mixtures that can meet the required 28-day compressive strengths (f’c) of
25, 30, 35 MPa, they must yield the required average compressive strengths [10]
f cr'  33.3MPa; f cr'  38.3 MPa; f cr'  43.3MPa , respectively.

Using the regression model, the values for the three variables (i.e., X 1, X2, and X3) are
determined and then calculated to find out the proportions of each GPCs mixture. The
results are shown in Table 2.
Table 2. Mixture proportion of GPCs with f’c of 25, 30, 35 MPa
Mix 1 Mix 2 Mix 3
GPCS mixture
(f'c= 25 MPa) (f'c= 30 MPa) (f'c= 35 MPa)
Fly ash (kg) 482.34 491.89 507.29
Coarse aggregate (kg) 4.75-19 mm 1670.90 1670.90 1670.90
Fine aggregate (kg) 0.15-4.75 mm 716.10 716.10 716.10
NaOH solution (kg) 57.66 56.49 52.12
Na2SiO3 solution (kg) 172.99 164.62 153.59

2.5 Verification of the designed mixtures

Based on the design proportions for the three mixtures (Table 2), three GPCs mixtures
were prepared in the laboratory. 30 cylinders (150 mm by 300 mm) were prepared
and tested for each GPCs mixture to verify the predicted model and statitical analysis.
The results obtained and shown in Table 3 suggest the design mixtures would meet
the required 28-day compressive strengths. Thus, the regression model is verified for
this study.
Table 3. 28-day compressive strength and corresponding statistics for GPCs mixtures
Standard Coefficient Specified
GPCs No. of Mean
Parameters deviation of Varia- value
mixture samples (MPa)
S (MPa) tion (%) (MPa)
Mix 1 30 34.80 3.23 9.271 27.29
Compressive
Mix 2 30 39.97 3.20 8.012 32.50
strength
Mix 3 30 44.85 3.34 7.455 37.06

3 Conclusions and Recommendations

This paper describes the development of a statistical model that can be used to select
proportions of a GPCs mixture to yield a targeted 28-day specified compressive
strength between 25 and 35 MPa. Based on the results of this study, the following
conclusions and recommendations can be drawn:
464 D. V. Dao and S. H. Trinh

• The model was developed based on the RSM using laboratory test results obtained
from GPCs mixtures that were made by varying three mixture properties, including
molar concentration of NaOH solution, mass ratio of alkali activating solution to fly
ash, and mass ratio of Na2SiO3 to NaOH;
• The model was then verified using three mixtures that were designed using the
regression model to yield the required 28-day specified compressive strengths of 25,
30 and 35 MPa. Based on the laboratory test results of these three GPC mixtures, their
28-day compressive strengths can meet the required strength requirements, suggesting
the model is verified.
In summary, the regression model will allow a designer to quickly determine the pro-
portions of a fly ash-based GPC mixture that utilizes steel slag as both coarse and fine
aggregates. Thus, fly ash-based GPC mixtures with steel slag aggregates will be used
as a sustainable construction material for various structural applications.

References
1. Davidovits J., Geopolymer: Chemistry & Applications. Institut Geopolymere. 3 rd edition,
2011;
2. M.S.H. Khana et al, "Utilisation of steel furnace slag coarse aggregate in a low calci-
umflyash geopolymer concrete.", Cement and Concrete Research 89, pp 220–229;
3. ACI211-91 “Standard Practice for Selecting Proportions for Normal, Heavyweight, and
Mass Concrete”;
4. Tran Viet Hung, Dao Van Dong and Nguyen Ngoc Long (2016), "Mix design for low cal-
cium of fly ash base Geopolymer concrete", ACF2016 - The 7th Int. Conference of Asian
Concrete Federation “Sustainable concrete for now and the future”, Hanoi, Vietnam;
5. Lloyd NA, Rangan BV. Geopolymer concrete: A review of development and opportuni-
ties. 35th Conf. Our World Concr. Struct., 2010;
6. Rangan BV., Design and manufacture of fly ash-based geopolymer concrete. Conf. Aust
2008; 34:37–43;
7. Ferdous MW, Kayali O, Khennane A., A detailed procedure of mix design for fly ash
based geopolymer concrete. Fourth Asia-Pac. Conf. FRP Struct. APFIS 2013, 2013, p. 11–
13;
8. Paul G. Mathews (2005), Design of Experiments with MINITAB ASQ Quality Press Mil-
waukee, Wisconsin;
9. ASTM E178 (2016), "Standard Practice for Dealing With Outlying Observations";
10. ACI 318 (2011), "Building code requirements for structural concrete ".
 Experimental Measurement and Modeling of the
Complex Poisson’s ratio of Bituminous Mixtures

Quang Tuan Nguyen 1,2[0000-0001-5318-0381] and Thi Thanh Nhan Hoang 1,2
1
Faculty of Civil Engineering, University of Transport and Communications (UTC), Vietnam
2
Research and Application Center for Technology in Civil Engineering, UTC, Vietnam
quangtuan.nguyen@utc.edu.vn, ttnhan.hoang@utc.edu.vn

Abstract. A complete characterization of linear viscoelastic properties of bitu-

minous mixtures requires the measurements of the complex Poisson’s ratio. The
complex Poisson’s ratio of bituminous mixtures is not easy to obtain because of
the high precision required and the limited accuracy of the experimental device.
In this paper, three dimensional cyclic tension-compression tests were per-
formed to measure the complex Poisson’s ratio. Two bituminous mixtures were
tested: one prepared with classical bitumen and one prepared with polymer-
modified-bitumen. In order to measure the complex modulus and the complex
Poisson’s ratio, the sinusoidal loadings were applied on the specimens at differ-
ent temperatures and frequencies. The experimental results show that the com-
plex Poisson’s ratios of bituminous mixtures vary as a function of reduced fre-
quency. The master curve of the complex Poisson’s ratio can be built using the
same shift factor for the master curve of the complex modulus. It was verified
that the Time-Temperature Superposition Principle and the Partial Time-
Temperature Superposition Principle can be applicable for the complex Pois-
son’s ratio of the classical bituminous mixtures and the polymer-modified-
bitumen mixtures, respectively. A linear viscoelastic model with a continuum
spectrum called 2S2P1D is used to simulate the 3D linear viscoelastic behavior
of tested bituminous mixtures.

Keywords: Bituminous Mixtures, Linear Viscoelasticity, Complex Poisson’s

Ratio, Linear Viscoelastic Model, Complex Modulus.

1 Introduction

Bituminous mixtures commonly used in road pavement construction consist of the

aggregates, filler (particles no larger than 70 microns) and bitumen. The bitumen
content ranges between 10% and 12% by volume of mixes. In the bituminous mix-
tures, the bitumen binder is mainly responsible for the cohesion while the mineral
skeleton constituted by the aggregates ensures the rigidity of the mixes. The proper-
ties of bitumen are characterized by large viscous effects and great thermal sensitivity.
Within the small strain amplitude domain of loading (less than 100 µm/m), the behav-
ior of bituminous mixtures can be considered as linear viscoelastic (LVE) [1]. In other
words, the materials stiffness is independent of the amplitude of the applied loading.

© Springer Nature Singapore Pte Ltd. 2020 465

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_72
466 Q. T. Nguyen and T. T. N. Hoang

In order to characterize the linear viscoelastic behavior of bituminous mixtures and to

provide necessary properties of material in road pavement design, the complex modu-
lus E* of this material is generally measured and analyzed. The complex modulus of
bituminous mixtures depends on the temperature and the frequency of the loading.
Many researches show that the time-temperature superposition principle (TTSP) can
be applicable on the complex modulus of bituminous mixtures [2,3]. However, the
measurements of the complex Poisson’s ratio are required in order to complete the
characterization of 3D linear viscoelastic properties of bituminous mixtures. The
complex Poisson’s ratio of bituminous mixtures is not easy to obtain because of the
high precision required and the limited accuracy of the experimental device. It is often
considered as a fixed constant in the literature.
This paper focuses on experimental measurement and modeling of the complex
Poisson’s ratio of bituminous mixtures. In this paper, the complex Poisson’s ratio is
measured using a 3D cyclic tension-compression test. The test temperature and the
frequency of loading are changed to check the time-temperature dependence and the
time-temperature superposition principle for the complex Poisson’s ratio. In addition,
a 3D linear viscoelastic model is presented and used to simulate the 3D linear viscoe-
lastic behavior of tested bituminous mixtures.

2 Experimental program

2.1 Materials and specimens

Two bituminous mixtures, called M1 and M2, were used in this study. The grading
curves are presented in Fig. 1. The grading curves of both M1 and M2 are 0/14 mm
and continuous. Bituminous mixture M1 is dosed at 5.6 % (by weight of aggregates)
with EVA (ethylene-vinyl acetate) polymer modified bitumen. The tested specimen of
M1 has an air void content of 1.3 %. The bitumen content and air void content of M2
specimen are 4.5% and 2.3%, respectively. The bitumen used for mixture M2 is un-
modified bitumen of 35/50 penetration grade. The tested cylindrical specimens are 75
mm in diameter and 140 mm high.

100.00
M1
80.00
M2
Passing (%)

60.00

40.00

20.00

0.00
0.01 0.10 1.00 10.00 100.00
Sieve Size (mm)

Fig. 1. Grading curves of tested bituminous mixtures.

Experimental Measurement and Modeling … 467

2.2 3D complex modulus tests

Test equipment is used to apply sinusoidal loading in tension and compression

(through zero) on cylindrical specimen of asphalt mixture at different frequencies and
different temperatures. The sinusoidal tension and compression loading is applied on
the specimen put in the thermal chamber using a hydraulic press (Fig. 2a). Axial
strain is measured in the middle part of the specimen (Fig. 2b) using three extensome-
ters. They are located at 120° around the specimen with an initial length of 75 mm
and measure the vertical relative displacement. Axial strain is obtained from the aver-
age of these three displacements. During the whole experimental campaign, monitor-
ing is made from this averaged axial strain. Radial strain is measured using two non-
contact sensors. These two sensors are fixed oppositely on a sample diameter and aim
at two targets glued on the sample (see Fig. 2b). A temperature probe placed in the
thermal chamber is used to control the ambient temperature inside this chamber. An-
other temperature sensor (PT100 type) gives the temperature at the surface of the
sample during the test.

Fig. 2. a) Hydraulic press and thermal chamber, b) Measurement of axial strain, radial strain
and temperature on specimen

One 3D complex modulus test was performed on each bituminous mixture. The
temperatures and frequencies used in the tests are presented in Table 1.

Table 1. Temperatures and frequencies used in the tests.

Test Temperature Frequency

-8.4°C, 1.5°C, 11.3°C, 21.3°C, 26.5°C, 0.03Hz, 0.1Hz, 0.3Hz,
M1
30.9°C, 40.5°C, 49.9°C 1Hz, 3Hz, 10Hz
-25.8°C, -16.4°C, -7.0°C, 2.4°C, 11.6°C, 0.03Hz, 0.1Hz, 0.3Hz,
M2
16.2°C, 20.7°C, 29.7°C, 39.0°C 1Hz, 3Hz, 10Hz

Sinusoidal axial strain ( ) (average of three extensometers) is applied, sinusoidal

axial stress ( 1) is measured from load cell and sinusoidal radial strain ( 2) is deduced
from the two non-contact sensors: 1(t) = 01sin( t), 1(t) = 01sin( t+ E) and 2(t) = –
sin( t+ ), where E is the phase angle between the axial strain and the axial stress,
and is the phase angle between the axial strain and the opposite of radial strain. 01,
468 Q. T. Nguyen and T. T. N. Hoang

02 and 01 are respectively the axial strain amplitude, the radial strain amplitude and
the axial stress amplitude. Complex modulus E* and complex Poisson’s ratio are
obtained considering equations 1 and 2, where |E*| and | | are respectively their
norms and j is the complex number defined by j2= –1.

E* = 01
ej E
E* e j E
(1)
01

* 02
ej *
ej
(2)
01

3 2S2P1D model

The 3D linear viscoelastic model with a continuum spectrum, called 2S2P1D (two
Springs, two Parabolic elements, one Dashpot) consists of a generalization of the
Huet-Sayegh model [4] in 1D case. It was shown that the 2S2P1D model is powerful
to simulate the 1D and 3D LVE behavior of bituminous binders, mastics and mixes
over a very wide range of frequencies and temperatures [5,6]. Due to the limit of the
paper, the analogical representation and the equations of this model are not presented.
The readers can find more details of 2S2P1D model in [5].

4 Analysis of the test results

For the materials which exhibit a thermorheologically simple behavior, which means
that the TTSP holds, the complex modulus can be represented by the two master
curves (norm and phase angle). The complex modulus is plotted as a function of the
reduced frequency using the shifting procedure at a reference temperature Tref. An
example of building the |E*| master curve for M1 is given in the Fig. 3a.

Master curve aT 0.50 -8.4°C

1.5°C
4
a) 11.3°C
10 0.45 21.3°C
Tref = 11.3°C Tref = 11.3°C 26.5°C
-8.4°C 0.40
E (MPa)

1.5°C 30.9°C
M1 Test Master curve 40.5°C
11.3°C
0.35
*

3 49.9°C
10 21.3°C
*

26.5°C
30.9°C 0.30 b)
aT aT
40.5°C M1
49.9°C 0.25
2
10 -6 -5 -4 -3 -2 -1 0 1 2 3 4 5 -6 -5 -4 -3 -2 -1 0 1 2 3 4 5
10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10
Reduced frequency (Hz) Reduced frequency (Hz)

Fig. 3. Master curves of |E*| (a) and | *| (b) at Tref = 11.3°C (M1 test)

The Fig. 3b presents the results of complex Poisson’s ratio * measurements. The
isotherms of M1 indicated that the obtained complex Poisson’s ratio depends on the
tested temperature and frequency and is not a constant as generally admitted in litera-
ture. | *| varies approximately between 0.25 for high frequencies and/or low tempera-
 Experimental Measurement and Modeling … 469

tures and 0.5 for low frequencies and/or high temperatures. It is interesting in the Fig.
3b that master curve of the complex Poisson’s ratio | *| can be obtained by the same
procedure with E*.

50000 60
M1 b)
Tref = 11.3°C
50 M2
a)
5000 40 M2 - 2S2P1D model
|E*| (MPa)

M1 M1 - 2S2P1D model

(°)
30
M2

E
500 20
M2 - 2S2P1D model
M1 - 2S2P1D model 10 Tref = 11.3°C

50 0
1E-6 1E-3 1E+0 1E+3 1E+6 1E-6 1E-3 1E+0 1E+3 1E+6
Reduced frequency (Hz) Reduced frequency (Hz)

Fig. 4. Master curves of |E*| (a) and E (b) at Tref = 11.3°C for both M1 and M2

Fig. 4 and Fig. 5 present the master curves of |E*|, E, | *| and for both M1 and
M2. The simulation using the 2S2P1D model was also shown. It should be noted that
for each material, the same shift factors aT was used to build all four master curves of
|E*|, E, | *| and . It can be seen in Fig. 4 that using the polymer modified bitumen in
M1 enhances the |E*| and reduces the phase angle E at high temperature/low frequen-
cy. In Fig. 5b, is not nil contrarily to what is usually assumed in the literature. The
values of are quite small and negative regardless of some measurements at very
high temperature. The complex Poisson’s ratio of both tested mixtures is very close.
The simulation results show a good fitting for both E* and , which confirms the
good performance of 2S2P1D model.

0.55 6
a) M1 b)
0.5 4 M2
0.45 M2 - 2S2P1D model
2
Tref = 11.3°C
0.4 M1 - 2S2P1D model
(°)
| *|

M1 0
0.35
M2 -2
0.3
M2 - 2S2P1D model -4
0.25
M1 - 2S2P1D model Tref = 11.3°C
0.2 -6
1E-6 1E-3 1E+0 1E+3 1E+6 1E-6 1E-3 1E+0 1E+3 1E+6
Reduced frequency (Hz) Reduced frequency (Hz)

Fig. 5. Master curves of | *| (a) and (b) at Tref = 11.3°C for both M1 and M2

The obtained master curves for both E* and indicated that the TTSP is verified
and applicable in the 3D case. However, some dispersion can be observed on the E
master curve of the material M1. This remark is more clear when presenting the E*
and * of M1 in the Black diagram (Fig. 6). It can be seen in Fig. 6 that at high tem-
peratures, the isotherms do not form an unique curve but make some “commas”. This
result can be explained by the effect of using the polymer modified bitumen in the
mixture. Only one master curve obtained for |E*| (not for E) indicated that the M1
material respects the Partial Time-Temperature Superposition Principle (PTTSP).
470 Q. T. Nguyen and T. T. N. Hoang

Fig. 6. Experimental results of E* (a) and *

(b) in Black diagram (M1 test)

5 Conclusions

From experimental tests and analysis, the following conclusions can be drawn:
─ The values of E* and * for two tested bituminous mixtures were measured ex-
perimentally. These values were shown to be time-temperature dependence.
─ Master curves were built for E* and * using the same shift factor aT. The TTSP
is verified and shown to be applicable in the 3D case.
─ Using polymer modified bitumen enhances the E* at high temperature and makes
the M1 mixtures respect only the PTTSP.
─ The 3D model 2S2P1D shows a good simulation for both E* and *.

Funding

This work was funded by Vietnam National Foundation for Science and Technolo-
gy Development (NAFOSTED) under Grant number 107.02-2018.305.

References
1. Airey G. D. and Behzad R.: Combined bituminous binder and mixture linear rheological
properties. Construction and Building Materials, Vol. 18, 535-548 (2004).
2. Nguyen Q.T., Di Benedetto H., Sauzéat C, Nguyen M.L., Hoang T.T.N.: 3D complex
modulus tests on bituminous mixture with sinusoidal loadings in tension and/or compres-
sion. Materials and Structures 50 (1) (2017).
3. Nguyen Q.T., Di Benedetto H., Sauzéat C., Tapsoba N.: Time Temperature Superposition
Principle Validation for Bituminous Mixes in the Linear and Nonlinear Domains. Journal
of Materials in Civil Engineering 25 (9), 1181–1188 (2013).
4. Sayegh, G.: Contribution à l’étude des propriétés viscoelastiques des bitumes purs et des
bétons bitumineux. Ph.D. Thesis, Université de Paris (France) (1965).
5. Di Benedetto H., Delaporte B., Sauzéat C.: Three-dimensional linear behavior of bitumi-
nous materials: experiments and modeling. International Journal of Geomechanics 7(2),
149–157 (2007).
6. Nguyen, Q.T., Di Benedetto, H. and Sauzéat, C.: Linear and nonlinear viscoelastic behav-
iour of bituminous mixtures. Materials and Structures 48(7), 2339-2351 (2015).
 Evaluation of Macro-Synthetic Fibre Reinforced
Concrete as a Sustainable Alternative for Railway
Sleepers

Christophe Camille1[0000-0002-5768-4509], Dayani Kahagala Hewage1[0000-0001-5952-9847],

Olivia Mirza1, Fidelis Mashiri1, Brendan Kirkland1 and Todd Clarke2
1
Western Sydney University, Locked Bag 1797, Penrith 2751, New South Wales, Australia
2
BarChip Pty Ltd, Sydney, New South Wales, Australia
c.camille@westernsydney.edu.au

Abstract. Sleepers are among the most essential components within the track
infrastructure and are characteristically made from timber, steel and prestressed
concrete. However, due to recent concerns in regard to the inferior quality, deg-
radation, durability, high-cost and environmental impact of the conventional ma-
terials, researchers are now focused towards sustainable alternatives such as com-
posite materials. Correspondingly, even if these new sleepers provide the ade-
quate strength characteristics towards the redistribution of high static and dy-
namic loads, their practical implementation remained fairly limited due to their
unknown long-term behaviour and high production cost. This paper presents a
review of the Macro-Synthetic Fibre Reinforced Concrete (MSFRC) sleepers
which uses BarChip fibres towards the optimisation of the mechanical properties.
In addition, the effects of different fibre dosages and aspect ratios will be thor-
oughly assessed. Further, the sustainability aspects of the MSFRC sleepers will
be discussed with respect to durability and resource minimisation. As a result,
the implementation of the recycled fibres within the concrete sections is expected
to reduce the overall carbon-footprint of the sleeper towards a characteristically
adequate and sustainable alternative material for railway sleeper applications.

Keywords: Synthetic fibre reinforcement, Railway sleepers, Sustainable perfor-

mance

1 Introduction

Typically known as transoms (i.e. bridge ties), sleepers are the transverse components
beneath the rails which critical functions are to maintain the track geometry while re-
distributing the wheel loads (i.e. vertical, lateral & longitudinal) at allowable stresses
onto the supporting ballast or girders [1, 2]. Nevertheless, with the increase in speed
and axle loads of trains as well as the unavailability of high-quality timber throughout
the years, engineers implemented the use of steel, concrete and polymer sleepers which
offered superior mechanical behaviour [3]. However, these sleepers are usually suscep-
tible to premature failures resulting from material degradation, corrosion and cracks
which considerably increase the maintenance frequency and cost. In recent years, nu-
merous developments in the field of railway sleepers have emerged, highlighting the

© Springer Nature Singapore Pte Ltd. 2020 471

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_73
472 C. Camille et al.

sustainable characteristics of composites material as an alternative to traditional timber,

steel and prestressed concrete sleepers. Indeed, these composite technologies could be
engineered with superior mechanical properties to further reduce the disposal and
maintenance cycles of the railway track components to a minimum. Nowadays, modern
composite sleepers predominantly incorporate fibre reinforcement aimed towards the
partial or complete substitution of the steel prestressed cables to further improve the
design efficiency in terms of weight, adaptability, mechanical properties, corrosion re-
sistance and sustainability.
Therefore, in an effort to continuously provide an innovative solution, the study
herein presented will focus on the reinforcing capabilities of macro-synthetic fibres (i.e.
BarChip) integrated in concrete sleepers towards efficiently reducing the cost while
optimising the strength, failure mechanisms and sustainability. The BarChip fibre cho-
sen possesses most of the critical characteristics for sleeper applications which included
good durability, strength (i.e. residual), corrosion resistance, environmentally friendly
and light weight.

2 Methodology

2.1 Concrete Mix Design

The high strength concrete mix herein presented was designed and further adjusted to
achieve a characteristic class strength of at least 50 MPa with a water-to-cement (w/c)
ratio of approximately 0.39. These limitations ensure that all the batches met the mini-
mum prestressed concrete sleepers’ strength, durability and workability requirements
as specified in the relevant Australian Standards [4]. Table 1 below summarises the
concrete mix design (i.e. M50) implemented towards the investigation of the mechani-
cal characteristics of macro-synthetic fibre reinforced concrete.

Table 1. Details of the Design Mix as per the M50 Class Strength

No. Component Unit Dosage

1 Builders Cement (30% fly ash) 445.5
2 Blue Metal – 10mm Coarse aggregate 869.8
3 Nepean Paving Sand (Coarse) 510.0
kg/m3
4 Newcastle Sand (Fine) 350.0
5 Total Water 175.2
6 High-Range Water Reducer (HRWR) 4.4 – 5.3
7 Water-to-Cement ratio (w/c) ~ 0.39

2.2 BarChip Fibres

The BarChip macro-synthetic fibre concrete reinforcement are predominately made
from a high performance polypropylene base material, providing structural reinforce-
ment (i.e. class II) in concrete, mortar and grout [5]. The macro-synthetic fibres herein
investigated are the BarChip 48 and BarChip MQ58 (Fig. 1) which compliance with
Evaluation of Macro-Synthetic Fibre Reinforced … 473

EN 14889-2 have been casted at six different fibre volume ratios ranging from 0.0%
(control) to 2.0%. The use of fibres is expected to increase the durability of the concrete
while concurrently reducing the carbon-dioxide (CO2) emissions. Table 2 below sum-
marises the characteristics of the BarChip synthetic fibres herein studied with their as-
sociated structural applications.
Table 2. Characteristics of the BarChip Fibres [5]

Tensile Young’s
Length Applica-
Fibre Base Material Strength Modulus
(mm) tions
(MPa) (GPa)
Virgin Poly- Industrial
BarChip 48 48 12
propylene floors,
640
BarChip Bi-Component Pavement,
58 10
MQ58 Polymer Precast

(a) (b)

Fig. 1. (a) BarChip 48, (b) BarChip MQ58

3 Experimental Results

The results generated from the experimental tests herein presented, summarise the fun-
damental mechanical behaviour of the MSFRC in terms of indirect tensile behaviour,
residual (post-cracking) flexural strength and carbon footprint reduction towards the
practical application of this innovative sleeper material.

3.1 Indirect Tensile Strength

The tensile behaviour of concrete reinforced with macro-synthetic fibres (i.e. BarChip
48 & BarChip MQ58) at different volume fraction is critical towards the understanding
of crack initiation and propagation in sleeper applications as illustrated in Fig. 2.
474 C. Camille et al.

Indirect Tensile Strength (MPa) 12

10

2
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0 2.2
Fibre Volume Ratio (%)

Barchip 48 Barchip MQ58

Fig. 2. Indirect tensile strength vs. Fibre volume ratio

The comparative results suggest that both types of fibre have a tendency to gradually
improve the ultimate tensile capacity of concrete as the fibre volume ratio is increased.
For instance, MSFRC embedded with BarChip 48 and BarChip MQ58 were both ob-
served to enhance the tensile strength by up to 83% and 54% respectively as the fibre
volume fraction was increased from a range of 0.2% to 2.0%.
Experimentally, the tensile strength (ft) of plain concrete is observed to vary between
10 – 15% of the characteristic compressive strength (fc). Comparatively, Fig. 3 high-
lights this particular tensile-to-compressive strength ratio wherein which the interaction
of the concrete matrix and the macro-synthetic fibres are individually acknowledged.
0.22
Tensile-to-Compresive strength

0.18

0.14
ratio (ft / fc)

0.10

0.06
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0 2.2
Fibre Volume Ratio (%)

BarChip 48 BarChip MQ58

Fig. 3. Tensile-to-compressive strength ratio for BarChip 48 & BarChip MQ58

From the range of 0.0% – 1.0% of fibres, both BarChip 48 and BarChip MQ58 spec-
imens performed within the predictable tensile strength limit wherein both the fibres
and the concrete matrix equally contributed to the strength growth. However, as the
fibre dosage is increased above 1.0%, the tensile strength of the MSFRC is observed to
predominantly increase due to the extra amount of fibres which bridging interaction
strengthen the concrete and reduces the propagation of coalesced micro-cracks.
Evaluation of Macro-Synthetic Fibre Reinforced … 475

3.2 Residual Flexural Strength

The flexural behaviour herein presented fundamentally identifies the post-cracking
characteristics of the concrete with increasing macro-synthetic fibre dosage. Indeed, the
post-cracking behaviour of MSFRC is observed to significantly increase as the fibre
volume ratio increases up to 2.0%. Overall, both types of fibres exhibited a positive
effect on the residual flexural behaviour of the concrete as summarised in Fig. 4 below.
For instance, BarChip 48 provided the greatest increase in residual strength at 2.0%
(i.e. 7.88MPa) as compared to BarChip MQ58 which peaked at 1.5% with a post-crack-
ing strength of 7.85MPa. This post-cracking behaviour of the MSFRC is justified by
the mechanics of crack formation and propagation wherein which the fibres interact
with the concrete matrix to stabilise and suppress crack growth [6].
10
Peak residual flexural strength (MPa)

0
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0 2.2
Fibre Volume Ratio (%)

BarChip 48 BarChip MQ58

Fig. 4. Post-cracking flexural strength comparison for BarChip 48 & BarChip MQ58

3.3 Carbon Footprint Reduction

Since the most traditional sleepers are designed using the permissible-stress (i.e. allow-
able-stress) design concept related to the elastic moment capacity, a reduction in the
cross-sectional area of the sleeper can be evaluated using the limit of proportionality.
Accordingly, MSFRC embedded with 0.6% of fibres (i.e. BarChip 48 & BarChip
MQ58) generated the highest flexural strength permitting a reduction in the cross-sec-
tion towards an identical structural capacity as compared to traditional concrete sleeper.
Indeed, standard size sleeper reinforced with macro-synthetic fibres enables a max-
imum cross-sectional area reduction of approximately 6% and 4% for BarChip 48 and
BarChip MQ58 respectively. As a result, the use of macro-synthetic fibre reinforcement
in concrete empowers a sustainable development in railway infrastructure towards re-
ducing 2.1 kg of CO2 emission per sleeper equivalent to around 7.1 tonnes per km of
double track as represented in Fig. 5.
476 C. Camille et al.

Concrete (with 30% fly

ash)

Concrete (with 30% fly

ash & BarChip Fibres)

108 110 112 114 116 118 120 122

Carbon emission (tonne / km of double track)

Fig. 5. Carbon emission of MSFRC sleepers

4 Conclusion

This paper discusses three key impacts of macro-synthetic fibre reinforcement on the
mechanical properties of concrete. An accurate experimental program has been devel-
oped in accordance with the relevant international standards to investigate the indirect
tensile strength, residual flexural behaviours and carbon emission of the MSFRC from
which the following conclusions were achieved:

• Indirect tensile strength – The indirect tensile strength results were observed to in-
crease with the fibre dosage up to 2.0% characterising the bridging interaction of the
fibres which strengthen the macro-synthetic fibre reinforced concrete;
• Flexural behaviour – The addition of fibres was observed to have a negligible impact
on the ultimate flexural strength (i.e. pre-cracking) of the specimens. However, the
fibres reinforcement improved the fracture mechanisms towards a more ductile be-
haviour, characteristically reducing the loss in capacity sustain after the initial
cracks. Overall, higher fibre dosages showed better performance in terms of residual
flexural strength (i.e. post-cracking), ductility and toughness;
• Carbon emission – The implementation of MSFRC enables a reduction in the cross-
section of the sleeper, thus minimising the quantities of concrete and its associated
carbon footprint towards a more environmentally friendly sleeper.

References
1. Kumar, D.K. and K. Sambasivarao, Static and Dynamic Analysis of Railway Track Sleeper.
International Journal of Engineering Research and General Science, 2014. 2(6).
2. Tzanakakis, K., The Railway Track and Its Long Term Behaviour. 1 ed. Springer Tracts
on Transportation and Traffic. 2013: Springer-Verlag Berlin Heidelberg.
3. Sadeghi, J. and P. Barati, Comparisons of the mechanical properties of timber, steel and
concrete sleepers. Structure and Infrastructure Engineering, 2012. 8(12): p. 1151-1159.
4. Standard Australia Online, Railway Track Materials - Part 14: Prestressed concrete
sleepers, AS1085:14:2012. 2012: SAI Global database.
5. BarChip. 2018 20 October 2018]; Available from: https://barchip.com/product/.
6. Amin, A., et al., Material characterisation of macro synthetic fibre reinforced concrete.
Cement and Concrete Composites, 2017. 84: p. 124-133.
 The repairing effects of lithium silicate based material to
the surface of hardened concrete

Nguyen Xuan Tung[1]; Ryoma Kitagaki[2] and Yoshiro Yamakita[3]

1
Civil Engineering Department, University of Transport and Communications
ngxuantung@utc.edu.vn
2
Graduate school of Engineering, Hokkaido University, Sapporo, Japan
ryoma@eng.hokudai.ac.jp
3
Bebrain Co,.LTD, Hokkaido, Japan
yamakita@bebrain1.com

Abstract. There are many damage types on concrete structures. Crack and spall-
ing on concrete structures are the most popular damages. The causes of these
damages are variable but mostly originate from the corrosion of reinforcement
that cause by the salty, humidity air, carbonation or ion chloride penetration via
appeared cracks on the concrete surfaces that lead to the expansion of the rein-
forcement cross section. It is necessary to protect the surface area of concrete in
order to prevent concrete structures from being deteriorated. The method of ap-
plying a surface protective agent has been developed as one of the methods to
protect concrete. In this study, the repairing effects of lithium silicate based sur-
face protective agent in high permeability and chloride resistance will be evalu-
ate. As a result, reduction of coefficient of water permeability and effective dif-
fusion coefficient of chloride ion of hardened concrete that had been repaired by
the lithium silicate based surface protective agent were confirmed.

Keywords: Lithium silicate Surface protective agent, Coefficient of water per-

meability, Chloride ion diffusion coefficient

1 Introduction

Concrete is a relatively inexpensive material and widely used in construction. However,

due to its mechanism and properties itself resulting to cracks, penetration of ion chloride
and water…, therefore the service life of concrete element can decrease. There are many
conventional repairing materials that were researched to deal with those effects such as
sodium silicate, tetraethyl orthosilicate (TEOS)….Sodium silicate is a healing agent
that introduced in to concrete while casting to seal and potentially heal microcracks.
The tetraethyl orthosilicate (TEOS) is Organic surface treatment agents that were used
to improve the surface quality of cement-based materials but one of their defects is
lower weather resistivity compared to the inorganic one [1]. In fact, the concrete mate-
rial mostly face problems during its service life under the effect of environment and
those two agents mentioned above are not capable of. Lithium Silicate (Hereinafter
referred as LS), is the product of the Bebrain Company at Hokkaido Japan, has the

© Springer Nature Singapore Pte Ltd. 2020 477

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_74
478 N. X. Tung et al.

chemicals named as Amorphous Silica and Alkali metal with the components as in the
Table 1:

Table 1: Content of Lithium Silicate

Content (wt %) Chemical Formula CAS No
Amorphous Silica 8.47 – 9.36 SiO2 7631-86-9
Alkali metal 3.6 – 4.15 (As Na2O) As Na2O 1313-59-3
The Lithium Silicate densifies and hardens the surface of concrete, while also dis-
persing and refining the micro cracks that can occur in young concrete. Reaction prod-
ucts seal the concrete interior, prevent the penetration of water, chlorides, and other
deleterious influences, and improve the durability of the concrete. This completely in-
organic product helps extend the longevity of structures while not bringing harm to
people or pollution to the environment.
Current silicate formulations on the market typically contain sodium and potassium.
While offering protection from the penetration of water, chlorides, etc., sodium and
potassium are substances that cause alkali‐aggregate reactions deleterious to the quality
of concrete, which means that they can promote the aging of concrete. Lithium Silicate,
however, uses lithium. This is a substance recognized for suppressing the alkali‐aggre-
gate reaction, though it has been criticized for not penetrating deeply into concrete.
R&D on Inner Protect has addressed this drawback and has enabled the product to pen-
etrate 40 mm deep. As a new, degradation prevention product, Lithium Silicate densi-
fies concrete and gives it the durability for long life.
In this paper, the effects of Lithium Silicate to the salty, water and chloride penetra-
tion to the surface of hardened concrete will be clarified via various tests and the test
results of LS coated and non-coated specimens will show the advantages of LS based
material to the surface of hardened concrete.

2 Lithium Silicate effects to surface of hardened concrete

2.1 Salty penetration effects

Used material and specimens

The materials used are shown in Table 2, and the concrete composition are shown in
Table 3. The specimens are the column of 10cm diameter and 20cm height. The speci-
men was cured up to 14 days of age and then a the modifying agent LS was applied on
the surface of concrete. After that, concrete was kept at wet condition for 14 days.
Besides, there was a commercially siliceous-based modifier (hereinafter referred to
as SC) will be used as surface agent for comparison with LS. Regards to the concrete
composition and surface modifying agents, the coating method for both LS and SC
were applied to the concrete surface with the amount of 0.15l/m2 then cured in sprayed
water for 90minutes after that the 2nd coating with amount of 0.1l/m2 and sprayed water
curing will be applied.
The repairing effects of lithium silicate … 479

Test method
The salty penetration effects of the surface modifier were identified in accordance
with JSCE-G 571-2003 standards. The 5cm_thickness specimen was cut out from the
middle of the sprayed water cured concrete column that had been applied the surface
modifier agents as mentioned in the previous paragraph

Table 2: Materials used

Classification A symbol Types and physical properties

OPC Ordinary cement: d* = 3.16; B* = 3320

Cement
BB Blast furnace cement: d = 3.04; B = 4760

Admixture Lignin sulfonate type AE water reducing agent

Sanitary crushed sand: 2.76 g/cm3 dry table, water absorption rate
Fine aggregate
1.49%
Coarse aggregate Static quartzite: surface dry 2.77 g/cm3, water absorption 0.98%

* D: density g/cm3, B: specific surface area cm2/g

Table 3: Concrete Composition

Type W/C Fine Unit amount (kg/m3) Admixture
(%) aggregate Water Cement Fine Coarse Addition
ratio (%) aggregate aggregate amount
(kg/m3)
OPC50 50 51 176 352 896 941 3.52
BB50 50 51 173 346 899 941 3.46
BB60 60 51 174 290 959 927 2.90

Results and discussion

The influence of surface modifying agents (SC and LS) on salty penetration is presented
by effective diffusion coefficients of chloride ion are shown in Fig. 1. The numerical
values in parentheses in the figure indicate the ratio R of the effective diffusion coeffi-
cient between the modifier-coated and non-coated specimens. In the case of OPC 50
with the composition that shown in Table 3, the effective diffusion coefficient of spec-
imen coated with SC was R = 79%, slightly less as compared with the uncoated spec-
imen. The effective diffusion coefficient of the surface modifier containing lithium sil-
icate coated specimen drastically decreased, R = 20%. According to the Concrete
Standards Presentation of Japan Society of Civil Engineers (Construction Edition), the
diffusion coefficient ratios of W/C = 40% and W/C = 30% concrete with respect to W/
C = 50% are 43% and 15%, respectively. In this experiment, the diffusion coefficient
ratio of OPC 50 concrete with LS coating and non-coating is 33%, which is very high
as salty penetration resistance effect.
On the other hand, the BB 50 concrete with the composition that shown in Table 3,
showed a high salty penetration resistance effect, with the diffusion coefficient is 1/10
less of OPC 50 in case of non-coated. However, when surface modifiers were applied,
480 N. X. Tung et al.

especially LS, the effective diffusion coefficient tended to decrease and it was recog-
nized that the salty penetration resistance was further improved.
Effective diffusion coefficient (cm2/s ×10-8)

Effective diffusion coefficient (cm2/s ×10-8)

OPC50 BB50

Fig. 1. Effective diffusion coefficients

2.2 Water penetration effects

Used material
The surface modifier LS used in a mixture for experiment with physical properties such
as specific gravity of 1.1, viscosity of 2.0 mPa/s and pH of 12.7. As for specimen, or-
dinary cement mixture with water-cement ratio of 0.5 and a sand-cement ratio of 2.0
was used.
Test specimen
The test specimen is in a 100mmmm diameter and 50cm height column. The specimen
was attached by steel wire fixed at equal intervals and electrode system that shown in
Fig. 2. The LS modifier was applied on the surface of specimen after 14 days of water
curing and then it was cured under air conditions until the test began. The non-coated
specimens were also subjected to the same curing process, and the testing age was set
at 28, 91, and 180 days.
a: 2.2mm; d: 10mm;
l: 34mm
d a

l Depth from the surface

10 mm
20 mm
30 mm
40 mm
Fixing material
Fig. 2 Electrode arrangement testing setting method
The repairing effects of lithium silicate … 481

Test method
Normally, determination of depth water penetration is usually done by visual inspection
with splitting method. In this test, the measurement method using padded electrodes by
Keihira [4] was applied. In this method, time and depth penetration were measured to
determine the permeability coefficient.
Measuring the inter-electrode resistance values by time was determined specific re-
sistance using the equation (1). Changes in specific resistance indicate the changes in
the moisture content at the electrode positions and also indicate the diffusion time of
water from the surface to the electrode position. Besides, the time of sharp decline at
electrode ends (hereinafter referred to as Te), where the saturation reached, equivalent
to the so-called penetration depth, where the water reached stable state. Therefore, the
permeability coefficient for each depth were obtained by using the penetration depth,
dm, in the equation (2) and Te was used to calculate the permeability coefficient at each
depth.
log⁡(𝑑/𝑎)
𝑅= 𝜌⁡⁡⁡⁡⁡⁡⁡⁡⁡⁡⁡⁡⁡⁡⁡⁡⁡⁡⁡⁡⁡⁡(1)
𝜋_1

Where, R: Resistance value (Ω), d: Electrode spacing (m), a: Radius of electrode

(m), l: Length of conduction part (m), ρ: Specific resistance (Ωm)’

Where: P: pressure (MPa); t: time (s); dm: depth of penetration (m); k: permeability
coefficient (m/s); ω0: unit volume of water (9.8 N/m3)
Results and discussion
Fig. 3 shows the change in 1E+06
NO
specific resistance in the permeability test
conducted at 91 days of age. The arrow ↑
in the figure indicates Te at the depth and
1E+05
shown in the square. Te was defined as the
6SHFLILFUHVLVWDQFHȍP

inflection point of the specific resistance

change rate. 10
20 30
The permeability coefficient for each 1E+04 40
depth obtained from equation (2) is shown 1E+06
LS 10mm 20mm
in Fig.4, with age of 91, 180 days and the
30mm 40mm
depth of LS coating 40 mm.
In the case of comparing 91 days and 1E+05
180 days of age, the permeability coeffi-
cient reduced for non-coated specimens. 10
However, the permeability coefficient of 20 30
1E+04
the LS coated specimen also reduced. 0 20 40 60 80 100 120 140
In measurement at 28 days of age, there ඥ‫ݐ‬௦ (mm)
was no change in the resistance value ob- Fig. 3. Specific resistance change in wa-
served except for the depth of 10 mm, and ter permeability test (91 days of age)
only the permeability coefficient at the
482 N. X. Tung et al.

depth of 10 mm was obtained. However,

Permeability coefficient K
the obtained permeability coefficient is
smaller than at age of 91 and 180 days for
both non-coated and LS-coated specimens
From the above results, it is confirmed
that the water permeability coefficient of
LS-coated specimen at any age and depth
was smaller than those without coating,
and improvement of water penetration re-
sistance of LS coating is also confirmed. It
seems that LS coated on the surface per- Depth (mm)
meates into the interior and densifies the Fig. 4. Permeability coefficient change
voids such as pores resulting the reduction in the depth direction
of the permeability coefficient.

3 Conclusions

The results obtained in this study are summarized as follows:

- In LS coating, salty penetration resistance was improved regardless of cement
type and water cement ratio.
- LS applied to the surface of the cement hardened body penetrates into the inside
of the cured body and improve salty penetration resistance
- The water permeability coefficient reduced to a depth of 40 mm by using the
surface modifier LS and t its effect increased with age of mortar.
- By installing the LS on the whole concrete elements, their surface properties
will be improved resulting the extension the service life of concrete structures
and reduce maintenance cost.

References
1. Guo Z., Hou P., Huang S., Ning X.,Xin C., Singh L.P, Strokova V., Nelyubova V.: Surface
treatment of concrete with tetraethyl orthosilicate, Na2Si03 and silane: comparision of their
effects on durability, Journal of Ceramics – Silikaty 62, (4)332 – 341 (2018)
2. Torii Kazuyuki: Diffusion coefficient of chloride ion in cement hardened body obtained by
electrical acceleration test method, Cement – Concrete, Journal No. 604, pp. 18-22, 1997.
3. Naganuma Y.: Study on penetration effect of concrete surface modifying agent using lithium
silicate, Proceedings of the Society of Resources and Materials, Spring Meeting of the So-
ciety of Resources and Materials, Material Collection, pp. 205-206, 2005 (in Japanese).
4. MIZUKI Suzuki et al .: Influence of highly permeable surface modifier on permeability and
durability of cemented cement, Conte Engineering Annual Meeting, Vol. 27, No. 1, pp. 829
- 835 , 2005 (in Japanese).
5. Katsura, Yoshino Toshiyuki, Tabata Masayuki, Kamada Eiji: Theoretical verification on
water content measurement in concrete by an alternating two-electrode method, Concrete
Engineering Annual Report, Vol. 16, No. 1, 1994 (in Japanese)
 Research on fabricating flowforms from ultra-high
performance concrete with local admixtures for use in
processing contaminated water

∗1 ∗1 1
Thi Thuy Ha Ung , Tuan Hung Pham , Tho Bach Leu

1
National University of Civil Engineering (NUCE), Ha noi, Vietnam
* uttha2002@gmail.com; hungpt@nuce.edu.vn

Abstract. In the recent years, flowforms have emerged as an effective natural

aeration solution to provide dissolved oxygen (DO) for treatment processes of
contaminated water. Flowforms are usually made from materials such as: cement,
concrete, ceramic, glass, composite stone, metals... Among which, concrete and
ceramic are most common due to the low production cost of the former and rela-
tively good durability of the later. However, plain concrete has some disad-
vantages such as: heavy weight, difficult maintenance and prone to corrosion of
the surface, i.e. peeling of fine aggregates on the surface resulting in loss of
smoothness needed, while ceramic surface can develop cracks over time. Recent
studies on ultra-high performance concrete (UHPC) with admixtures available in
Vietnam such as silica fume and fly ash showed that, with high workability, this
material would satisfy well the requirements for fabrication of flowforms in terms
of smooth surface, high mechanical durability, corrosive resistance of finish
product. The research team has successfully fabricated flowforms prototype from
UHPC and tested it in practice. The test results on waste water in two years
showed that UHPC has a distinct advantage over the most common materials in
flowforms fabrication: it preserves the smooth surface, unlike the etched surface
of cement or cracked surface of glass enameled ceramic (tested in same condi-
tion). In addition, flowforms made from UHPC also has the advantage in produc-
tion cost, compared to ceramic and a good potential in reuse of industrial wastes.
This article will cover the process of fabricating flowforms from UHPC with sil-
ica fume and fly ash admixtures as well as the results of testing flowforms proto-
type in waste water treatment.

Keywords: Water treatment, ultra-high performance concrete, UHPC, fly ash,

silica fume, flowforms, aeration, dissolved oxygen, DO.

1 Introduction

With the main function of providing oxygen for aerobic processes, aeration is one of
the most important steps wastewater treatment. Aeration solutions are usually classified
into two groups: natural and artificial (forced). In recent years, one of the natural aera-
tion solutions called flowforms has gained increasing interest due to the development

© Springer Nature Singapore Pte Ltd. 2020 483

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_75
484 T. T. H. Ung et al.

trend of environmentally friendly pollution treatment systems. These devices are simple
to manufacture, easy to operate and maintain with low energy consumption, and can
blend into the natural landscape [1].
Flowform is a vessel of many different shapes, which can be like a bowl or a pot,
which creates a vortex flow that allows the water flow to regenerate oxygen from the
air, helping enhance water purification to bring water back to its natural state [2].
The production of flowforms is found with a variety of materials. Each type of ma-
terial has certain influences on the nature, pattern and rhythm of the water flow [3].
However, they must possess following properties: smoothness of finish surface; dura-
bility in natural climatic conditions; corrosion and chemical resistance; crack free dur-
ing operation. There are numerous materials which can meet to different extents the
above requirements, namely: cement, concrete, ceramic, enameled ceramic, thermo-
plastic resin, glass, composite stone, metal, ... Along with those materials, ultra-high
performance concrete (UHPC) has received increasing interest during recent years.
This kind of material has good climate-resistance, non-shrinking properties and high
smoothness of surface [4] as well as easy workability and very high strength (usually
over 150MPa). Comparing to popular materials used for production of flowforms
UHPC has obvious advantages in robustness, surface smoothness and corrosion re-
sistance. In addition, flowforms made from UHPC also has the advantage in production
cost, compared to enameled ceramic and potential in reuse of industrial wastes. This
study will cover the process of fabricating flowforms from UHPC with silica fume and
fly ash admixtures as well as the results of testing flowforms prototype in waste water
treatment.

2 Basics of flowforms

Flowforms are set up in a cascade, through which water flows, from one to the next.
The flowforms' shape coaxes the water (which would otherwise flow along the axis of
the cascade by gravity) into eddies and vortices, a pulsating figure of eight pattern in
each form before it drops into the next a mimicking of the action water undergoes in its
natural undisturbed flow in, say, a stream, as opposed to captive water in typical chan-
nels or pipes.
Flowforms take many sizes and shapes, with certain common characteristics: a
smaller entry "bowl" which receives water from the
source or the flowform above, from which water
flows into two larger basins where the figure of “8”
movement takes place, then an outlet through
which the flow drops to the next flowform. The size
and positioning of the entry bowl, its outlet, the
larger basins and their outlet and the shaping of the
surfaces over which the water flows is what does
the "coaxing" referred to above. Depends on design
shape and size, flowforms can allow flow rates
ranging from 12L/min to 250L/min of water [5]. Fig. 1. Flow patterns in flowforms [1]
Research on fabricating flowforms from ultra-high … 485

The principle of the flowforms lies in it creating "figure of 8" movement in the water
(Fig .1). Running through a flowforms cascade, water streams repeatedly through each
vessel’s side-to-side following “figure of 8” patterns, thereby multiplying the actual
length the water travels by some 10 times that of the length of the physical flowforms
cascade [6]. As the results, oxygen is drawn into the water, and thus improves condi-
tions for micro-organisms to help break down waste in a natural way, consequently
enhances water quality.

3 Materials and methods

3.1 Materials

As described above, UHPC was used to fabricate flowforms. Composition and ratio of
aggregates used in UHPC fabrication are: sand 45.5%; cement 31.8%; fly ash (taken
from Pha Lai thermal power plant) 9%; silica fume 4.5%; water 7.9%; superplasticizer
additive 1.3%. The workability of concrete mixes is determined through the flowing
ability of concrete. With this mixing ratio, UHPC achieved flow diameter of 209 mm,
compressive strength at 28 days of 134.5 MPa [7].
The flowform design chosen in this study have the shape as shown in Fig. 2 and the
dimensions: 270 x 270mm in plan, 90mm in height. Using the same prototype, 3 types
of flowform were made of different materials: white cement (weighted 2.5kg), UHPC
(weighted 2.45kg) and glass enameled ceramic (weighted
1.8kg).
The fabrication procedure of UHPC flowforms con- 270mm
sists of the following steps: (1) forming and shaping a
flowform prototype; (2) testing the prototype with water;
(3) creating mold from composite material; (4) preparing 270mm
a mixture with the ratio of concrete ready-mix, water and
pigments at 81.5%, 16.3% and 2.2%, respectively; (5)
pour and unload mold; (6) curing by immersion in water
for 28 days in order to achieve optimal intensity and avoid
surface cracking. Fig. 2. Flowform sample

3.2 Experimental setup

During 2 years, a cascade of 7 flowforms (2 made from glass enameled ceramic, 2 made
from white cement and 3 made from UHPC) has been the aeration solution for a series
of experiments including:

 Research on aeration capacity of flowforms: water samples were collected from

Linh Quang lake – an urban lake heavily polluted by surrounding residential area.
Typical values of chemical oxygen demand (COD), total nitrogen (TN) and total
phosphorous (TP) during rainy season in August 2016 were as high as 227 mg/L, 34
mg/l and 4 mg/l respectively, while pH was 7.25 and dissolved oxygen (DO) was
486 T. T. H. Ung et al.

not detected. The experiment was conducted in 7 batches to study the aeration per-
formance of the flowforms;
 Research on application of a system of flowforms and biological pond for improving
urban lake water polluted by domestic wastewater: samples were collected from the
same Linh Quang lake. The experiment was carried out from August 2016 till July
2017 with pause intervals;
 Research on application of a system of flowforms, constructed wetlands and biolog-
ical pond for improving lake water polluted by leachate: water samples were artifi-
cially made by mixing leachate from a lab-scale landfill diluted by tap water at the
volume ratio 1:30. Typical values of COD, NH4⁺, NO3‾ were 190mg/L, 87.5mg/L
and 4.5mg/L respectively, while pH was 7.81 and DO was 0.6mg/L. The experiment
was carried out from August 2017 till July 2018 with pause intervals.

All of above experiment setups were placed outdoors, thus the flowforms were exposed
directly to weather conditions.

4 Results and discussions

4.1 Aeration performance

As seen on the Fig. 3 below, at the experimental conditions the cascade of 7 flowforms
could raise DO value from 0mg/L to an average value of 5.2mg/L; the average increase
of DO value by each flowform was approximately 0.7mg/L. During this experiment,
the flowforms swapped position every day to find out whether the aeration performance
was affected by the material they were made of. The data presented on Fig. 3 also
demonstrates that no significant influence of material was observed.

6,00

5,00 5,16
23.8.16
4,81
4,33 24.8.16
4,00
3,74
DO, mg/L

25.8.16
3,00 2,91 26.8.16

2,00 2,00 27.8.16

28.8.16
1,00
0,86 29.8.16

0,00 Average
0 1 2 3 4 5 6 7 8
Flowform order from top to bottom

Fig. 3. Dissolved oxyen change over the flowform vessels

Research on fabricating flowforms from ultra-high … 487

4.2 Durability of materials

Fig. 4 and Fig. 5 below show how the materials used for fabricating flowforms held up
with time. After two years of exposing to wastewater and weather impacts, small cracks
have developed on the surface of the white cement samples and even more pronounced
cracks found on enameled ceramic ones. However, the surface of the UHPC samples
stays smooth without noticeable cracks.

White cementt Enameled ceramic UHPC

Fig. 4. Surface of new flowforms

White cement Enameled ceramic UHPC

Fig. 5. Surface of the flowforms after two years in operation

5 Conclusions

This study demonstrates that flowforms can be an effective aeration measure for pro-
cessing contaminated water. Besides, they are very suitable to those situations when a
low-cost, environment- and landscape-friendly solution is needed. On the other hand,
UHPC with the utilization of local fly ash and silica fume as admixtures not only has
proved its durability over time, but also shows a potential of a “green” option for tack-
ling industrial wastes. The modest results of this study would contribute to stimulate a
wider application of UHPC flowforms as an cost-effective and reliable aeration solution
in urban context.
488 T. T. H. Ung et al.

References

1. Wilkes, John.: Flowforms: The Rhythmic Power of Water, Floris Books, Edinburgh
(2003).
2. Wilkes, AJ.: Flowforms: the rhythmic power of water. Floris Books, 2nd edition, Ed-
inburgh (2005).
3. Joachim Schwuchow, John Wilkes ARCA, Costantino Giorgetti Eng. Iain Trousdell
BA, Alison Trousdell BA: Flowform Water Research, Published by the Healing Wa-
ter Institute (2008).
4. Benjamin A. Graybeal: Material Property Characterization of Ultra-High Perfor-
mance Concrete. US Department of Transportation Federal Highway Administra-
tion, Federal Highway Administration Research and Technology, 186p (2006).
5. Trousdell, Iain, Virbela Flowforms & Biodynamic Preparations Stirring. Hastings,
NZ (1991).
6. Wilkes, John.: Flow Design Research Relating to Flowforms. in: Chaos, Rhythm
and Flow in Nature. The Golden Blade 46. Floris Books. Edinburgh (1993).
7. Nguyen Cong Thang, Nguyen Van Tuan et all.: Research on manufacturing ultra-
high performance concrete using silica fume and fly ash mixtures available in Vi-
etnam. Journal of Construction Science and Technology. Issue 2/2013, p24-36
(2013).
 Research on application of flowforms in combination
with planted constructed wetland for improving water
quality of urban polluted lakes

Thi Thuy Ha Ung∗1, Tuan Hung Pham∗1, Tho Bach Leu1, Thi Hien Hoa Tran1, and
Hong Nhung Chu1

1
National University of Civil Engineering, Hanoi, Vietnam – Vietnam
* uttha2002@gmail.com; hungpt@nuce.edu.vn

Abstract. With continuously increasing urbanization in Vietnam, urban lakes

pollution by domestic wastewater has become a serious issue, which results in
degradation of lake water quality and consequent deterioration of landscape as
well as environment for urban residents. Finding low-cost, simple and sustainable
solutions would help to overcome that issue in Vietnam’s context. This study
aims at proposing a solution which includes low-cost, nature- and landscape-
friendly processes such as wetlands and flowforms for improving polluted urban
lakes. The study was carried-out on a pilot system consisting of a cascade of
flowforms and a planted constructed wetland with flow recirculation. Water sam-
ples used in this study have been collected from an urban lake in Hanoi, which
receives wastewater from a densely populated residential area. The research re-
sults have demonstrated that this solution could achieve high treatment efficien-
cies: Color - 92%, Chemical oxygen demand (COD) - 92%, Ammonia (NH4+-N)
- 98% and Total Phosphorus (TP) - 87%. The effluent quality could meet require-
ment by Vietnam’s National technical regulation on surface water quality QCVN
08-MT:2015/BTNMT, level B1 in terms of the above mentioned parameters.

Keywords: Urban, lake, pollution, water, landscape, flowform, wetland, low-cost.

1 Introduction

Recently in Vietnam, the utilization of low-cost or simulating nature processes such as

wetlands and flowforms for improving urban water environment has drawn an increa-
sing interest due to their simplicity, sustainability and friendly landscape.
Planted constructed wetland (CW) is a system of vegetation, soil, water and micro-
bial populations that uses natural processes to carry out water treatment. The processing
takes place in CWs largely due to decomposition by microorganisms along with plant
uptake to remove organic matter, nitrogen, etc... Amongst the factors influencing CWs
performance, dissolved oxygen (DO) plays a crucial role, since it is vital for microbial
activities [1]. It has been consistently documented in many reports that the higher the
oxygen content the faster organic matter can be degraded in CWs [2]. Besides, due to

© Springer Nature Singapore Pte Ltd. 2020 489

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_76
490 T. T. H. Ung et al.

competition for oxygen between organic matter decomposition and nitrification pro-
cesses, lack of DO would inhibit nitrification or even affect the total nitrogen removal
efficiency to a large extent [3]. Amongst the solutions for enhancing DO content in
wastewater engineering, flowforms have emerged as an effective, low-cost and land-
scape-friendly option which has been widely applied in Europe, US and Australia [4].
Flowforms are vessels that seek to emulate the swirls of vortices of the mountain
stream enabling water to re-oxygenate (Fig. 1). When arranged in a cascade, flowforms
would provide strong vortices resembling a figure of “8” (Fig. 1) and help the aeration
to occur effectively in a small
space, while creating a harmoni-
ous movement of water that sim-

400m
ulates nature rhythms.
This paper presents the results
of the research on application of 400m
flowforms (Fig. 2) in combina-
tion with a horizontal subsurface
flow CW (HSFCW) for treat- Fig. 1. Flow patterns in Fig. 2. Flowform shape
ment of urban lake water pol- flowforms [4]
luted by domestic wastewater.

2 Materials and methods

2.1 Materials
The pilot system for experiments is as shown in Fig. 3 and includes the following com-
ponents:
- Flowforms cascade: consists of 7 vessels made of ultra-high performance concrete
with dimensions 400 x 400mm in plan, 140mm in height and shape as shown in Fig. 1.
- HSFCW: dimensions B×L×H = 0,55×1,2×0,4m. Crushed stone was used as media
and arranged into 3 layers. Bottom layer: 10cm in thickness and 3,5÷5cm in size; mid-
dle layer: 15cm in thickness and 1,5÷3cm in size; top layer 15cm in thickness and
0,5÷0,7cm in size. The water depth in the CW was maintained at 35cm. 20L of thick-
ened activated sludge from a local domestic wastewater treatment plant was used for
seeding at the acclimatization period. Umbrella papyrus (Cyperus alternifolius), a ubiq-
uitously found specie in Vietnam, has been selected to plant at a density 100 trees/m2.
The plants with the height ranged from 30cm to 60cm were taken from the same lake
where water samples were collected in order to facilitate their acclimatization. The ac-
climatization period lasted 15 days with the inlet flow of 150L/h. From the 7th day, the
growth in length of the plants at 1cm/d÷1,5cm/d were observed.
- Storage tank: dimensions B × L × H = 0,5 × 1,2 × 0,5m;
- Flow stabilizator: dimensions D x H = 190 x 500mm, made of PVC and used for
maintaining a constant flow to the experimental setup;
- Equipment: a pump used for feeding the experimental system; the valves V1, V2,
V3, V4 used for regulating the influent rate, pump flow rate and drainage respectively.
Research on application of flowforms in combination … 491

V2
Flow stabilizator
V1

Inlet Flowforms
Collecting Plant
well 2 device

Pump Collecting
well 1
V3
Storage tank V4 Horizontal subsurface flow CW Flowform cascade

Fig. 3. Pilot system components

2.2 Methods

Experimental setup
Lake water samples were fed into the experimental setup with recirculation flow as
follows: Storage tank  Flowforms  HSFCW  back to Storage tank.
The aeration performance of the flowform as shown in Fig.1 has been previously
studied by the authors of this paper [5]. The highest aeration rates were obtained at flow
rates ranging between 150L/h and 200L/h. Therefore, the flow rates of 150L/h (in 1st
stage) and 200L/h (in 2nd stage) were chosen for feeding the pilot system. The experi-
ment lasted from March 20 to May 27, 2018 with three batches of 10 days in each stage.

Sampling and data analysis

Lake water samples were collected from Kim Lien lake (Hanoi) – a typical urban lake,
polluted by domestic wastewater. Samples quality parameters are shown in Table 1.
Samples were collected from Storage tank once a day and analyzed for pH, Color,
COD, NH4+-N, NO3‾-N, TP in accordance with standard methods [6]. At each stage,
the results of 3 batches were averaged and presented on the Table 1 below.

Table 1. Quality parameters of lake water samples

Stage 1 Stage 2 QCVN 08-MT:2015/

Parameter
Spl 1 Spl 2 Spl 3 Spl 4 Spl 5 Spl 6 BTNMT, level B1
o
Temperature ( C) 29,7 28,9 31,2 26,5 28,3 26,3 -
pH 7,0 6,8 6,9 6,9 6,7 7,1 5,5-9
COD (mg/L) 166,6 174,9 181,3 279,2 248,5 165,7 30
DO (mg/L) 0,2 0,3 0,2 ND ND ND ≥ 4
-
NO3 -N (mg/L) 0,15 0,09 0,06 0,06 0,12 0,05 10
NH4+-N (mg/L) 32,6 30,2 29,4 23,6 24,2 26,8 0,9
TP (mg/L) 1,57 1,62 1,71 2,56 2,48 2,56 PO43¯-P ≤ 0,3
Color (Pt-Co) 308,3 304,4 276,3 450,5 350,0 299,6 -
Remark: Spl – Sample; ND – Non-detectable.
492 T. T. H. Ung et al.

DO concentration before and after CW was measured every hour (from 7:00 to 17:00),
daily average values were used for assessment of DO change through CW.
Removal efficiencies of the experimental setup were calculated using formula:
𝑉0 − 𝑉𝑙𝑜𝑠𝑡
𝐶0 − 𝐶1
𝑉0
𝐸= × 100% (1)
𝐶0
Where C0: initial concentration (mg/L); C1: concentration at the end of each run (mg/L);
V0: initial volume of lake water sample (L); Vlost: volume of water being lost by sam-
pling, evaporation and transpiration (L).

3 Results and Discussion

3.1 Dissolved oxygen change in CW

Fig. 4 below denotes that after the first day of operation, average DO concentration at
the CW inlet reached 3,9mg/L to 4,8mg/L and remained stable throughout each stage.
This has proved the high aeration efficiency of the flowforms. The DO value at the
outlet from CW fluctuated from 2,9mg/L to 3,7mg/L, which indicates that there was
sufficient DO for both organic matter decomposition and nitrification to occur in CW.
06
Stage 1 ,
05 CW inlet
04 Stage 1 ,
DO, mg/L

CW outlet
03
Stage 2 ,
02 CW inlet
01 Stage 2 ,
CW outlet
00
0 1 2 3 4 5 6 7 8 Days

Fig. 4. DO change in CW

3.2 Performance of the experimental setup

Data on Fig. 5 and Fig. 6 indicate that, although the influent rates at each stage of ex-
periment were significantly different (150L/h in the 1st compared to 200L/h in the 2nd),
performances in both stages followed a similar trend.
COD removal: during both stages, a sharp decrease of COD concentration in the first
4 days and slower decline from the 5th day was observed. On the 7th day, COD concen-
tration reached 28,1mg/L in 1st stage and 26,1mg/L in 2nd stage. Removal efficiencies
were as high as: 90,7% in 1st stage and 92,6% in 2nd stage. Such trend shows that bio-
decomposition was the main mechanism of COD removal, although filtration through
media layers and accumulation in root zone should have also take place.
NH4+-N removal: as can be seen on Fig. 5 and Fig. 6, there was a consistency in
changes of NH4+-N and NO3‾-N concentrations. During the first 4 days of operation,
Research on application of flowforms in combination … 493

NH4+-N content decreased by more than 10 times, down to 1,9mg/L and 1,7mg/L in 1st
and 2nd stages respectively, while NO3‾-N content rose sharply from 0,1mg/L to
5,5mg/L in 1st stage and from 0,08mg/L to 4,3mg/L in 2 nd stage. After the 7th day, the
average ammonia removal efficiency could reach 98% in both stages with NH4+-N con-
tent as low as 0.8mg/L in 1st stage and 0,4mg/L in 2nd stage, while NO3‾-N content
remained less than 5,6mg/L in 1st stage and 4,5mg/L in 2nd stage.

Fig. 5. Performance of the experimental setup during the 1st stage

Fig. 6. Performance of the experimental setup during the 2nd stage

While the DO data characterized the aerobic condition in CW during both stages, the
data of NH4+-N indicate that such low concentrations of ammonia were the result of the
nitrification process. Since denitrification process was not likely to happen, the low
content of NO3‾-N concentrations would link mostly to plant uptake.
494 T. T. H. Ung et al.

Compared to the results of Li et al. [7], both COD and NH4+-N removal efficiencies
after 7th day were at the same level – about 90% and 98% respectively. This points out
that, although being a passive and simple device, the flowforms can perform as good
as air compressor - aeration tubes system used in study by Li et al. [7].
TP removal: TP decreased rapidly in the first 3 days in both stages, then gradually
decreased to 0,3mg/L in 1st stage and 0,6mg/L in 2nd stage. Such trend of TP content
would be described that it was removed mainly by accumulation in filter media over
accumulation by aerobic microorganisms and plant uptake. The TP removal efficiency
reached 89% in 1st stage 1 and 85% in 2nd stage.
Color removal: data show a very good removal efficiency at about 92% and a stable
level at about 40Pt-Co from the 7th day of the operation in both stages.
The experiment results have demonstrated very good performance of the Flowforms
- HSFCW pilot system in treatment of polluted lake water samples: after 7 days of
operation, all the parameters COD, NH4+-N, NO3‾-N and TP (in terms of PO43‾) would
meet the requirement by QCVN 08-MT:2015/ BTNMT, level B1 (see Table 1 above).

4 Conclusions

The initial results proved that although being very simple, flowforms are a very effec-
tive option for aeration. Coupled with flowforms, CW systems could deliver high per-
formance in removing organic matters, nutrients and color, thus could be applied for
improvement of urban lakes, polluted by domestic wastewater. These systems could be
placed next to or integrated into the lakes embankment and would serve not only en-
hancing water quality, but also shaping lake landscape, especially in Vietnam’s context.

Acknowledgement: The authors would like to thank for funding support from National
University of Civil Engineering (NUCE) under Grant No 16-2019/KHXD-TD.

References
1. Wu H., Zhang J., et all: A review on the sustainability of constructed wetlands for wastewater
treatment: Design and operation. Bioresource Technology, 175, 594-601 (2015).
2. Pan J., Fei H., Song S., et all: Effects of intermittent aeration on pollutants removal in subsur-
face wastewater infiltration system. Bioresource Technology, 191, 327-331 (2015).
3. Hu Y., Zhao Y., et al.: High rate nitrogen removal in an alum sludge-based intermittent aeration
constructed wetland. Environmental Science & Technology, 46(8), 4583-4590 (2012).
4. Schwuchow J., Wilkes J., Trousdell I.: Energizing water – Flowform technology and the power
of nature, Sophia Books (2010).
5. Ung T. T. H., Leu T. B. et al.: Research on application of flowforms to improve water quality
and create environment-friendly landscapes at polluted lakes. Grant No134-2017/KHXD-TD.
National University of Civil Engineering (2018).
6. APHA, AWWA, WEF: Standard methods for the examination of water and wastewater. 19 th
Edition, American Public Health Association, Washington, D.C., USA (1995).
7. Li F., Lu L., Zheng X., et all: Enhanced nitrogen removal in constructed wetlands: Effects of
dissolved oxygen and step-feeding. Bioresource Technology, 169, 395-402 (2014).
 A model of local kinetics of sorption to understand the
water transport in bio-based materials

N. Reuge1, F. Collet1, S. Pretot1, S. Moissette1, M. Bart1, C. Lanos1

1 Université Européenne de Bretagne, Laboratoire de Génie Civil et Génie Mécanique, Equipe
Matériaux Thermo Rhéologie, IUT Rennes, 3 rue du Clos Courtel, BP 90422 35704 Rennes,
France

Abstract. The classic models describing the hygric mass transfers inside porous
materials seem unsuitable in the case of bio-based materials. They are based on
the assumption of instantaneous local equilibrium between relative humidity
and water content [1]. These two parameters evolve according to the diffusive
fluxes following the sorption isotherms. This study shows that it leads to predict
much shorter times of stabilization than those experimentally obtained. A new
approach is presented here, it frees from the local instantaneous equilibrium in-
troducing a local kinetics to describe the transformation of water from vapor
state to liquid state and vice versa. The local kinetics of sorption is coupled with
the well-known hysteresis phenomenon. It is adjusted from bibliographic data
[2] giving mass evolution of three hemp concretes under adsorption / desorption
conditions. 1D cylindrical simulations allows an excellent fitting on the experi-
ments. Finally, a semi-empirical model is proposed, allowing to determine the
kinetics parameters more easily. The effect of the local kinetics model on the
hygrothermal transfers occurring through a bio-based wall is then studied.

Keywords: bio-based porous materials, hemp concrete, local kinetics, sorption,

hygric transfer, modeling.

1 Introduction

Water sorption in porous media involves complex and coupled phenomena such as
vapor / liquid water mass transport by Fickian diffusion and equilibrium isotherms of
adsorption / desorption associated with hysteretic phenomena [3].
Classic simulation tools of hygrothermal transfers in building materials are based
on the assumption that for a given local relative humidity ( ), the associated equilib-
rium local water content (w) is reached instantaneously. Here, various bio-based ma-
terials are studied: it is demonstrated that such an assumption leads to serious incon-
sistencies. Thus, a local kinetics of sorption may exist. It had already been established
for bio-based materials [4, 5] as well as for traditional cement compounds [6].
The first part of this study is a summary of the hygric characterization performed
on three selected hemp concretes [2] and on the bio-based materials developed during
the ISOBIO project [7]. Then, the theoretical background necessary to model the

© Springer Nature Singapore Pte Ltd. 2020 495

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_77
496 N. Reuge et al.

water sorption is presented. A new expression for the local kinetics of sorption is
proposed and validated for the three studied hemp concretes in the following part.
Then, a semi-empirical analytical model is proposed, allowing to estimate straightly
the kinetic parameters. The final part studies a bio-based multilayered wall developed
during the ISOBIO European project: simulations performed with a 1D Cartesian
code based on the local kinetics model (TMCKIN) is compared against measurements
carried out on a demonstrator (HIVE, Wroughton, UK).

2 Materials characterizations

2.1 Classic properties

The three studied hemp concretes are made of hemp shiv mixed with lime based min-
eral binders. Their properties had been characterized in [2]. The first one is a precast
hemp concrete (PHC) also known as EASY-RTM, the second one is a sprayed hemp
concrete (SHC) and the third one is a molded hemp concrete (MHC).
The ISOBIO materials studied are: a BCBTM lime-hemp render (BCB), a
CAVACTM Rigid panel made of hemp shiv and an organic binder (CAV), a Biofib
Trio flexible from CAVACTM made of hemp wool (BIO), an OSB3, a ProclimaTM
INTELLO membrane (INT), a Lignicell CSBTM panel made of straw (CSB) and a
CLAYTECTM clay-hemp plaster (CLA). Their properties has been studied in [7].
Table 1 is a compilation of the classic properties of all these materials at dry state:
the bulk densities , the open porosities 0, the vapor diffusion resistance factors µ 0,
the thermal conductivities 0 and the specific heat capacities Cp0.
The Proclima INTELLO membrane is a hygrovariable material evolving very sig-
nificantly as a function of the relative humidity RH following a logistic power law: µ
= 1000/(7.33.10-3+1.8.RH7.64). The evolution of the thermal conductivities as a func-
tion of the water content has been modeled by the self-consistent scheme [8].

Table 1. Materials classic properties

Mat. 0 µ0 0 Cp0
(kg.m-3) (-) (-) (W.m-1K-1) (J.kg-1.K-1)
PHC 450 0.68 4 - -
SHC 425 0.66 4 - -
MHC 425 0.77 4 - -
BCB 530 0.55 9 0.13 1006
CAV 197 0.43 9 0.07 2100
BIO 28 0.35 2.5 0.039 1800
OSB3 567 0.81 130 0.13 1600
INT 85 0.085 1.37.105 2.74 2500
CSB 505 0.67 23 0.10 1700
CLA 1392 0.29 10 0.62 1040
A model of local kinetics of sorption to understand … 497

2.2 Isotherms of sorption

The sorption isotherms were measured according to a discontinuous method: the time
dependent water content W was determined at successive stages of increasing (and
then decreasing) relative humidity at 23°C [2]. Note that for the ISOBIO materials,
only the adsorption isotherm have been determined [7]. The measured equilibrium
water content Weq are fitted here by the Van Genuchten model (VG):

(1)
where Wsat is the maximum equilibrium local water content (i.e. 1000 0) and h and
are adjustment coefficients. The adjustment coefficients are given in Table 2.

Table 2. VG model: adjustment coefficients

Mat. hads (-) ads(-) hdes (-) des (-)

PHC 166 2.05 258 1.65
SHC 378 1.77 500 1.46
MHC 185 1.98 75 1.66
BCB 8524 1.38 - -
CAV 4830 1.435 - -
BIO 198852 1.47 - -
OSB3 25410 1.325 - -
INT 2091 1.42 - -
CSB 12966 1.334 - -
CLA 12000 1.36 - -

3 A model of local kinetics of sorption

3.1 Mass transport governing equations

Assuming that the rate of sorption Rs is very fast compared to the vapor diffusive flux
(i.e. instantaneous equilibrium w / assumption), the mass transport can be described
by the so-called Künzel single governing equation [1]:

where w T
is given by the isotherm of sorption.
Now, considering that the instantaneous equilibrium w / assumption is not justi-
fied, there are two mass balance equations to consider:

(3,4)
498 N. Reuge et al.

Note that the coupling with a hysteretic model (e.g. [3]) can easily be done: the rever-
sal points are obtained when the sign of Rs changes.

3.2 Sorption rate

A kinetics is usually expressed as a kinetic constant multiplied by a driving force. The
simplest consistent expression that can be conceived for the sorption rate is the fol-
lowing:
Rs k0 weq w (5)
where k0 is the local kinetic constant of sorption (adsorption or desorption) and weq is
the equilibrium local water content given by the sorption isotherm at . [4] came to
establish the same expression. However, the value of the kinetic constant k0 has to be
adjusted to quite different values as a function of the hygric conditions. Consequently,
a more complex driving force has been established in our studies introducing a kinetic
order of 2:
2
Rs k0 weq w (6)

4 Validation of the local kinetics model

The model has been validated comparing the sorption curves measured by [2] per-
formed on 5 cm diameter cylindrical samples of PHC, SHC and MHC to 1D cylindri-
cal calculations. As evidenced by the results reported in Fig. 1a (adsorption) and 1b
(desorption) for PHC, considering a value of k0 of 2 day-1/(kg.m-3), the adjustments
are globally very good. In adsorption conditions, the stabilizations of the water con-
tent occur from 5 days (at low RH) to several week at high RH. If the Künzel equation
is considered instead, calculations lead to stabilizations in less than one day!
For SHC and MHC, the adjustments are also very good and lead to k0 of 0.65 day-
1
/(kg.m-3) and 0.5 day-1/(kg.m-3) respectively.
-3 -3

Fig. 1. Temporal evolution of the global water content W in a sample of PHC for (a) increasing
/ (b) decreasing RH steps - Experiments (points) and simulations (lines), k0 = 2 day-1/(kg.m-3).
A model of local kinetics of sorption to understand … 499

5 A semi-empirical model for kinetic parameters identification

From [9] focusing on the hydration / dehydration of hydrated minerals, the global
kinetics of water adsorption / desorption of a sample can be well described by:
(7)
-1
where K0 (day ) is a global kinetic factor. Considering a kinetics order of 2, the fol-
lowing modified expression can be proposed:
(8)
where K0 is in day-1/(kg.m-3). Now, adjusting this model on the sorption curves, the
agreement is good and the following values of K0 are obtained: 1.75, 0.6 and 0.45 day-
1
/(kg.m-3) for PHC, SHC and MHC respectively. This is quite close to the values of
the local kinetic constant k0 deduced from the 1D cylindrical calculations.

6 Simulations of the HIVE demonstrator

An instrumented ISOBIO reference wall has been set up at the HIVE demonstrator
(Wroughton, UK) by the University of Bath (UK). This reference wall consisted in
(from outdoor to indoor): BCB, 25 mm; CAV, 50 mm; BIO, 145 mm; OSB3, 12 mm;
INT, 0.25 mm; BIO, 45 mm; CSB, 40 mm and CLA, 15 mm. RH and T sensors were
set up outdoor and indoor and at various positions inside the wall. The measurements
have been performed in winter 2018 during 18 days, i.e. 434 hours.
Some samples of these materials have been characterized [7], allowing, thanks to
the semi-empirical model (eq. 8), to determine the following kinetic constant K0 (day-
1
/(kg.m-3): 0.0014 for BCB, 0.25 for CAV, 10 for BIO and 0.007 for CSB.
The simulations have been performed with two 1D Cartesian tools: TMC based on
the Künzel equation and TMCKIN based on the local kinetics model. Fig. 2 shows the
RH measurements and the results of the simulations at the interface BIO / OSB3:
Contrary to TMC, TMCKIN predicts the good RH dynamics.
RH (%)

Fig. 2. Temporal evolution of RH at BIO / OSB3 interface – measurements and simulations

(TMC and TMCKIN).
500 N. Reuge et al.

7 Conclusion

This study has allowed to understand, model and validate a new phenomenon: the
local kinetics of sorption. An expression for the sorption rate has been established and
validated.
Beside a homemade code named TMC based on the classic approach, a new 1D
simulation tool named TMCKIN has been developed taking into account the local
kinetics of sorption. Finally, the data acquired at the HIVE demonstrator inside the
ISOBIO multilayered reference wall have been successfully simulated with
TMCKIN. TMC underpredicts the dynamics of the RH variations (at the hour / day
scale) while TMCKIN leads to results in good agreement with the measurements.

Acknowlegments

This work has been performed and funded in the framework of the European pro-
ject ISOBIO – (http://isobioproject.com) within the scope the of the research and
innovation program Horizon 2020 (agreement No. 636835). Special thanks goes to
Andrew Shea and his team of University of Bath (UK) for having performed the
measurements at the HIVE demonstrator.

References
1. Künzel, H.M.: Simultaneous Heat and Moisture Transport in Building Components – One-
and Two-Dimensional Calculation Using Simple Parameters. Fraunhofer IRB Verlag Sutt-
gart, ISBN 3-8167-4103-7 (1995).
2. Collet, F., Chamoin, J., Pretot, S., Lanos C.: Comparison of the hygric behaviour of three
hemp concretes. Energy and Buildings 62, 294–303 (2013).
3. Huang, H.C., Tan, Y.C., Liu, C.W., Chen C.H.: A novel hysteresis model in unsaturated
soil. Hydrological Processes 19(8), 1653-1665 (2005).
4. Nyman, U., Gustafsson, P.J., Johannesson, B., Hägglund, R.: Numerical method for the
evaluation of non-linear transient moisture flow in cellulosic materials. Int. J. Numer.
Meth. Engng 66, 1859–1883 (2006).
5. Frandsen, H.L., Damkilde, L., Svensson, S.: A revised multi-Fickian moisture transport
model to describe non-Fickian effects in wood. Holzforschung, Vol. 61, pp. 563–572,
Copyright by Walter de Gruyter, Berlin, New York (2007).
6. Zeng, Q., Zhang, D., Li K.: Kinetics and equilibrium isotherms of water vapor adsorp-
tion/desorption in cement-based porous materials. Transp. Porous Med. 109, 469–493
(2015).
7. Collet, F., Pretot, S., Colson, V., Gamble, C.R., Reuge, N., Lanos, C.: Hygric properties of
materials used for ISOBIO wall solution for new building. Accepted for publication in
ICBBM 2019 proceedings, Belfast, UK.
8. Collet, F., Pretot, S.: Thermal conductivity of hemp concretes: Variation with formulation,
density and water content. Building Materials 65, 612–619 (2014).
9. Nguyen 2009 Nguyen K.-S.: Comportement thermos-chimique de matériaux minéraux :
application à la protection incendie. PhD thesis, LGCGM, Univ Rennes (2009).
 Air permeability of cover concrete quality of precast box
culverts affected by casting direction

May Huu Nguyen1, 2, Kenichiro Nakarai1, *, Saeko Kajita1

1
Department of Civil and Environmental Engineering, Graduate School of Engineering, Hiro-
shima University, 1-4-1, Kagamiyama, Higashi-Hiroshima, Hiroshima 739-527, Japan
2
Department of Bridge and Tunnel, Faculty of Civil Engineering, University of Transport
Technology, 54 Trieu Khuc, Thanh Xuan, Hanoi 100000, Vietnam
* Corresponding author
E-mail addresses: maynh@utt.edu.vn (M.H. Nguyen), nakarai@hiroshima-
u.ac.jp (K. Nakarai), m186051@hiroshima-u.ac.jp (S. Kajita)

Abstract.
Casting direction is an external factor that can influence the properties, includ-
ing air permeability of concrete structures. In the current study, the effect of this
factor on the air permeability of precast box culverts was investigated. The Tor-
rent air permeability test was used to measure coefficient of air permeability kT
for two box culvert specimens produced using ordinary Portland cement, water-
to-binder ratio of 0.485, different casting directions (vertical and horizontal) at
the age of 3 months. The obtained results show the different cover concrete qual-
ities between the surfaces of the horizontal casting specimen, whereas the short-
comings during casting process were detected of the vertical casting specimen.

Keywords: Cover concrete, Air permeability, Box culvert, Casting direction.

1 Introduction

The transport properties are increasingly acknowledged as durability index for evaluat-
ing cover concrete quality [1], and can be classified into absorption, permeability, and
diffusion [2]. Therein, air permeability was pointed out as one of the appropriate in-
dexes that can provide useful information relating to the capillary porosity, pore system
of the cover concrete for characterizing the durability of concrete structures [3–4], and
it is affected by several parameters [5].
Casting direction is an external factor that can influence the properties of concrete struc-
ture such as segregation, bleeding, and hence permeability [6–9]. Casting direction can
result in differential accumulation of setting of aggregate leaving a motor layer on the
top of concrete [10]. Additionally, the shortcomings during production operations can
occur due to unskilled in some stages at the workplace [11-12]. These issues are signif-
icant influence on transport properties, especially the air permeability. Therefore, eval-
uation of air permeability affected by casting direction, on practical standpoints, is
meaningful task in controlling quality of concrete structure.

© Springer Nature Singapore Pte Ltd. 2020 501

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_78
502 M. H. Nguyen et al.

Accordingly, this study aims to evaluate the influence of casting direction on the air
permeability of two precast box culverts. The Torrent air permeability test was used to
measure coefficient of air permeability, kT, for two precast box culvert specimens pro-
duced using ordinary Portland cement, water-to-binder ratio of 0.485, different casting
directions at the age of 3 months. Counter-maps was used to describe the distribution
of kT on measured surfaces.

2 Experimental program

2.1 Materials and mix proportions

Table 1 illustrates the properties of the materials employed in this study. Tap water was
used to mix the cement components.

Table 1. Material properties

Material Type Properties

Cement (C) Ordinary Portland Density: 3.16g/cm3
cement Specific surface area: 3320 cm2/g
Fine aggregate (S) Crushed quartz Surface-dry density: 2.57 g/cm3
Coarse aggregate (G) Crushed gravel Surface-dry density: 2.72 g/cm3
Expansive agent (Ex) Ettringite type Density: 2.95 g/cm3
Specific surface area: 3000 cm2/g
Chemical admixture (Ad) Superplasticizer Polycarboxylic acid compound
The specimens were prepared by mixing the cement, aggregates, expansive agent,
chemical admixture, and tap water using the proportions presented in Table 2.

Table 2. Material mix proportions and measured parameters of the specimens

Slump Air s/a 2 Unit mass (kg/m3)

W/B 1
(cm) (%) (vol%) W C S G Ex Ad
0.485 18.0 2.0 45.0 173 307 801 1044 50 2.32
1
B: Binder includes cement (C) and expansive agent (Ex).
2
a: total volume of aggregates, s: volume of fine aggregate

2.2 Specimen preparation

The ready-mixed concrete was procured from the manufacturer. To investigate the ef-
fect of casting direction, a specimen was cast in vertical as precast used in the factory,
whereas another was cast in horizontal as cast-in-place on the site. The dimensions were
86 × 86 × 200 cm for both box culvert specimens (Fig. 1).
Regrading curing condition, after casting, all specimens were covered with blue plastic
sheets for 3 h, excluding surface II of horizontal specimen, and then cured with the
steam condition. In details, the temperature was increased at a steady rate 20 °C/h to 60
°C, and it was maintained at 60 °C for 3 h before returning to 20 °C over an 8-h period.
Air permeability of cover concrete quality … 503

At the 1-day age, specimens were exposed to a room with the average relative humidity
and temperature being about 62% RH and 16.6 oC during the experiments.

Fig.1. Geometries of the box culvert specimens and sketch test locations (cm)

2.3 Experimental methods

As shown in Fig. 1, the specific surfaces were given, namely, I to IV. The surfaces I, II
were employed as representatives to other surfaces. 15 test locations were then selected
on these surface (see Fig. 1). The air permeability tester, “PermeaTORR AC” (Active
Cell) developed by Torrent were employed to determine the coefficient of air permea-
bility kT [13]. The kT was measured at the ages of 30, 60, and 90 d to evaluate the
cover concrete quality according to age of the specimens.

3 Results and discussion

3.1 Changes in kT on measured specimens

Fig. 2. Changes in kT on measured surfaces

The variation of average kT values (geometric mean) on measured surfaces and stand-
ard deviation are shown in Fig. 2. The results indicated that kT on all surfaces slight
reduced from age of 30 to 60 d after that increased until 90 d. The first reduction can
504 M. H. Nguyen et al.

be explained due to the increase of the gross volume of the formed cement hydrates that
gradually fill part of the pore spaces originally occupied by water [14–15]. For the in-
creasing in kT after 60 d, it can be associated with the increase in the open pore content
caused by water evaporation [16–18]. In addition, the kT for surface II was larger
than surface I for horizontal specimen, whereas those were almost same for vertical
specimen. The large standard deviations imply the large difference on the obtained re-
sults that were presented in the next parts.

3.2 Evaluation cover concrete quality of the box culvert specimens

Horizontal box culvert specimen

As shown in Fig. 2, the kT for surface II was larger than the surface I. This result could
be explained by the negative segregation on the surface II-the top surface [9,19], and
poor curing condition due to it was not covered by blue plastic sheets. Based on the sug-
gested range of kT index values for quality classification of cover concrete, it can be
concluded that they were good quality (kT < 0.1 × 10-16 m2) surfaces excluding normal
quality areas on the surface II (kT > 0.1 × 10-16 m2) (see Fig. 3).

a)

b)
Fig. 3. Distribution of kT for horizontal box culvert at 90 days: a) surface I, b) surface II.

Vertical box culvert specimen

As plotted in Fig. 2, the kT values on the surface I were almost similar to the surface
II. It is quite plausible due to their casting and curing process were the same. Based on
these figures, some measurement points which had high kT values were detected in the
lower half of the specimen (from bottom to center). This observation can be mainly
explained by the shortcomings in compaction step during casting process. Based on kT
index, it can be concluded that they were good quality (kT < 0.1 × 10-16 m2) surfaces,
excluding normal quality areas on the surface I and II (see Fig. 4).
Air permeability of cover concrete quality … 505

a) b)
Fig. 4. Distribution of kT for vertical box culvert at 90 days: a) surface I, b) surface II.

4 Conclusions
Following conclusions are drawn based on the above-mentioned measurements:
• The cover concrete quality of tested surfaces were good excluding the normal qual-
ity areas on surface II and on the lower half of the horizontal and vertical specimen,
respectively.
• The effect of casting direction was reflected through existed normal quality areas
on these tested surfaces that could be explained by a combination of the negative
bleeding phenomenon and poor curing condition, as well as shortcomings during
casting and compaction, respectively.
On the basis of the findings in this study, the following recommendations are proposed:
• In the case of horizontal box culvert, all surfaces should be covered during applying
steam curing to prevent the water evaporation phenomenon, and then can improve
the quality of cover concrete.
• In order to prevent the similar shortcomings detected herein, in the case of vertical
box culvert, different technical should be combined during casting process to im-
prove the compaction step such as layer dividing, outer vibrations, and so on.

Acknowledgements
This study was supported by the All Japan Box-Culvert Association. The authors are
grateful to Mr. Hideaki Karasawa, Mr. Yuji Tadokoro and other members of the asso-
ciation for assisting with the specimen preparation in experiments.
506 M. H. Nguyen et al.

References
1. X. Pan, Z. Shi, C. Shi, T.C. Ling, N. Li, A review on concrete surface treatment
part I: types and mechanisms, Constr. Build. Mater., 132, 578-590 (2017).
2. M.H. Medeiros, P. Helene, Surface treatment of reinforced concrete in marine en-
vironment: influence on chloride diffusion coefficient and capillary water absorp-
tion, Constr. Build. Mater., 23(3), 1476-1484 (2009).
3. X. Pan, Z. Shi, C. Shi, T.C. Ling, N. Li, A review on surface treatment for concrete–
part 2: performance, Constr. Build. Mater., 133, 81-90 (2017).
4. G.K. Gandhi, D. Sharma, S. Misra, Quantifying changes in surface characteristics
of concrete due to progressive deterioration, J. Mater. Civ. Eng., 29 (8) (2017).
5. N. Hearn, Self-sealing, autogenous healing and continued hydration: what is the
difference? Mater. Struct., 31(212), 563-567 (1998).
6. M.S. Chin, M.A. Mansur, T.H. Wee, Effects of shape, size, and casting direction of
specimens on stress-strain curves of high-strength concrete. Materials Journal,
94(3), 209-219 (1997).
7. ACI Committee 238, Report on Measurements of Workability and Rheology of
Fresh Concrete, Farmington Hills (2008)
8. I. Navarrete, M. Lopez, Estimating the segregation of concrete based on mixture
design and vibratory energy, Constr. Build. Mater., 122, 384-390 (2016).
9. D. K. Panesar, B. Shindman, The effect of segregation on transport and durability
properties of self consolidating concrete, Cem. Concr. Res., 42(2), 252-264 (2012).
10. D. Bonen, S.P. Shah, Fresh and hardened properties of self-consolidating concrete,
Prog. Struct. Eng. Mat., 7,14-26 (2005).
11. B. Ray, P. Ripley, D. Neal, Lean manufacturing-A systematic approach to improv-
ing productivity in the precast concrete industry. PCI journal, 51(1), 62 (2006).
12. C. I. Goodier, Skills and training in the UK precast concrete manufacturing sector,
(2008).
13. R. J. Torrent, A two-chamber vacuum cell for measuring the coefficient of perme-
ability to air of the concrete cover on site, Mater. Struct. 25 (6), 358–365 (1992).
14. N. Hearn, Self-sealing, autogenous healing and continued hydration: what is the
difference?, Mater. Struct. 31 (212), 563–567 (1998).
15. T.C. Powers, Structure and physical properties of hardened Portland cement paste,
J. Am. Ceram. Soc. 41 (1), 1–6 (1958).
16. A. Abbas, M. Carcasses, J.-P. Ollivier, Gas permeability of concrete in relation to
its degree of saturation, Mater. Struct., 32 (1), 3-8 (1999).
17. W. Kubissa, M.A. Glinicki, Influence of internal relative humidity and mix design
of radiation shielding concrete on air permeability index, Constr. Build. Ma-
ter., 147 (Suppl. C), 352-361 (2017).
18. P.A.M. Basheer, É. Nolan, Near-surface moisture gradients and in situ permeation
tests, Constr. Build. Mater., 15 (2–3), 105-114 (2001).
19. I. Navarrete, M. Lopez, Understanding the relationship between the segregation of
concrete and coarse aggregate density and size, Constr. Build. Mater., 149, 741-748
(2017).
 Effects of Amounts and Moisture States of Clay-Brick
Waste as Coarse Aggregate on Slump and Compressive
Strength of Concrete

Phuong Trinh Bui1a, Xuan Nam Nguyen1b, My Ngoc Tang1c, Yuko Ogawa2d,
and Kenji Kawai2e
1
Ho Chi Minh City University of Technology, VNU-HCM, Ho Chi Minh, Vietnam
2
Hiroshima University, Higashi-Hiroshima, Japan
a
buiphuongtrinh@hcmut.edu.vn, b1412387@hcmut.edu.vn,
c
1612256@hcmut.edu.vn, dogaway@hiroshima-u.ac.jp,
e
kkawai@ hiroshima-u.ac.jp

Abstract. Effects of clay-brick waste (CBW) from demolition in construction

on slump and compressive strength of concrete were investigated to promote
the reuse of such waste at Vietnam for concrete technology towards sustainable
development. The replacements of coarse aggregate by CBW were 0, 10, 20,
and 40% by volume. To evaluate the effects of moisture states of CBW on
properties of concrete, two states of CBW with a size range of 5–25 mm were
prepared: (1) under oven-dry condition and (2) under saturated-surface dry con-
dition. Results showed that the higher the replacement of CBW, the lower the
slump of fresh concrete. The moisture states of CBW also affected slump of
fresh concrete and compressive strength of hardened concrete. Although CBW
under saturated-surface dry condition reduced the compressive strength at the
age of 3 days, internal water released from CBW promoted the cement hydra-
tion after 3 days, leading to an increase in compressive strength of concrete at
the ages of 7 and 28 days. Consequently, CBW can be reused as coarse aggre-
gate for making concrete towards sustainable development of construction ma-
terials.

Keywords: Clay-Brick Waste, Compressive Strength, Slump.

1 Introduction

Concrete is widely used as a construction material owing to its benefits such as high
performance, effective cost, and flexibility. However, traditional concrete production
is not an environmentally friendly process because its components such as cement and
natural aggregate have negative impacts to the surrounding [1]. The major environ-
mental impacts of natural aggregate mining are the conversion of land use, loss of
habitat, noise, dust, blasting effects, erosion, and so on [2]. Therefore, clay-brick
waste (CBW) from demolition in construction has been considered as an environmen-
tally friendly substitute for natural aggregates in concrete production. Generally, use
of this waste reduces the strength of concrete because it is comparatively weaker and

© Springer Nature Singapore Pte Ltd. 2020 507

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_79
508 P. T. Bui et al.

more porous than normal aggregate [3]. Recently, internal curing has been one of
promising methods to improve the strength and durability of concrete. It was found
that when internal curing with this waste up to 40% by volume of coarse aggregate is
applied to concrete, its strength is improved significantly [4, 5].
The purpose of this study was to investigate effects of amounts and moisture states
of this waste as coarse aggregate on slump and compressive strength of concrete in
order to promote the reuse of such waste at Vietnam for concrete technology towards
sustainable development.

2 Experiments

2.1 Materials

Portland cement (C) Type I (Nghi Son Cement Corp., Vietnam) confirming to ASTM
C150 was used. Its chemical compositions are listed in Table 1. River sand (RS) from
Dong Nai river and crushed stone (CS) from Hoa An quarry in Dong Nai province
were prepared as fine and coarse aggregates, respectively. Densities of C, RS, and CS
were 3.02, 2.62, and 2.70 g/cm3, respectively. After collecting from demolition in
construction at Ho Chi Minh city, Vietnam, CBW was crushed with a size range of 5–
25 mm in order to replace partially CS for concrete production, as seen in Fig. 1 (a).

Table 1. Chemical compositions of cement (% by mass).

Compositions CaO SiO2 Al2O3 Fe2O3 MgO Na2O K2O SO3 LOI
% 64.05 22.46 4.16 1.98 1.38 0.55 0.08 2.51 1.56

(a) (b)

Fig. 1. CBW after crushing (a) and particle size distributions of aggregates (b).

Particle size distributions of aggregates including CS, CBW, and RS are shown in
Fig. 1 (b). In addition, the crushing values and fineness moduli of CS and CBW were
12.99 and 21.07%, and 7.78 and 7.40, respectively. These indicates that CBW as
coarse aggregate had finer particles and was weaker than CS. CBW had the high wa-
ter absorption value of 16.65%. To evaluate the effects of moisture states of CBW on
Effects of Amounts and Moisture States of Clay-Brick … 509

properties of concrete, its two states were prepared: (1) under oven-dry condition and
(2) under saturated-surface dry condition. Oven-dry (D) condition was the condition
of aggregate from which all moisture was removed by heating in an oven at 105 oC to
constant weight. Meanwhile, saturated-surface dry (SSD) condition was the condition
of aggregate in which the surfaces of the particles were dry but the inter-particle voids
were saturated with water. Its density under D condition was 1.91 g/cm3 while that
under SSD condition was 2.21 g/cm3. [It is noted here the water content of CBW un-
der D condition was 0]. Water (W) and water-reducing agent (WRA) were also em-
ployed to make concrete mixtures.

2.2 Mixture proportion

A control mixture proportion with 0% CBW replacement was designed as per
ACI211.1-91 and ACI318. A dosage of WRA of 1.2% by mass of cement was pre-
pared not only to reduce partial amount of mixing water but also to make control mix-
ture with designed slump of 82 cm and compressive strength at 28 days of 40 MPa
which are suitable for industrial concrete structures. To investigate the effects of
amounts and moisture states of CBW on properties of concrete, four mixture propor-
tions having a constant water-to-cement ratio (W/C) of 0.41 were prepared in addition
to the control mixture proportion, as seen in Table 2. The replacements of coarse ag-
gregate by CBW were 0, 10, 20, and 40% by volume. Concrete mixtures were desig-
nated the following codes: B00, B10-D, B10-SSD, B20-SSD, and B40-SSD. The
numbers after B show the percentage replacements of CS by CBW (0%: B00, 10%:
B10, 20%: B20, and 40%: B40). D and SSD show moisture states of CBW (D: under
D condition and SSD: under SSD condition).

Table 2. Mixture proportion of all concretes.

Mixture Replacement kg/m3
ratio of CBW W/C s/a W C RS CS CBW WRA
(vol%)
B00 0 0.41 0.45 178 433 795 1001 0 5.2
B10-D 10 0.41 0.45 178 433 795 900 71 5.2
B10-SSD 10 0.41 0.45 178 433 795 900 82 5.2
B20-SSD 20 0.41 0.45 178 433 795 800 163 5.2
B40-SSD 40 0.41 0.45 178 433 795 600 327 5.2
s/a: sand-to-aggregate ratio

2.3 Mixing, slump testing, casting and curing condition

After the concrete components had been mixed in a mixer, slump of fresh concrete of
each mixture proportion (n = 1) was measured immediately. Then, the concrete mix-
tures were cast into cubic specimens (100×100×100 mm). After casting, the placing
surfaces of specimens were covered with polyethylene sheets and cured at room tem-
perature. All specimens were demolded 24 h after casting and then cured under rela-
tively sealed condition with polyethylene sheets until the designated test ages.
510 P. T. Bui et al.

2.4 Compressive strength test

Three cubic specimens for each concrete mixture were tested to determine their com-
pressive strength after aging for 3, 7, and 28 days.

3 Results and discussion

3.1 Effects of amounts of CBW

Slump. Effects of amounts of CBW under D and SSD conditions on slump of fresh
concrete are shown in Figs. 2 (a) and (b), respectively. According to Fig. 2 (a), 10%
replacement of CBW under D condition significantly reduced slump of fresh con-
crete. First reason is due to the lower fineness modulus of CBW than that of CS (i.e.
7.40 < 7.78). Second reason should be explained in Section 3.2. Similar to CBW un-
der D condition, the replacement of coarse aggregate by CBW under SSD condition
also reduced slump of fresh concretes (see Fig. 2 (b)). In addition, the higher the re-
placement of CBW, the lower the slump of fresh concrete. This is mainly due to lower
fineness modulus of CBW than that of CS.

(a) (b)

Fig. 2. Effect of amounts of CBW under D (a) and SSD (b) conditions on slump of fresh con-
cretes.

Compressive strength. Effects of amounts of CBW under D and SSD conditions on

compressive strength of hardened concrete at the ages of 3, 7, and 28 days are shown
in Figs. 3 (a) and (b), respectively. According to Fig. 3 (a), at the age of 3 days, the
compressive strength of B10-D was almost the same as that of B00. However, at the
age of 7 days, the compressive strength of B10-D was slightly higher than that of
B00. This may be due to a large amount of mixing water absorbed by CBW under D
condition in B10-D during mixing, which resulted in relatively lower W/C than B00.
However, this tendency was reverse at the age of 28 days. According to Fig. 3 (b), at
the ages of 3 days, concretes containing CBW under SSD condition (B10-SSD, B20-
SSD, and B40-SSD) had the lower compressive strength than concrete containing no
CBW (B00). This is due to higher crushing value of CBW than that of CS (i.e.
21.07% > 12.99%). However, at the ages of 7 and 28 days, concretes containing CBW
under SSD condition (B10-SSD, B20-SSD, and B40-SSD) had the higher compres-
Effects of Amounts and Moisture States of Clay-Brick … 511

sive strength than concrete containing no CBW (B00). This was also observed in the
previous studies [4, 5]. It confirms that water from CBW under SSD condition pro-
moted the cement hydration in such specimens after 3 days. Moreover, the compres-
sive strength of B10-SSD at the ages of 3, 7, and 28 days was higher than that of B20-
SSD and B40-SSD. It can be said that the adverse effect of the higher CBW may ex-
ceed the internal curing effect.

(a) (b)

Fig. 3. Effect of amounts of CBW under D (a) and SSD (b) conditions on compressive strength
of hardened concretes at the ages of 3, 7, and 28 days.

3.2 Effects of moisture states of CBW

Slump. Effect of moisture states of CBW on slump of fresh concretes is shown in Fig.
4 (a). With the specific amount of CBW replacement, fresh concrete containing CBW
under D condition had lower slump than that containing CBW under SSD condition.
It can be said that CBW under D condition absorbed the large amount of mixing water
during mixing, leading to significantly decrease workability of fresh concrete. This is
also a reason why CBW under D condition substantially reduced slump of fresh con-
crete as observed in Fig. 2 (a).

Compressive strength. Effect of moisture states of CBW on compressive strength of

hardened concrete at the ages of 3, 7, and 28 days is shown in Fig. 4 (b). At the age of
3 days, compressive strength of B10-D was higher than that of B10-SSD. The higher
compressive strength of B10-D was due to CBW under D condition which absorbed
excess water during mixing, which resulted in relatively lower W/C than B10-SSD.
However, this tendency was not observed at 7 and 28 days. It may be due to internal
water gradually released from CBW in B10-SSD which promoted the cement hydra-
tion in such concrete, leading to an increase in compressive strength of hardened con-
crete after 3 days of curing. Meanwhile, the internal water which CBW under D con-
dition absorbed during mixing in B10-D could be released very early, resulting in the
higher compressive strength at 3 days, but not at later ages (i.e. at 7 and 28 days).
512 P. T. Bui et al.

(a) (b)

Fig. 4. Effect of moisture states of CBW on slump of fresh concretes (a) and compressive
strength of hardened concretes at the ages of 3, 7, and 28 days (b).

4 Conclusions

The following conclusions can be drawn:

CBW reduced slump of fresh concrete and this reduction was increased with in-
creasing the amounts of CBW.
Although CBW under SSD condition reduced the compressive strength at 3 days of
hardened concrete, internal water released from CBW promoted the cement hydration
after 3 days, leading to the higher compressive strength of concrete containing CBW
than that of concrete containing no CBW.
The compressive strength at 3, 7, and 28 days of concrete with 10% replacement
with CBW under SSD condition was higher than that of concrete with 20 and 40%
replacements with CBW.
CBW under D condition significantly reduced slump of fresh concrete and com-
pressive strength of hardened concrete at the age of 28 days.

References
1. Wong, C.L., Mo, K.H., Yap, S.P., Alengaram, U.J., Ling, T.C.: Potential use of brick
waste as alternate concrete-making materials. Journal of Cleaner Production 195, 226–239
(2018).
2. Langer, W.H., Arbogast, B.F.: Environmental Impacts of Mining Natural Aggregate. In:
Fabbri A.G., Gaál G., McCammon R.B. (eds) Deposit and Geoenvironmental Models for
Resource Exploitation and Environmental Security. Nato Science Partnership Subseries: 2
(closed) (Series 2. Environment Security), vol 80, pp. 151–169. Springer, Dordrecht
(2002).
3. Zhang, S., Zong, L..: Properties of concrete made with recycled coarse aggregate from
waste brick. Environmental Progress & Sustainable Energy 33(4), 1283–1289 (2014).
4. Suzuki, M., Meddah, M.S., Sato, R.: Use of porous ceramic waste aggregate for internal
curing of high-performance concrete. Cement Concrete Research 39(5), 373–381 (2009).
5. Bui, P.T., Ogawa, Y., Nakarai, K., Kawai, K.: Internal curing of Class-F fly-ash concrete
using high-volume roof-tile waste aggregate. Materials and Structures 50, 203 (2017)
https://doi.org/10.1617/s11527-017-1073-z
Investigation of the use of reclaimed asphalt pavement as
aggregates in roller compacted concrete for road base
pavement in Vietnam

Thi Huong Giang Nguyen1, Tien Dung Nguyen², Trung Hieu Tran1, Van Dong Dao1,
Xuan Cay Bui2 and Mai Lan Nguyen3[0000-0001-8966-5209]
1
University of Transport Technology, Trieu Khuc, Hanoi, Vietnam
giangnth@utt.edu.vn, hieutrantrung@utt.edu.vn, dongdv@utt.edu.vn
2
University of Transport and Communications, Cau Giay, Hanoi, Vietnam
nguyen.tiendung@utc.edu.vn, bui.xuancay@utc.edu.vn
3
IFSTTAR, Allée des Ponts et Chaussées. Bouguenais, France
mai-lan.nguyen@ifsttar.fr

Abstract. When reinforcing existing asphalt pavement while the pavement ele-
vation must be maintained, old surface asphalt layers are generally milled to
apply the new ones. The use of the reclaimed asphalt pavement (RAP) recov-
ered from road deconstruction is very important. It is an actual need due to the
rise of asphalt cost and the lack of natural aggregates. It respects also sustaina-
ble development. On the other hand, pavement rutting is the most issue in Vi-
etnam. In order to reduce this issue, it is necessary to find efficient solutions.
Roller compacted concrete (RCC), with well-known advantages (simple, eco-
nomical and high stiffness modulus), is found as a very promising technique.
For these purposes, a science and technology project has been granted by the
Ministry of Transport of Vietnam. It aims to investigate the use of RAP as ag-
gregates in roller compacted concrete (RCC) for road base pavement. In this
project, RAP from two different resources is selected and their characteristics
are evaluated. They are then used for the RCC mix design with three different
RAP contents (0, 40 and 80% in masse of aggregates) and two different ce-
ments (PCB30 and PC40). Specimens are fabricated in laboratory to determine
traditional mechanical properties (compressive strength, tensile splitting
strength and elastic modulus) according to their curing periods. The results on
one RAP show that the studied mixes have potential performances and can be
used for road base pavement. Finally, one mix has been chosen to be used for
construction of a full-scale experimental pavement.

Keywords: Reclaimed Asphalt Pavement, Roller Compacted Concrete, Labora-

tory Tests, Experimental Site.

1 Introduction
The solution of removing or utilizing Reclaimed Asphalt Pavement (RAP) is an im-
portant issue in pavement construction and management. Since 1970s, the need to
reuse RAP has become practical due to the increase in asphalt cost and the scarcity of

© Springer Nature Singapore Pte Ltd. 2020 513

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_80
514 T. H. G. Nguyen et al.

natural aggregates. These economic and environmental issues have promoted the
development of technologies for producing asphalt mix using RAP aggregates. How-
ever, most of these recycling processes use only about 20 - 35% of RAP in the mix.
More effective use of this material resource needs to be considered. On the other
hand, in order to reduce pavement rutting which is the most issue in asphalt pavement
in Vietnam, Roller compacted concrete (RCC), which has well-known advantages
(simple, economical and high stiffness modulus), is found as a very promising tech-
nique. For these purposes, a science and technology project has been established. It
aims to investigate for the first time in Vietnam the use of RAP as aggregates in roller
compacted concrete (RCC) for road base pavement.
The following sections present the selection of RAP aggregates and their properties
characterization, the mix design and testing results for one RAP resource and a de-
scription of the experimental site.

2 RAP selection and their properties characterization

2.1 RAP selection and preparation procedure
RAP aggregates are selected from two different resources:

 RAP1 is selected from Tien Son – Bac Ninh industrial park.

 RAP2 is selected from Phap Van – Cau Gie highway.

In order to design 0/12.5 mix, before characterizing these RAP aggregates, they are
prepared through a systematic procedure which includes crushing, screening and clas-
sification stages according to particle sizes (RAP for diameter d ≤ 12.5 mm; LA: large
aggregates for 4.75 mm ≤ d ≤ 12.5 mm; SA: small aggregates for d < 4.75 mm).

2.2 Characterization of RAP properties

Different properties of RAP are determined, including grading, binder content, dry
density, water absorption and impurity content. This characterization is very im-
portant for the mix design process in the next step. Results are presented in Table 1.

Table 1. Properties of two studied RAP aggregates.

(RAP: d ≤ 12.5 mm; LA: 4.75 mm ≤ d ≤ 12.5 mm; SA: d < 4.75 mm)
Properties RAP1 RAP2
RAP1 LA SA RAP2 LA SA
Binder content (%) 3.03 1.89 3.53 4.42 3.53 5.76
Dry density (g/cm3) 2.51 2.56 2.53 2.57
Water absorption (%) 0.8 1.0 0.6 0.6 0.7 0.4
Impurity content (%) 2.3 1.8 3.2 1.9 1.1 2.5
Investigation of the use of reclaimed asphalt pavement … 515

Grading. It is determined according ASTM C33 specifications. For each type of ag-
gregates, a mean value is determined from six samples. Results presented in Fig. 1
show that both RAP aggregates do not meet the requirement of grading according to
design guidelines for RCC. They have many coarse aggregates and less fine elements.
This result shows that when producing RCC using these two types of RAP, it is nec-
essary to combine with additional new aggregates to control the grading of the mix as
required and that more new fine elements need to be added.

a) 100 b) 100
80 80

Passing (%)
Passing (%)

60 60

40 40
20 20
0 0
0,01 0,1 1 10 0,01 0,1 1 10
Diameter (mm) Diameter (mm)
Max Min Max Min
SA-RAP1 0/4.75 RAP1 0/12.5 SA-RAP2 0/4.75 RAP2 0/12.5
LA-RAP1 4.75/12.5 LA-RAP2 4.75/12.5

Fig. 1. Grading curves of RAP aggregates and grading envelops for: a) RAP1; b) RAP2.

Binder content. For each RAP aggregates size, two samples are used for binder ex-
traction and binder content is determined according to TCVN 8860-2: 2011. The re-
sults show that: for both RAP, LA (4.75/12.5 mm) has the lowest binder content and
SA (0/4.75 mm) has the highest one; RAP2 has higher binder content than RAP1.
RAP2 with 4.42 of binder content is more representative for most of RAP recovered
from highway deconstruction. This result shows similarities with studies in the litera-
ture [1].

Dry density. It is determined according to TCVN 7572-4: 2006.

Water absorption. Water absorption of RAP aggregates is determined according to

TCVN 7572-4: 2006. RAP2 has lower water absorption than RAP1. This result could
be explained by the higher binder content in RAP2 than in RAP1.

Impurity content. Content of impurities in RAP aggregates is determined according

to TCVN 7572-9: 2006. Impurity content of small aggregates is higher than large
aggregates. RAP1 aggregates have a higher content of impurities than RAP2. Impuri-
ty contents of RAP1 and RAP2 aggregates satisfy the technical requirements of con-
crete aggregates according to TCVN 7570: 2006.

Considering the limited length, only results with RAP2 are presented here after.
516 T. H. G. Nguyen et al.

3 RCC mix design and testing

3.1 Mix design
Three RAP2 contents are studied: 0%, 40% and 80% by weight of aggregates (respec-
tively 35% and 69% of aggregates + hydraulic binder mass). The aggregate fractions
of RCC mixes are adjusted to achieve the required gradation (Fig. 2a).
Two types of cement are used in this study: PCB30 and PC40. The cement content
was fixed at 13% in mass of granular part for all mix proportions. The optimum water
content was determined based on the maximum density criterion from modified proc-
tor tests (ASTM D1557) at various water contents (Fig. 2b).
The proportions of RCC mixes which determined on the basis of the optimum wa-
ter content are given in Table 2.

a) 100 RCC-0%RAP b) 2.4

RCC-0%RAP
80 RCC-40%RAP RCC-40%RAP
Dry density (Mg/m3)

RCC-80%RAP
Passing (%)

2.3 RCC-80%RAP
60 Min
Max
40
2.2
20

0 2.1
0,01 0,1 1 10 3 4 5 6 7 8 9
Diameter (mm) Water content (%)

Fig. 2. Design of RCC mixes: a) Gradation of aggregates for the RCC mixes; b) Dry density –
water content curves of mixes.

Table 2. Different RCC mix proportions

Component RCC-0%RAP RCC-40%RAP RCC-80%RAP

Coarse LA-RAP2 (4.75/12.5) 0% 26% 30%
Fine SA-RAP2 (0/4.75) 0% 9% 39%
Crushed gravel (4.75/12.5) 44% 13% 0%
Alluvial sand (0/4.75) 35% 30% 9%
Cement (PCB30 or PC40) 13% 13% 13%
Fly ash 9% 9% 9%
Water content 6.0% 6.0% 6.3%
Water/cement ratio 0.46 0.46 0.48

3.2 Mix testing

Cylindrical specimens of 150 mm diameter and 300 mm height were prepared to per-
form the mechanical tests (compressive strength, elastic modulus and tensile splitting
strength) based on ASTM C39, ASTM C496 and ASTM C469 standards, respective-
Investigation of the use of reclaimed asphalt pavement … 517

ly. The specimens were demoulded after 24 hours, then cured under water at 27±20C
and tested at 7 and 28 days. Results are presented in Fig. 3.

35 35 RCC-PC40-7 days
Compressive strength (MPa)

RCC-PC40-7 days
30

Elastic modulus (GPa)

30 RCC-PC40-28 days RCC-PC40-28 days
25 RCC-PCB30-7 days 25 RCC-PCB30-7 days
RCC-PCB30-28 days RCC-PCB30-28 days
20 20
15 15
10 10
5 5
0 0
0 20 40 60 80 0 20 40 60 80
a) RAP2 content in mass of aggregate (%) b) RAP2 content in mass of aggregate (%)
3.5
RCC-PC40-7 days
Tensile splitting strength (MPa)

3 RCC-PC40-28 days
2.5 RCC-PCB30-7 days

2 RCC-PCB30-28 days

1.5
1
0.5
0
0 20 40 60 80
c) RAP2 content in mass of aggregate (%)
Fig. 3. Performances of RCC mixes tested at 7 and 28 days: a) Compressive strength;
b) Elastic modulus; c) Tensile splitting strength.

Compressive strength (Rc). The influence of the RAP content appears clearly for all
mixes. The compressive strength decreases considerably when increasing the RAP
content (Fig. 3a). Moreover, the cement type has also a significant impact on Rc. Rc
increases when PCB30 cement was replaced by PC40 cement.

Elastic modulus (Ec). The same conclusion as for the Rc can be formulated: incorpo-
rating RAP in RCC decreases the elastic modulus (Fig. 3b). The cement type affects
also the elastic modulus. The replacement of PCB30 cement by PC40 gives an in-
crease of about 14-22% of Ec at 7 and 28 days.

Tensile splitting strength (Rt). RAP content and cement type have the same effect
on Rt of RCCs as that presented for Rc and Ec (Fig. 3c). The results show that Rt de-
creases with incorporation of RAP and increases with replacement of PCB30 cement
by PC40 cement.

4 Experimental site
Based on mix testing results, the mix using 40% RAP2 and PC40 cement was chosen
for construction of an experimental site. The last one is situated at Vinh Yen city on a
dedicated site for experimenting new pavement technologies. It has 3.5m wide and
518 T. H. G. Nguyen et al.

20m long. During the construction, different measurement were made including vol-
umetric density, moisture content, degree of compaction and bearing capacity above
foundation and subgrade (Fig. 4a). RCC material was laid and compacted according
to the calculated number of compaction passes (Fig. 4b). After curing periods of 7, 14
and 28 days, samples will be drilled in order to test mechanical properties. Falling
Weight Deflectometer (FWD) measurement will be realized for monitoring of pave-
ment performance.

Fig. 4. Construction of experimental site: a) Subgrade bearing capacity; b) Compaction of RCC

5 Conclusion
Two RAPs have been recovered and prepared for the fabrication of RCC mixes. The
characterization of RAP properties show that it is possible to make RCC using the
aggregates from these RAP with additional natural aggregates in small size part. Me-
chanical test results show that all tested mechanical properties (compressive strength,
elastic modulus and tensile splitting strength) decrease with incorporation of RAP
aggregates and increase with replacement of PCB30 cement by PC40 cement.
The construction of the experimental site shows the feasibility of recycling of RAP
material in RCC. The evaluations of the experimental pavement using FWD meas-
urements, which will be carried out in the following months, will allow controlling in-
situ performance of the pavement made of the investigated RCC-RAP material.

References
1. Cosentino, P.J., Kalajian, E.H., Bleakley, A.M., Diouf, B.S., Misilo, T.J., Petersen, A J.,
Krajcik, R.E., and Sajjadi, A.M.: Improving the Properties of Reclaimed Asphalt Pave-
ment for Roadway Base Applications. Florida Institute of Technology, Melbourne. (2012).
2. D. Yuan, S. Nazarian, L. R. Hoyos and A. J. Puppala: Cement Treated RAP Mixes for
Roadway Bases, Texas Department of Transportation in cooperation with Federal High-
way Administration. (2010).
3. Kevin Bilodeau, et al.: Effect of reclaimed asphalt pavement on cement treated materials
for road base layer 5th Eurasphalt & Eurobitume Congress, Istanbul. (2012).
 Comparison between Critical Path Method (CPM) and
Last Planners System (LPS) for Planning and Scheduling
METRO Rail Project of Ahmedabad

Viraj Parekh1 , Karan Asnani1, Yashraj Bhatt1 and Rahul Mulchandani1

1 Civil Engineering Department, Institute of Technology, Nirma University, Ahmedabad, India

Abstract. The purpose of this paper is to compare Critical Path Method (CPM) and Last
Planner System (LPS) with respect to Planning and Scheduling of METRO Rail Project,
Ahmedabad, Gujarat. Critical Path Method emphasises on updating the network for
tracking the progress as well as to identify the delays. Last Planner System works on the
weekly schedules prepared from the Master Plan and Look-ahead schedules to avoid the
delays. One of the stretch from North-South Corridor was selected for the study from
Vijaynagar to Usmanpura. The data such as activities, duration of activities, sequence
and inter-relation of activities etc. were collected to prepare the network as well as weekly
schedules. The network was updated and original network was compared with the up-
dated one and the delays were spotted for the stretch selected. Weekly plans were also
prepared for the selected stretch from the look-ahead schedule and Master Plan. PPC
(Percent Plan Complete) were calculated to track the progress as per planned schedule.
The data were collected by conducting interviews of various personnel and visual obser-
vations. Both the approaches (CPM and LPS) have been applied on the selected stretch
by action research process. The delays were calculated and studied for both the methods
and it was observed that Last Planner System is more appropriate to use for big infra-
structure projects like this to avoid time-overrun and consecutive cost over-run. Re-
sources can be well utilized with the Last Planner System, too. The type of this paper is
a Case Study.

Keywords: Construction Planning and Scheduling, Last Planner System, Con-

struction Project Management

1 Introduction

Planning and Scheduling is the back-bone for any Construction Project as execution
is purely based on the planning and scheduling. If Planning and Scheduling is done
haphazardly then execution of the project will be affected and the project would observe
time overrun and consecutive cost overrun. Critical Path Method (CPM) and Program
Evaluation and Review Technique (PERT) are being used for planning and scheduling
in the construction industry since last 5 decades. Over the period of time, certain limi-
tations of CPM and PERT was observed. Though the network representation in CPM
and PERT is an integral part of Project, Last Planner System aids in better clarity about
the work on executional basis.

© Springer Nature Singapore Pte Ltd. 2020 519

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_81
520 V. Parekh et al.

A Network is created in Critical Path Method connecting all the activities. Forward
Pass and Backward Pass gives the Critical Path of the Network. All the activities lying
on Critical Path are considered as Critical Activities; having float value zero. Rest of
all the activities are considered to be non-critical activities. When the project is not
executed as per the planned schedule, Critical Path of the Network may get changed.
In that case, certain activities may become critical which were non-critical earlier.
Change in Critical Path and Critical Activities disturb the entire planning and schedul-
ing with respect to time, resources, cost, etc. and it can be considered as the major
limitation of Critical Path Method. Once the original schedule gets disturbed, it is very
difficult to bring back on track and the project observes time overrun and cost overrun.
Last Planner System (LPS) was developed to increase the effectiveness of planning
and control by making programmes more predictable, thereby improving the chances
of delivering them on time [1]. The LPS has been implemented in a large number of
projects in several countries since 1992 [2]. Many reports and research papers have
confirmed that the technique has achieved remarkable improvements, including better
planning and control, improved work flow reliability, increased productivity, promo-
tion of team building; improved quality and safety, enhanced predictability, thereby
reducing the duration and cost of projects (e.g. [3], [4], [5], [6], [7] and [8]). The liter-
ature, however, shows no evidence of LPS practical applications within India as well
as on METRO Rail Projects. Therefore, this study is concerned with the application of
LPS in METRO Rail Project of Ahmedabad in the state of Gujarat, India. The study
also compares the traditional method of planning and scheduling CPM with LPS re-
garding the METRO Rail Project of Ahmedabad.
Ahmedabad Metro is an under construction mass transit system being built for the
Ahmedabad City. It is being built by Gujarat Metro Rail Corporation (GMRC) Limited;
a Special Purpose Vehicle constituted to implement all the Metro Projects in the state
of Gujarat. The total length of the Ahmedabad Metro Rail Project (Phase-I) is about
40.03 km out of which 6.35 km is underground tunnel construction and the rest is ele-
vated construction. The project will connect 4 corners of the Ahmedabad city with 2
corridors (North-South and East-West) and 32 stations.

2 Methodology
Feasibility study for the Ahmedabad METRO Rail Project was done two times.
Therefore, since beginning of the project, delays were observed. In such situation, plan-
ning and scheduling of the project must be thorough to reciprocate the delays. To pre-
pare the planning and scheduling of METRO Rail Project, data collection was done at
the first place. Data was collected by interviewing respective personnel of METRO Rail
Project. List of Activity, Duration of Activity and Interdependency was collected for
the stretch of Vijaynagar to Usmanpura in North-South Corridor. Based on the collected
data presented in Table 1, a network was prepared (Fig.1). Duration was calculated
from the Network and that came as 42 days. The Network was updated after 28 days
and updated network was also created. By performing forward pass and backward pass
in the updated network, the duration on critical path was calculated and it came to be
Comparison between Critical Path Method (CPM) … 521

56 days. A comparison was done between planned work and actual work based on the
network prepared and updated. Through this comparison, delays were rectified.

Table 1. Collected data for the stretch of Vijaynagar to Usmanpura

Nota- Name of the Activity Dura- Immedi-
tion tions ate Pre-
(Days) ceding
Activity
A. Point Establishment for Piles & 1 -
Piers
B. Excavation of Pile under Pier 5 A
No.159&160
C. Reinforcement & Concreting of 5 B
Piles (Total &Piles)
D. Crib Arrangement (Welding, Cut- 2 -
ting etc.)
E. Crib placement near Pier No.158 3 D
F. Reinforcement for Cross Arm No.1 3 E
G. Formwork for Cross Arm No.1 1 F
H. Concreting for Cross Arm No.1 1 G
I. Base Preparation for Pile Cap 2 C
J. Reinforcement for Pile Cap 2 I
K. Formwork for Pile Cap and Con- 2 J
creting
L. Reinforcement Preparation for Pier 2 J
159-160
M. Formwork and Concreting work for 2 L
Pier No.159-160
N. Excavation of piles under pier 159- 2 C
160
O. Reinforcement and Concreting of 2 N
Piles (4 Piles)
P. Excavation of Pile No.163 3 O
Q. Reinforcement and concreting (6 2 P
Piles)
R. Pile Cap for Pile No. 161&163 7 Q
S. Pier Construction for 161&163 8 R
T. Cross Arm No.2 & Reinforcement 4 S
Preparation at 163
U. Formwork and Concreting of Cross 3 T
Arm No.2
V. Pier Cap Formwork Arrangement 2 M
W. Pier Cap Reinforcement and Con- 2 V
creting
522 V. Parekh et al.

X. Pile Load Test (Survey & Setup) 5 O

Y. Setup Inspection and implementing 4 X
required changes
Z. Reporting of Test Results and Un- 2 Y
loading the Pile Loads

Fig. 1. Network of planned schedule.

Considering the principles of Last Planner System, Weekly Work Plans were prepared
from the Look-ahead Plan and Percent Plan Complete (PPC) was calculated for each
week as well as each day (total 42 days). Table 2 represents Weekly Work Plan and
PPC calculation for week 1. Weekly Work Plan and PPC for all 6 Weeks are created
and calculated as attached in Annexure-I.
Weekly PPC can be maximum 100% in the ideal condition i.e. 100% for 7 days. For
calculating PPC on daily basis, Weekly PPC was divided by 7 i.e. 100/7 = 14.28% PPC
for a single day. PPC can be calculated as;
Completed Work
Percent Plan Complete (PPC) = x 100%
Planned Work

3 Data Analysis
Data was analysed by comparing the actual schedule planned and actual execution hap-
pened on the site. By comparing this, the delays have been identified and presented
with PPC. Delays were calculated for both weekly basis as well as daily basis. For the
weekly delays, value of the weekly PPC is less than 14.28%. For the daily delays, value
of PPC is 0% for the activities yet to be started.
Comparison between Critical Path Method (CPM) … 523

Table 2. Weekly Work Plan and PPC Calculation for Week 1

Daily P.P.C.
DAY Activity (planned) for Week 1 Activity (executed) for Week 1
(In %)
1 1. According to plans and drawings 1. All are As planned 14.28
establish the centre points of piles
& pile caps for pier no.- 159 &160
2. Rig machine preparation
3. Pier Arm casting completed(curing
and
finishing)
2 1. Excavation of pile 1&2 under pier 1. Material acquiring 10.71
no. 159 & 160 for crib is delayed
2. Reinforcement preparation of piles 2. Other as planned
and Reinforcement lowering
3. Pile concreting
4. Material acquiring for cribs
3 1. Excavation of pile 3 & 4 under pier 1. Excavation of Pile 4 not 4.71
no.159 & 160 done due to problem
2. Reinforcement preparation of piles in Rig Machine and
and Reinforcement lowering other activities also
3. Pile concreting delayed because they
4. Crib material arrangements (weld- are interdependent
ing and Cutting etc.)

4 1. Excavation of pile 5 & 6 under pier 1. Material Acquired for 7.14

no.159 & 160 Crib arrangement Started
2. Reinforcement preparation of piles 2. Repairing of Rig
and Machine
3. Reinforcement lowering
4. Pile concreting
5. Crib placement near to pier
5 1. Pile cutting excluded portion 1. Pier arm preparation only 7.14
2. Side slope preparation for pie cap others are delayed
3. R/F preparation for pier arms
6 1. P.C.C for pile cap base 1. Pier arm preparation only 7.14
2. Formwork preparation for pier others are delayed
arms
7 1. Formwork preparation for pile cap 1. Pier arm preparation only 4.00
2. R/F preparation for pile cap others are delayed

Remarks : Out of 6 piles only 3 piles completed

Weekly PPC 55.12
524 V. Parekh et al.

4 Result and Discussion

A stretch from Vijaynagar to Usmanpura was selected in the METRO Rail Project of
Ahmedabad to compare the planning and scheduling based on CPM and LPS. Actual
data was collected from the site and compared both the methods with respect to time.
After the analytical work, these conclusions were made. The schedule was prepared for
42 days but it was delayed and extended up to 56 Days. A pattern was observed in the
PPC of every week that it is decreasing from week 1 to week 4. Weekly PPC was 0%
for Week 5 and Week 6.

5 Conclusion
If the project would have been executed based on the principle of LPS then this kind of
delays would have been rectified from week 1 itself and project would be saved from
time overrun and cost overrun. Table 3 summarize the conclusion by comparing CPM
and LPS with criteria of Planning and Scheduling.

Table 3. Comparison of CPM and LPS

Criteria for better Planning and Scheduling
judgement CPM LPS
Delays identified After 4th week On 1st Week
Weekly delays No Yes
available
Daily delays avail- No Yes
able
Monitoring the Tedious for doing Smooth for doing
Progress on weekly basis on weekly basis

References
1. Ballard, G. (2000), “The Last Planner System of production control”, PhD thesis,
School of Civil Engineering, University of Birmingham, Birmingham.
2. Ballard, G. and Howell, G. (2003), “An update to Last Planner”, Proceedings of the
11th IGLC Conference, Blacksburg, VA.
3. Ballard, G., Howell, G. and Casten, M. (1996), “PARC: A case study”, Proceedings of
the 4th IGLC Conference, Birmingham, reprinted in Alarcon (1997).
4. Garnett, N., Jones, D. and Murray, S. (1998) “A strategic application of lean thinking”,
Proceedings of the 6th IGLC Conference, Brazil.
5. Junior, A., Scola, A. and Conte, A. (1998), “Last Planner as a site operations tool”,
Proceedings of the 6th IGLC Conference, Guaruja, Sao Paulo.
6. Ballard, G. (2000), “The Last Planner System of production control”, PhD thesis,
School of Civil Engineering, University of Birmingham, Birmingham.
7. Fiallo, C. and Revelo, V. (2002), “Applying LPS to a construction project: a case study
in Quito, Ecuador”, Proceedings of the 10th IGLC Conference, Gramado.
8. Thomassen, M., Sander, D., Barnes, K. and Nielsen, A. (2003), “Experience and results
from implementing Lean Construction in a large Danish contracting firm”, Proceedings
of the 13th IGLC Conference, Blacksburg, VA.
 Study and propose the size of cement concrete slabs for
airport road surface in Vietnam conditions

Nguyen Duy Dong1, Nguyen Vu Viet2 and Nguyen Quoc Van3

1,2,3
Military Technical Academy, Ha Noi, Viet Nam;
1
Nguyenduydong255@gmail.com; 2 Vuducviet6666@gmail.com.

Abstract: Cement concrete slabs (CCS) for road pavement has many ad-
vantages such as high Young modulus and long life expectancy, but this type
of road pavement needs transverse joints (contraction joints, expansion joints
and construction joints) that adversely affect the using process. Recently,
many scientists have been studying solutions to increase the size (length and
width) of cement concrete slabs in order to increase the smoothness of the
road pavement and reduce the amount of raining water penetrating into joints.
In this paper, the authors study and propose solutions to use reinforcement to
reduce the amount of joints on cement concrete pavement which bring a better
using process of this road pavement type.

Keywords: Cement concrete slabs, Road pavement, Slab size.

1. Introduction

The cement concrete pavement (CCP) has been using in almost countries in over the
world for a long time. The United State, in 1913, began constructing a road section
with concrete pavement. In Germany, the CCP was started to construct in the 1880s
and in 1934 it was widely used in the expressway construction. In Vietnam, the CCP
is also commonly used in airports.
Together with the advantages, during the operation process, this type of pavement
surface has the disadvantages of making the passengers very uncomfortable when the
means of transport passing through the CCP joints. The contraction joint and expan-
sion joint are very necessary for the pavement surface but they reduce the evenness
and make the passenger uncomfortable when the aircrafts or means of transport oper-
ating on the pavement.
This fact has been posing the necessary question how to minimize the total length of
contraction joints and expansion joints for the airport CCP.

2. CCP in the aiports of Vietnam

By 2019, the Vietnam Aviation Industry (VAI) have managed and operated 22 air-
ports. Currently, most the airports use CCP type at the RWY, taxiways and aprons.

© Springer Nature Singapore Pte Ltd. 2020 525

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_82
526 N. D. Dong et al.

Since the airports have been constructed and used for a long time, most airports have
to be upgrade and repaired such Cat Bi, Noi Bai, etc. In order to meet the demand for
operation, many new airports have been constructed, such as Phan Thiet, Van Don
(Quang Ninh), etc. The airports to be upgraded, repaired and constructed are designed
and executed domestically in accordance with the Vietnamese standards [1], ICAO
and current standards. The common slab dimensions are 4 × 5 and 5 ×5 (m) with a
thickness of 30 – 35 cm for CCP. For the slab with a thickness of 36 – 40 cm, used as
reinforced concrete (RC), some airports already used the slab with a thickness of 4.5
× 10 (m) or 5 × 10 (m). CC slab is usually placed on a cement stabilized base layer
with a sand bedding layer (fig.1.).

1
1
2
2
3 3
4
4
5
5

6 6 =

7 7 =

Fig. 1. Typical RC structure with grade of M350/45 at A area.

3. Calculation of contraction joint and expansion joint spacing in

airport concrete pavement

The increase or decrease in the average temperature in the slab causes expanding or
contracting of concrete slabs, and while the different surface and bottom temperature
will cause bending of the slab. When the deformation of the concrete slab is hindered
by the temperature field uniformly distributed in the depth of the slab or horizontal
displacement is prevented by friction between the slab bottom and the base, and the
temperature stress will occur in the slab, according to the linear theory of thermo elas-
ticity we have:
 T   t , z   ms (1)

where:  T - temperature stress in the depth z in the slab;

 t, z - temperature stress due to the temperature uniformly distributed by the slab depth;
 ms - temperature stress due to the friction between the concrete slab and the
base, when using a separation layer, this stress may be omitted.
3.1 Calculation of the temperature stress due to the temperature uniformly dis-
tributed in the slab depth
Study and propose the size of cement concrete slabs … 527

When the temperature of the slab surface is greater than the temperature of the slab
base (at the daytime), the compressive stress will appear on the slab surface, the ten-
sile stress will appear at slab base, on the contrary, when the temperature of the slab
surface is smaller than the temperature of the slab base (at night), the tensile stress
will appear on the slab surface, and the compressive stress will appear on the base
slab.
An approximate formula is given in Branch Standard 22TCN 223 – 95 for calculated
the temperature gradient linearly distributed as per the concrete slab depth: t= 0.84h
depends on the slab thickness, where h is the concrete slab thickness (in China, it is
required to take the value within t = 0.82 0.85 h).
The upward bending temperature stress at the slab edge and slab centre can be calcu-
lated [2], [3], [4], [5]:
1
 c t  C x .Et . .t ;
2
Et
 ot  (C x  C y ). t ,
2(1   2 )
 Et t (2)
t  ,
2(1   2 )
where:
ct – temperature stress at slab edge; ot- temperature stress at slab centre;
t – temperature stress at the slab center when the restrained slab is completely
deformed;  - concrete coefficient of thermal expansion,  = 1.10-5;
Et - modulus of elasticity of concrete;  - Poisson's ratio for concrete;
t – temperature gradient (approximately taken with t = 0.84h);
Cx, Cy – coefficient depending on slab edge sizes (see fig. 2. ):
2cos.cosh  B ; B, L - slab edge size (C ,
Cx,y  1  (tg  tgh), with   x
sin2  sinh2 L 8
Cy can be taken as per the chart).

Fig. 2. Chart for determination of Cx, Cy coefficients

L0 – elastic property of concrete slab; L  4

Ebt h3 ; with C - base coefficient.
0
12C
528 N. D. Dong et al.

3.2 Calculation of temperature stress due to friction of slab bottom

When the average temperature of the concrete slab decreases, the slab will be con-
tracted, and when the temperature increases, expansion of slab will occur. In the pro-
cess of contraction and expansion like this, if friction exists too much between the
bottom of the concrete slab and the base surface, there will be frictional force restrict-
ing the slab movement.
For concrete slabs with a free edge, the movement due to temperature change tends to
shift to the centre of the slab when the temperature decreases, and shifts from the
centre of the slab when the temperature increases. For the structure of the multi-
layered concrete slab, it is necessary to arrange a separation layer between concrete
layers to reduce friction between concrete layers when they move due to temperature
change, so if the actual friction between layers is small, and it can be omitted in the
calculation.
When the temperature changes per the seasons as decreases or increases, the concrete
slab is contracted or expanded. During the movement process of the concrete slab,
frictional force between the surface of the lower layer and the bottom of the concrete
slab will cause the tensile stress and eccentric compression with the deviation e, cal-
culated by formula (3):
T Mx T 6e (3)
 
t  e (1  )
Bh Jx Bh h
where:
T – frictional force of the slab bottom, equal to half of the weight of the slab
multiplied by friction coefficient f.
B – With of concrete slab, cm; L – Length of concrete slab, cm;
h – thickness of concrete slab, cm;  - density of concrete,  = 0.0025kg/cm3;
f – friction coefficient between concrete slab and base layer.
At the middle depth e = h/2, the eccentric tensile stress will be:
T
t  (1  3)
Bh
Frictional force of the slab bottom, and this for is born by reinforcement steel:
T = Fa Ra ma
where: Fa – reinforcement steel area; Ra – calculated strength of reinforcement steel;
ma – work condition factor of reinforcing steel equal to 0.9 – 0.95
Calculated for unfavourable case when the slab is contracted, the tensile stress at the
bottom of the plate is:
4T (4)
 
t  2 L f
Bh
L  [] / (2f) (5)
where: f – friction coefficient, and its values are taken as per Table 1.

Table 1. Friction depending on type of base materials

Type of bases Thickness of Shear strain Friction coeffi-
concrete slab limit, mm cient f
Sand 30 0.8 – 1.8 0.9
Study and propose the size of cement concrete slabs … 529

45 1.8 – 2.2 0.85

Clay soil 30 1.0 1.6
Layer separated 30 1.7 – 2.0 1.3
with bitumen sheet 45 2.0 1.1
Crushed stone 30 4.0 2.4

3.3 Calculation example

Calculated data and designs are taken as follows

- Type of hydrogeological condition: Type 2, [1];

- Climate Region for airport pavement design: Region II, [1];
- Type of subgrade: Fine sand layer: K = 8 (kG/cm3), E = 100 (Mpa);
- Type of calculated aircraft: of B747 - 400ER aircraft.

The pavement structure is as follows. M350/45 RC layer, bar diameter of D = 18

mm; M150/25 layer, 20cm thick; Crushed-stone aggregated layer, 30 cm thick; Sub-
grade K = 8 (kG/cm3). Studied for the slab with the thickness of 30cm to 40 cm as
shown in Table 02.

Table 02. Calculated results of concrete slabs according to their thickness

Concrete slab Elastic properties of Length of RC Recommended length
C1 – C2
thickness concrete slab (cm) slab (m) of RC slab (m)
30 87.7 0.25 – 0.33 10.4 – 13.8 10.00
32 90.5 0.23 – 0.31 11.1 – 14.7 10.50
34 93.3 0.22 – 0.29 12.2 – 16.3 11.50
36 96.1 0.20 – 0.27 12.7 – 16.9 12.0 – 13.0
38 98.8 0.19 – 0.26 13.7 – 18.3 12.0 – 14.0
40 101.5 0.18 – 0.25 14.9 – 19.9 12.0 – 15.0

4. The contraction joint with the spacing of 10m that have been ap-
plying at some airports

In 2004 - 2005, when designing to upgrade A3 apron (Noi Bai International Airport)
into a heavy load bearing apron, AEC Company used reinforced concrete slabs of 38
cm thick with a slab size of 5x10 m. Next, in the following years, some airports in
Vietnam have already used the RC pavement with a size of 5×10 m or 5.5 ×10 m in
the areas with Category A sub - areas (heavy load bearing), such as 150 m2 at the
beginning of the RWYs, the taxiway areas, the taxi tracks on apron A1 of Noi Bai
ARPT (Fig.3.), etc. Currently, Van Don ARPT is being constructed, and the RC
pavement with slab sizes of 4.5x10 m, 5x10 m and 5.5 x10 m is also used (Fig.3.).
530 N. D. Dong et al.

B
FLAT OF 1/ 2 STEEL BELOW FLAT OF 1/ 2 STEEL ABOVE

SECTION B - B

A A

SECTION A - A

Fig. 3. Layout plan of reinforcement steel of 5×10 m RC slab for taxi track of apron A1

5. Conclusion

After calculating the length of RC pavement with the pavement thickness of 30cm to
40cm as shown in Table 2, and in combination of the actual monitoring situation in
the past years regarding the application of the airport pavement when using the slabs
with large sizes (CC slab with reinforcement steel), currently the slabs still operate
perfectively without any damages. Simultaneously, thanks to the development of the
current construction technology, we recommend that the joint space of 12 – 15 m on
RC panels for 36 – 40 cm thickness slabs. In addition, the slab width, depending on
the paver capacity and actual project condition, of 5 m to 6 m should be selected.

References
1. TCVN 10907 – 2015, Civil Aerodrome -Pavement - Specifications for Design (2015).
2. Kazda, Antonín, and Robert E. Caves, eds. Airport design and operation. Emerald Group
Publishing Limited, (2010).
3. Nguyen Duy Dong, Studying the influence of contraction joint on bearing capacity of cement
concrete slab of airport pavement, Research topic, level of Institute of Military Technology,
Hanoi (2003).
4. Packard, Robert G. Design of concrete airport pavement, 1973.
5. Decision No. 3230/QĐ-BGTVT dated 14/12/2012 of the Ministry of Transport on temporari-
ly stipulating jointed cement concrete pavement in construction of transport works.
 Experimental study of the thermomechanical behavior of
the carbon textile reinforced refractory concrete
subjected to the constant load and temperature heating

Manh Tien TRAN 1,2, Xuan Hong VU 1 and Emmanuel FERRIER 1

1
Université de Lyon, Université Lyon 1, Laboratory of Composite Materials for Construction
(LMC2), 82 bd Niels Bohr, F-69622 Villeurbanne, France
2
Hanoi University of Mining and Geology (HUMG), Material Resistance Depart, 18 Pho Vien,
Bac Tu Liem, Hanoi, Vietnam

Abstract. The carbon textile reinforced refractory concrete is an alternative ma-

terial to strengthen or reinforce the structure in an elevated temperature envi-
ronment or usually subjected to fire. This composite is one perfect combination
of carbon textile reinforcement which ensures high capacities (mechanical
strength and stiffness) at elevated temperature, and refractory concrete matrix
which ensures thermal protection and loading retransmission into textile rein-
forcement. The aim of this paper is to identify the rupture temperature and ex-
posure duration of carbon TRC in thermomechanical condition. This paper pre-
sents original experimental results on carbon TRC specimens subjected to 5 ap-
plied force levels (from 10% to 75% in comparison with the maximal strength
of this material at room temperature) and temperature increasing. As results, the
studied carbon TRC specimen can exhibit up to the temperature of 670 °C in 23
minutes corresponding with an applied force level of 50%. The rupture temper-
ature and exposure duration of carbon TRC decreased progressively with in-
creasing of applied force level. In comparison with the experimental result on
carbon textile specimen, the contribution of the refractory matrix as thermal
protection was found and analyzed. The exposure duration of the carbon TRC
in case of fire could be calculated thanks to the temperature heating curve ac-
cording to the standard ISO – 834.

Keywords: Textile - Reinforced Concrete (TRC), Carbon Textile, Refractory

Concrete, Elevated Temperature, Thermomechanical Behavior.

1 Introduction
The textile-reinforced concrete composite (TRC) is an alternative material to
strengthen or reinforce the civil engineering structures. This composite is one reason-
able combination between the textile reinforcement and the cementitious matrix. The
textile reinforcement ensures TRC mechanical strength, to against the weakness and
cracking sensitivity of cementitious matrix thanks to its high strength and Young’s
modulus. On the other hand, the cementitious matrix ensures loading retransmission
between the textiles and protection from the environment. In case of fire or exposure
to high temperature in the civil engineering structures, the high-temperature resistance
of TRC composite is an important factor for structure strength durability. In TRC

© Springer Nature Singapore Pte Ltd. 2020 531

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_83
532 M. T. Tran et al.

material, fire resistance of textile reinforcement influenced greatly on that of TRC.

With high strength and Young’s modulus, the reinforcement by using the carbon tex-
tile was performed successfully in several previous research [1-2]. Furthermore, the
contribution of the cementitious matrix as thermal protection is also important for the
fire resistance of TRC composite. Refractory matrix is the perfect choice for fire ap-
plication. This matrix is based on calcium aluminate cement (CAC) [3] or other ce-
ment with high alumina content [4]. The refractory matrix has been used successfully
to manufacture the basalt TRC composite for elevated temperature application in [5].
So, it would be interesting to combine carbon textile with a refractory matrix to manu-
facture a good TRC for fire application. This paper presents an experimental study on
thermomechanical behaviour of carbon TRC subjected to constant tensile loading and
temperature increasing (as in case of fire). Firstly, carbon TRC specimens were tested
in tensile at room temperature in order to identify its resistance. Then, the carbon
TRC specimens were subjected to 5 applied force levels (from 10% to 75% in com-
parison with its strength at room temperature) and temperature increase in order to
determine the rupture temperature and exposure duration at high temperature. As
results, the studied carbon TRC specimen can exhibit up to the temperature of 670 °C
in 23 minutes corresponding with an applied force level of 50%. The rupture tempera-
ture and exposure duration of carbon TRC decreased progressively with increasing of
applied force level. In comparison with the experimental result on carbon textile spec-
imen, the contribution of the refractory matrix as thermal protection was found and
analyzed. The exposure duration of the carbon TRC in case of fire could be calculated
thanks to the temperature heating curve according to the standard ISO – 834 [6].

2 Experimental works
2.1 Equipment used
The test machine (TM20kN - 1200°C) has a mechanical capacity of 20kN and
equipped a cylindrical furnace to generate temperature around of specimens. The
furnace can provide the temperature up to 1200°C with a maximum heating rate of
30°C/min. The temperatures in the furnace and on the surface of the test sample are
controlled by thermocouples. The laser sensor equipped on the machine is used to
measure the longitudinal deformation of specimens for the test at room temperature.
This equipment was used in some authors’ previous research [1,7].
2.2 Carbon TRC and specimens
The carbon TRC (called F.GC1 in this study) is based on the refractory matrix and
reinforced by GC1 carbon textile. The refractory matrix consists of a silico-
aluminous-calcic synthetic aggregate (maximum diameter less than 1.25 mm) and the
CAC cement. The high aluminate content of this cement (about 50%) gives good
thermal characteristic to the matrix. The dimensions of F.GC1 specimens are 740 mm
x 65 mm x 11.5 mm (length x width x thickness). The reinforcement ratio calculated
by the volume fraction between the GC1 carbon textile and the composite (Vf/Vtotal) is
0.88%. Table 1 shows the properties of the materials used in this study.
Experimental study of the thermomechanical … 533

Table 1. Materials used for carbon TRC composite

Composition of refractory concrete GC1 Carbon textile

3
Aggregate (kg/m ) 1676 Ultimate strength (MPa) 2617
Cement (kg/m3) 669 Young’s modulus (GPa) 256
3 3
Superplasticizer (kg/m ) 4.34 Density (g/cm ) 3.43
3 Grid geometry (long x trans)
Viscosity modifier agent – VMA (kg/m ) 0.51 46x41
(mm x mm)
Epoxy
Water (kg/m3) 234.2 Type of coating
resin
Water/cement ratio 0.35 Cross-sectional area (mm2) 1.85
2.3 Test procedure
The procedure of thermomechanical test at constant force (TMCF) consists of two
phases as presented in Fig 1. The first is to apply the quasi-static loading monoton-
ically to the specimen up to the applied force level with the loading rate of 200
μm/min. The second is to maintain this force and increase the temperature around the
TRC specimen with the temperature rate about 30 °C/min from room temperature
until the rupture. The temperature corresponding to the rupture point of the test spec-
imen is identified as the rupture temperature (Tr). The time of the second phase is
called the exposure duration at high temperature (see Fig 1).

Fig 1. Loading path for thermomechanical tests in TMCF regime

2.4 Summary of specimens and tests

Table 2 shows the list of the specimens and the tests carried out. There are 18 tests
carried out on the specimens of carbon TRC at room temperature and different load-
ing levels varying from 10% to 75%.
534 M. T. Tran et al.

Table 2. List of tests carried on F.GC1 specimens at room temperature and at TMCF regime

Specimens Dimensions of the specimens [cross Loading levels Number

section, S (mm2); length, l (mm)] (%) of tests
F.GC1 – 25°C – a,b,c 3
F.GC1 – 10% – a,b,c 10 3
F.GC1 – 25% – a,b,c 25 3
S=11x65(mm2); l=740(mm)
F.GC1 – 50% – a,b,c 50 3
F.GC1 – 65% – a,b,c 65 3
F.GC1 – 75% – a,b,c 75 3
Total of tests 18

3 Results
3.1 Test results at room temperature
Fig 2 shows the "stress-strain" curves of the mechanical behavior of F.GC1 at room
temperature. From Fig 2, it can be found that F.GC1 samples give a strain hardening
behavior with three phases as presented in the literature. The average ultimate tensile
strength of F.GC1 specimens is 12.67 MPa, corresponding with a maximum axial
strain of 0.866%. Table 3 shows all values at the typical points of F.GC1 composite at
room temperature.

Fig 2. Mechanical behavior of F.GC1 at room Fig 3. Temperature increasing as a function of

temperature time with different levels of applied force
Table 3. Results of direct tensile tests carried out on F.GC1 specimens at room temperature

Specimens Force Ultimate strength Ultimate strain

(kN) (MPa) (%)
F.GC1 – 25°C – a 9.38 12.88 0.789
F.GC1 – 25°C – b 9.67 13.16 0.899
F.GC1 – 25°C – c 9.19 12.24 0.910
Average 9.41 12.76 0.866
Standard deviation 0.24 0.48 0.066
Experimental study of the thermomechanical … 535

3.2 Test results in the TMFC regime

Fig 3 shows the results of the test performed on F.GC1 specimens in TMCF condition
with different levels of applied force. As results, F.GC1 specimens can subject to
thermomechanical action up to the temperature of 760 °C corresponding with the
imposed force of 1kN (11% compared to its ultimate force at room temperature).
From Fig 3, it can be found that the rupture temperature of F.GC1 gradually decreases
when the imposed force level increases. With an applied force level of 75% (7.1 kN),
its rupture temperature is 274 °C. All values obtained from the TMCF tests are pre-
sented in Table 4 below.
Table 4. Results of thermomechanical tests carried out on F.GC1 specimens in TMCF regime
Applied Force level Rupture tem- Exposure duration Temperature
Specimens
force (KN) (%) perature (°C) (Min) rate (°C/min)
F.GC1 – 11% 1.0 11 759.7 (13.8) 27.9 (1.3) 33.2 (0.5)
F.GC1 – 25% 2.4 25 711.1 (21.2) 24.2 (0.5) 32.8 (0.1)
F.GC1 – 51% 4.8 51 667.6 (23.4) 23.4 (0.4) 33.5 (0.3)
F.GC1 – 64% 6.0 64 424.8 (1.3) 15.1 (0.2) 32.1 (0.4)
F.GC1 – 75% 7.1 75 273.6 (37.3) 10.7 (1.3) 29.2 (0.2)

4 Discussion
Fig 4 presents the evolutions of rupture temperature of F.GC1 depending on the ap-
plied loading level. With the applied force below the critical value of matrix cracking,
the F.GC1 specimens gave good values of rupture temperature and exposure duration.
In comparison with the experimental results on GC1 specimens, an improvement in
fire resistance can be found in Fig 4 with the same applied force level. When the ap-
plied force increases, heat transfer in the matrix happens more quickly, depending on
the appearance, density, and opening of cracks on the specimen corps. So, the contri-
bution of the refractory matrix as thermal protection is decreased clearly. With ap-
plied force level higher than 50%, F.GC1specimens is completely cracked and their
cracks are widened on its corps under tensile force. So, the refractory matrix almost
loses its contribution as thermal protection for the GC1 textile.

Fig 4. Evolution of the rupture temperature as Fig 5. Evolution of the exposure duration at a high
a function of the applied loading level temperature according to the applied loading level
536 M. T. Tran et al.

Concerning the results of temperature exposure duration, the similar evolution of its
value as a function of applied force level could be found. In the comparison of the
experimental results with the temperature curve in ISO-834, it could predict the expo-
sure duration of F.GC1 in case of fire according to that. Fig 5 presents the evolution
of the exposure duration at high temperature as a function of applied force level. It
compares obtained experimental results and predicted results as following the ISO–
834 curve. From Fig 5, it could be observed a difference about 10 min between both
values.
5 Conclusion
As results obtained, some conclusions could be drawn for this work. The reinforced
textile concrete F.GC1 has better capacity in TMCF condition in comparison with
GC1 carbon textile and the Handmade- Carbon Fibre Reinforced Polymer (H-CFRP)
composite [7] thanks to the contribution of the refractory matrix as thermal isolation.
This matrix contribution depends on the appearance, density, and opening of cracks
that are results of mechanical loading corresponding with each level of applied force.
The exposure duration of F.GC1 at high temperature in case of fire could be predicted
from the obtained experimental results.
Acknowledgments
This research was performed with the financial subvention of the European Regional Develop-
ment Fund of the European Union and of the Rhône-Alpes-Auvergne region, France for the lot
5 “Characterization of composite materials” of the PRORETEX II research project. This project
is the collaborative research project between four industrial partners (SULITEC - project lead-
er; FOTIA; ER2I; CIMEO) and two academic partners (UCBL/LMC2; ENISE/LTDS). We also
would like to thank the team of technicians (Mr. JANIN, Mr. COTTET) from the Civil Engi-
neering Department at IUT Lyon 1 and LMC2, University Lyon 1 for their technical support.
References
1. M.T. Tran, X.H. Vu, E. Ferrier, Experimental and analytical analysis of the effect of fibre
treatment on the thermomechanical behaviour of continuous carbon textile subjected to
simultaneous elevated temperature and uniaxial tensile loadings, Construction and Build-
ing Materials 183, 32–45 (2018).
2. S. M. Raoof, D. A.Bournas, TRM versus FRP in flexural strengthening of RC beams: Be-
haviour at high temperatures, Construction and Building Materials 154, 424-437 (2017).
3. W.G. Bareiro, F. A. Silva, E.D. Sotelino, O.F.M. Gomes. The influence of alumina content
on the chemical and mechanical behavior of refractory concretes fired at different tempera-
tures, Constr. Build. Mater 187, 1214–1223 (2018).
4. V.A. Abyzov. Refractory Cellular Concrete Based on Phosphate Binder from Waste of
Production and Recycling of Aluminum. Procedia Engineering 206, 783–789 (2017).
5. D.A.S. Rambo, Y. Yao, F.d.A. Silva., R.D.T. Filho, B. Mobasher. Experimental investiga-
tion and modelling of the temperature effects on the tensile behavior of textile reinforced
refractory concretes, Cem. Concr. Compos 75, 51-61 (2017).
6. ISO 834-1: Fire-resistance tests–Elements of building construction–Part 1: G. requirements
7. P. L. Nguyen, X. H. Vu, E. Ferrier. Characterization of pultruded carbon fibre reinforced
polymer (P-CFRP) under two elevated temperature-mechanical load cases: Residual and
thermo-mechanical regimes. Constr. Build. Mater 165 (2018) 395-412.
 A study on improvement of early - age strength
of super sulfated cement using phosphogypsum

Nguyen Ngoc Lam1

1
National University of Civil Engineering, Hanoi, Vietnam
lamnn@nuce.edu.vn

Abstract. Supersulphated cement (SSC) is a newly developed non-burned ce-

mentitious material. It is a kind of environmentally-friendly cementitious mate-
rial due to its energy-saving, low-carbon emission, and waste-utilization. Com-
pared with ordinary Portland cement (PC), SSC has many advantages such as
low hydration heat, excellent property of resistance to sulfate, using less clinker
but more gypsum. However, the setting time of this binder is quite long, so the
strength development is slow at early age. The aim of this paper is to improve
the early-age strength for SSC binder samples with 30% Phosphogypsum, 60%
ground-granulated blast-furnace slag, the rests are cement or lime and activator
Na2SO4 with different content. The research results show that compressive
strength at 3-day age can be improved almost twice. Microstructural analysis by
infrared spectroscopy revealed that ettringite formed earlier in the case of using
activators and improved the microstructure and strength at early age of the
binder. However, the strength of the SSC at the later age was reduced when
compared to the control sample without activator. The results also show that the
optimum content of activator Na2SO4 for improving SSC strength at early age is
1%.

Keywords: Super Sulphated Cement, Compressive Strength, Activators.

1 Introduction

Currently, the study of the use of industrial waste products such as phosphogyp-
sum, blast furnace slag and some other waste products to make building materials has
economic - technical benefits. One of the new binders has the potential to replace
cement in practice is Super Sulfated Cement which composes of ground granular blast
furnace slag, gypsum, a small amount of alkali activator (cement, clinker cement,
lime) [1-5]. This type of binder allows the consumption of a large amount of gypsum
in general and phosphogypsum in particular, so that it is suitable to the guidelines and
policies of state management agencies in the issue of environmental protection and
sustainable development [5-9]. Singh et al., studied binders with high durability on
the basis of anhydrous phosphogypsum, fly ash, finely ground furnace slag and Port-
land cement [8]. The results show that this binder is not only suitable for manufactur-
ing construction mortar, unburnt bricks and fiberglass reinforced composite materials,
but also can be replaced up to 25% of cement content in concrete, without causing

© Springer Nature Singapore Pte Ltd. 2020 537

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_84
538 N. N. Lam

any loss of strength. Ding Sha [3] have made SSC containing 45% phosphogypsum,
48% finely ground slag, 2% steel slag and 5% cement clinker, the results show the
strength of concrete at 28 days can reach 38.6MPa, this value is equivalent to concrete
sample using a 100% cement as a binder. Some other applications of this type of
binder can be mentioned as roadbed in China [10], production of unburnt bricks and
block bricks [11]. However, super sulfated cements have a big disadvantage, that is
the strength at early age is quite low, so improving the strength at early age contrib-
utes to improving the practicality and feasibility in application. Using SSC creates the
opportunity to reuse industrial waste in large quantities, contributing to reducing
product costs, improving the meaning of sustainable construction development, reduc-
ing environmental pollution.

2 Materials and experimental methods

2.1 Experimental materials

The main materials used in this study are PC40 cement (PC), phosphogypsum
(PG), ground granulated furnace slag (GGBFS), quick lime (QL) and hardening acti-
vator (Na2SO4). Phosphogypsum is obtained at DAP Dinh Vu chemical fertilizer
plant, Hai Phong, Vietnam, dried in an oven at 60°C then crushed in a vibrating ball
mill for 10 minutes, the density of PG was 2.34 g/cm3, the average particle size is
11µm. GGBFS has an average particle size of about 10.2 m. Portland cement used in
the study was PC40 But Son, with Blaine fineness of 3890 cm2/g, with compressive
strength after 3 days and 28 days respectively of 29.8 MPa and 52.2 MPa. Quick lime
powder is added to binder to activate GGBFS, contribute to the formation of C-S-H at
the early age in binder. Standard sand satisfies TCVN 6227: 1996 for determining the
strength of binder. Sodium sulfate activator (Na2SO4) to enhance the early age of this
SSC binder. The water used in the research meets the technical requirements of Viet-
namese standard TCVN 4506: 2012 of mixing water for mortar and concrete.

2.2 Experimental method

The formulation of SSCs used in this study is shown in Table 1:

Table 1. Formulation of SSCs

Name of Percentage of materials in SSC,% by weight

N°
Composition PG GGBFS) PC QL
1 CP1 30 60 10 0
2 CP2 30 60 0 10
3 CP3 30 60 5 5

Compressive strength is determined according to TCVN 6016: 2011. The ratio of

sand/binder (S/B) by weight is 3: 1, the ratio of Water/Binder is fixed at 0.5. After
mixing, the fresh mortar mixture will be cast in 40x40x160 mm mold. After 2 days,
A study on improvement of early - age strength … 539

the mortar samples will be removed and cured in a sealed ball bag to the age of testing
the compressive strength at 3 days, 7 days and 28 days. The infrared spectroscopy
method is also used to determine the hydration minerals formed in SSCs in both cases
of with and without hardening activator Na2SO4. The evaluation of the main hydration
products of this binder is determined through the characteristic peaks of ettringite, C-
S-H and chemically bound water.

3 Results and discussion

3.1 Compressive strength development of SSCs

The compressive strength of SSCs with time in 2 cases of with and without harden-
ing activator Na2SO4 is presented in table 1:

Thus, it can be seen that the inherent disadvantage of SSCs is that the compressive
strength at the early-age is quite low at 3 days, only 16-26% of 28-day strength, while
at 7 days about 60-70%. Therefore, the study to improve the strength of SSC at early
age will contribute to the feasibility of this type of binder in application.
In general, when adding Na2SO4 activator to the binder, all binder samples have 3-
day age compressive strength higher 1.5 to 2 times than that in case of without
Na2SO4. It is explained by the fact that Na2SO4 has high solubility, SO42- ion will
react quickly with Ca(OH)2 and GGBFS, so it will soon form ettringite so that the
compressive strength of 3 days is improved and increased faster.
 540 N. N. Lam

























Fig. 1. Compressive strength of SSC binders with and without Na2SO4 at 3 days and 28 days

From the results in Figure 1, it can be seen that when using 0.5%-1,0% Na2SO4, the
3-day compressive strength is increased insignificantly by about 1.2-2.0 times when
compared to SSC samples without Na2SO4. However, the results also show that when
increasing Na2SO4 from 1% to 1.5%, the compressive strength at 3 days is not signifi-
cantly improved. In general, it can be seen that as the more the content of Na2SO4
increases, the more the early age compressive strength is high. However, the com-
pressive strength tends to decrease slightly in the later age. On the other hand, the 3-
day age strength of the binder sample using 1% Na2SO4 is insignificantly smaller than
the sample using 1.5% Na2SO4, but compressive strength at 28 days has little attenua-
tion. Therefore, the content of 1% Na2SO4 is proposed to use to improve the compres-
sive strength for SSC at early age.

3.2 Microstructure of SSC by infrared spectroscopy method IR

This research used the infrared spectroscopy analysis method to understand and
explain the effect of hardening activator on the microstructure formation of SSC. The
binder samples hydration was stopped at 3 days of age by acetone, then crushed to be
smaller than 0.09 mm. The experimental results are shown in Figure 2 to Figure 4.

Fig. 2. IR spectra of 3 days hardened CP1 with 1% Na2SO4 and without Na2SO4
A study on improvement of early - age strength … 541

  
,

" ,
" $#


(#

"!,
" $#


"!,
" $#


"%

"
!!"

##)

#&"%
)%

!%
!&'
!&"

"#&

!$#
(&

"#$

!
&&

%)$

%

"!!

$
#&
#"
"(
"$
"
!&
!"
(
$


 
!

Fig. 3. IR spectra of 3 days hardened CP2 with 1% Na2SO4 and without Na2SO4

  
-
$!-
# %$

$"-
# %$


$!-
# %$


$!-
# %$


"#
$


$!-
# %$

$"-
# %$


$"-
# %$


$$*!

""#!

$'#&
"'(!
"'#!
*&!

#$'!
"%$!

''!
&*%

#$%!

%!!!
$'!!
$#!!
#)!!
#%!!
#!!!
"'!!
"#!!
)!!
%!!


 
"
Fig. 4. IR spectra of 3 days hardened CP3 with 1% Na2SO4 and without Na2SO4

The results of the determination of infrared spectra of 3 SSC compositions clearly

show the difference between the mixtures without Na2SO4 and with 1.0% Na2SO4.
When using activator, the adsorption intensity of hydration products is stronger, this
explains the reason why the binder with 1.0% Na2SO4 hardened faster and have com-
pressive strength at early age. Specifically at the age of 3 days, the compositions
without Na2SO4 do not have typical C-S-H peaks at wave numbers 660cm-1, 594 cm-1,
and 860cm-1. On the other hand, the characteristic peak of the SO42- in ettringite at
1120cm-1 in the case of using Na2SO4 is also stronger. Another important difference is
that chemically bound water at wave numbers from 3000cm-1 to 3800cm-1 in hydra-
tion products of compositions using activator is stronger than the compositions not
using activator.
542 N. N. Lam

4 Conclusions

Based on the findings, some conclusions can be drawn as follows:

The amount of cement and quick lime used in SSC binder is very low (5-10% by
weight), 90% of the remaining materials in binder are industrial waste, the compres-
sive strength at 28 days can reach approximately 50 MPa.
The use of Na2SO4 increases the rate of hardening at early - age but also has the
potential to reduce the compressive strength of the binder in long-term, but the degra-
dation of strength is negligible with a Na2SO4 content of 1%.
Studying microstructure by infrared spectroscopy method clearly explained the hy-
dration process of binder at 3 days of age, and the effect of activator on the formation
of hydration products at early age.

References
1. Matschei, T., Bellmann F., and Stark J.: Hydration Behaviour of Sulphate- Activated Slag
Cements. Advances in Cement Research, 17(4), 167-178 (2005).
2. Kuryatnyk T., Angulski da Luz C., Ambroise, J.: Valorization of Phos-phogypsum as Hy-
draulic Binder. Journal of Hazardous Materials 160 (2), 681–687 (2008).
3. Ding Sha, S. Zonghe, Chen Wei, Lu Jianxin, and Tian Sufang.: Properties of Supersul-
phated Phosphogysum slag Cement (SSC) Concrete. Journal of Wuhan University of
Technology 29(1), 109–113 (2014).
4. Mun, K.J., Hyoung W.K., and Lee C.W.: Basic Properties of Non- Sintering Cement Us-
ing Phosphogypsum and Waste Lime as Activator. Construction and Building Materials
21(6), 1342 – 1350 (2007).
5. Taneja, C.A., Singh M., Tehri S.P., and Raj T.: Super sulphated cement from waste phos-
phogypsum. In 12th International conference on silicate industry and silicate science, Bu-
dapest - Hungary, pp. 821-827 (1977).
6. Mulla M. S., K.V., Gundakalle V. D., Thakare R. B.: Phosphogypsum applications in ce-
ment concrete. In: International Workshop and Conference on Construction and Materials,
CONMAT 2003, Department of Civil Engineering, IIT Kharagpur, India, pp. 677-687,
January 9-11, 2003.
7. Manjit S., Garg M.: Phosphogypsum-Fly ash cementitious binder - Its hydration and
strength development, Cement and Concrete Research 4(25), 752-58 (1995).
8. Singh M., Garg M.: Cementitious Binder from Fly Ash and Other In-dustrial Wastes. Ce-
ment and Concrete Research 29(3), 309-314 (1999).
9. Garg M., Singh M., Kumar R.: Some aspects of durability of a phos-phogypsum-lime-fly
ash binder. Construction and Building Materials 10(4), 273-279 (1996).
10. Weiguo, S., Mingkai Z., Qinglin Z.: Study on Lime–Fly Ash–Phosphogypsum Binder.
Construction and Building Materials 21(7), 1480-1485 (2007).
11. Sunil K.: Perspective Study on Fly Ash–Lime–Gypsum Bricks and Hollow Blocks for
Low Cost Housing Development. Construction and Building Materials 16(8), 519-525
(2002).
 Effect of calcium sulfate type and dosage on early
strength and porosity of self-leveling underlayments

Nguyen Ngoc Lam1, Elodie Prud’homme2 and Jean-François Georgin2

1
National University of Civil Engineering, Hanoi, Vietnam
2
Lyon University, Insa Lyon, France
lamnn@nuce.edu.vn

Abstract. In recent years, self-leveling underlayments (SLUs) are often used due to
high flow ability and self-smoothing properties. Besides the strength, other characteris-
tics of SLUs such as workability, rapid drying, rapid hardening, shrinkage compensa-
tion, etc...are required. This paper presents the effect of calcium sulfate type and dosage
on early-age strength and porosity of SLUs based on ettringite binders. The raw materi-
als used for making SLUs binders are calcium aluminate cement (CAC), Portland ce-
ment (PC) and calcium sulfate (C$Hx). Two types of calcium sulfates (hemihydrate and
anhydrite) were used in this research. The ettringite formation is the reaction that con-
trols the mechanical properties at early age of SLUs such as rapid hardening, early
strength gain, etc. It was found that using hemihydrate in SLUs gives higher compres-
sive strength than using anhydrite at the same content of calcium sulfate. On the other
hand, the pore size of SLUs will decrease with increasing the amount of calcium sulfate
in SLUs thanks to the higher amount of ettringite.

Keywords: Self-leveling Underlayments, Ettringite Binder, Calcium Sulfate,

Early-age

1 Introduction

In recent years, self-leveling underlayments (SLUs) are often used for a concrete floor
in various fields, i.e. office buildings, apartments, stores, schools, hospitals, factories,
parking lots, vessels and repairing work due to high flow ability and self-smoothing
properties [1-3]. Besides the strength, other characteristics of SLUs such as workabil-
ity, rapid drying, rapid hardening, shrinkage compensation, etc., are required. There-
fore, it is necessary to meet the increasing demand for SLUs that is becoming diversi-
fied. One of the innovative binders for SLUs is ettringite binder. This type of binder
includes calcium aluminate cement (CAC), different sources of calcium sulfate
(C$Hx) and Portland cement (PC). This combination influences on rapid setting and
hardening and self – desiccation shrinkage compensation [4,5]. There are three types
of C$Hx – anhydrite, dihydrate and two modifications as alpha-hemihydrate and beta-
hemihydrate – which were used with the mixed binders for different SLU composi-
tions. After contacting water, anhydrite and dihydrate will harden very slowly. How-
ever, if alpha-hemihydrate and beta-hemihydrate are mixed with water, they react

© Springer Nature Singapore Pte Ltd. 2020 543

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_85
544 N. N. Lam et al.
more quickly and harden. Furthermore, the rate of dissolution and hydration speed of
beta-hemihydrate is quicker than alpha-hemihydrate [1,6]. Thus, performances of
SLUs themselves differ largely depending on the sulfate calcium used for SLUs. In
general, the kinetics of hydration, mechanical properties development and volume
variations of ettringite mixture depend essentially on the CAC/C$Hx ratio and on the
type of calcium sulfate (anhydrite, hemihydrate or gypsum) [7-9].
The aim of this work is to investigate the development of strength and porosity of
SLUs at early-age with different ratio of CAC/C$Hx and different source of C$Hx.

2 Materials and Methods

2.1 Materials

The binder of SLUs in this research consists of the calcium aluminate cement (CAC),
Portland cement (CEM I 42.5) and two types of calcium sulfate: anhydrite (A) or -
hemihydrate (P). The skeleton of SLUs composes of silica sand, slag and limestone
fillers. The formulation of SLUs is presented in Table 1:

Table 1. Formulation of self-leveling underlayments.

Raw materials Percentage composition of raw materials, %

75CAC/25C$Hx 90CAC/10C$Hx
CAC 24.26 29.11
C$Hx 8.09 3.23
CEM I - 42,5 3.97 3.97
Silica sand
Carbonate powder 33.41 33.41
Slag
Superplasticizer
Viscosity agent 5.46 5.46
Anti-shrinkage agent
Retarder + Accelerator 0.238 0.238
Water 24.57 24.57

2.2 Experimental methods

Strength: The fresh mixture of SLUs is poured into the mold and placed at 20°C in
endogenous condition. In order to determine the mechanical properties from 2h to 24h
after mixing such as compressive strength, 3-point flexural strength, the SLU is un-
molded just before mechanical tests.
Mercury intrusion porosimetry test: Porosity was measured on fragments of spec-
imens (40 x 40 x 160 mm), which were first stopped hydration by grinding the sam-
ples of self leveling underlayment at the required times and putting the pieces ob-
 Effect of calcium sulfate type and dosage … 545

tained in an acetone bath for 2 days then they were filtered in a Buchner and placed in
a dryer for 2 days to extract excess acetone and limit carbonation.

3 Results and discussion

3.1 Development of early-age compressive strength of SLUs

Since SLUs harden very quickly, the strength of SLUs was determined as soon as
possible, from 2 hours up to 24 hours after mixing. Fig. 1 shows the development of
gth and flexural strength of 4 SLU

Fig. 1. Compressive strength of the 4 SLUs with time.

a) Compressive strength b) Flexural strength

The results in Fig. 1 indicates that the strength of SLUs increases quickly at early age,
After 24h, the compressive strength and flexural strength of the compositions
75CAC/25C$Hx reach 18-20 MPa and 5 MPa respectively, while the strengths of the
compositions 90CAC/10C$Hx are lower, about 13-17 MPa for the compressive
strength and 4 - 4.5 MPa for the flexural strength. It also was found the strength de-
velopement of SLUs containing the -hemihydrate is also faster than the SLUs with
anhydrite. This can be explained by the fact that the CAC reacts with hemihydrate to
form ettringite more rapidly than that with anhydrite.

3.2 Mercury Intrusion Porosimetry (MIP)

In this research, the Mercury Intrusion Porosimetry was used to investigate how the
source of calcium sulfate and its content (CAC/C$Hx) influence the porosity and the
pore distribution of SLUs at 2h, 5h, 10h and 24h after mixing.
 546 N. N. Lam et al.

Total porosity
Total porosity development of SLUs with different types of calcium sulfate at differ-
ent ratio CAC/C$Hx was shown in Fig. 2 and Fig. 3.










 





























 










































 



 


 



Fig.2. Total porosity Fig. 3. Relationship between the strengths

of SLUs and the total porosity of SLUs
a) Compressive strength b) Flexural strength
and total porosity and total porosity

The total porosity is an inversely proportional function of hydration time. The pore
volume of the SLUs decreases quickly with time due to hydration products filling in
the pores, especially during the first 10 hours. Afterward, the porosity of SLUs begins
to decrease slowly.
The relationship between the total porosity and strength development of SLUs is in
Fig. 3. The results shows that the strengths of SLUs with different source and content
of calcium sulfate are fairly linear with the total porosity. The strength increases when
the porosity decreases due to an increasing amount of hydrates which fill the pore and
improves mechanical properties of SLUs.
It is noted that for the same value of the porosity, the strength of compositions
75CAC/25C$Hx is always higher than that of the compositions 90CAC/10C$Hx; the
compressive strength of the composition 75CAC/25P is the highest, while the highest
flexural strength is for the composition 75CAC/25A. The composition 90CAC/10A
has the lowest strength. In the previous section on the total porosity of SLUs, the total
porosity among these compositions is not clear, therefore the strengths of these SLUs
may depend principally on their pore distribution.

Pore Distribution
The pore distribution of the 4 SLUs are shown in Fig. 4. Only the results of 2h and
24h are presented in this paper.
Effect of calcium sulfate type and dosage … 547

40 0.6
35
0.5
30
0.4

dV/dlog(r)
25
Pore entry, [%]

20 0.3
15
0.2
10
5 0.1
0 0
0.001 0.01 0.1 1 10 0.001 0.01 0.1 1 10
Pore entry radius, [ m] Pore entry radius, [ m]
90CAC/10A-2h 75CAC/25A-2h 90CAC/10A-2h 75CAC/25A-2h
75CAC/25P-2h 90CAC/10P-2h 75CAC/25P-2h 90CAC/10P-2h
40 0.6
35 0.5
30
Pore volume, [%]

0.4
25
dV/dlog(r)

20 0.3
15
0.2
10
5 0.1
0 0
0.001 0.01 0.1 1 10 0.001 0.01 0.1 1 10
Pore entry radius, [ m] Pore entry radius, [ m]
90CAC/10A-24h 75CAC/25A-24h 90CAC/10A-24h 75CAC/25A-24h
75CAC/25P-24h 90CAC/10P-24h 75CAC/25P-24h 90CAC/10P-24h

Fig. 4. Pore distribution of SLUs at 2h and 24h.

a) Cumulative pore volume at 2h b) Pore size distribution at 2h
c) Cumulative pore volume at 24h d) Pore size distribution at 24h

In Fig. 4, it is obvious that there is a decrease in the critical size of the pore with the
progress of hydration and with increasing ratio CAC/C$Hx. This ''critical'' size is the
radius of the pore from which there is a significant variation in the mercury intrusion
process. With the advancement of hydration, the intergranular space decreases by
filling hydrates, which induces a radius ''threshold'' to become smaller. The critical
pore size of SLUs 75CAC/25C$Hx or SLUs containing hemihydrate is smaller than
that of SLUs 90CAC/10C$Hx or SLUs containing anhydrite. It means the pore size of
these compositions is also finer. That is the reason why the strength of SLUs
75CAC/25C$Hx or SLUs containing hemihydrate are higher. It was also found that
the pore distribution shapes are similar for all 4 SLUs. The main pore family is
macropores (with pore radius greater than 0.025 micrometre). A monomodal pore size
distribution of 0.01- 1.0 micrometre is clearly present for all SLUs at early-age.
548 N. N. Lam et al.

4 Conclusion

According to the experimental work undertaken, the following conclusions can be

established:
The early-age strength of SLUs with 25C$Hx/75CAC is better than that of of SLUs
with 10C$Hx/90CAC. For the same value of the porosity, the strength of the SLUs
75CAC/25C$Hx is higher than that of the SLUs 90CAC/10C$Hx thanks to the smaller
pore size.
The total porosity of SLUs quickly decreases at early-age, especially before 10h. The
type of calcium sulfate and the ratio C$Hx/CAC does not influence the total porosity
but influences the pore distribution.
The strength of the SLUs containing hemihydrate is more important than that of the
compositions containing anhydrite.

References
1. Onishi, K., and Thomas A.B.: Investigation into relations among technological properties,
hydration kinetics and early age hydration of self-leveling underlayments. Cement and
Concrete Research, 40 (7), 1034 - 1040 (2010).
2. Kighelman.: Hydration and Structure development of ternary Binder System as used in
Self-levelling compounds. Thesis, Ecole Polytechnique Fédérale de Lausanne (2007).
3. Radek Magrla, Dominik Gazdi , and Jan Novák.: Testing self-levelling floor screeds on
the basis of ternary binder. Advanced materials Research, 897, 17-20 (2014).
4. Le Saout, G., Lothenbach B., Taquet P., Fryda H., and Winnefeld F.: Hydration study of a
calcium aluminate cement blended with anhydrite. In Calcium aluminates - Proceedings
of the international conference, Avignon - France (2014).
5. Lutz, H., and Bayer R.: Dry Mortars', Ullmann’s encyclopedia of industrial chemistry. 11,
541-579 (2010).
6. Daimon, M., Rhee K.H. and Kondo R.: On the hydration mechanisms of calcium sulfate
hemihydrate. Journal of the Ceramic Association of Japan 78, 277–82 (1970).
7. Bayoux, J.P., Bonin A., Marcdargent S., Mathieu A. and Verschaeve M.: Study of the hy-
dration properties of aluminous cement and calcium sulfate mixes. In: Calcium Aluminate
Cement , E & FN Spon, London, pp. 320 - 334 (1990).
8. Bizzozero, J., Gosselin C, and Scrivener K.L.: Expansion mechanisms in calcium alumi-
nate and sulfoaluminate systems with calcium sulfate. Cement and Concrete Research 56,
190-202(2014).
9. Scrivener, K.L., and Capmas A.: Chapter 13: Calcium Aluminate Cements, in: P.C.
Hewlett (Ed.), LEA's Chemistry of Cement and Concrete, Arnold, London, (1998).
 Heat resistant mortar using Portland cement and waste
clay bricks

Nguyen Ngoc Lam1

1
National University of Civil Engineering, Hanoi, Vietnam
lamnn@nuce.edu.vn

Abstract. The waste materials of clay bricks are usually come in different
ways. Some are created in factories during and after the production process as a
result of human mistakes, inappropriate materials, or a mistake in production
process, some others are formed in transportation and distribution stage and fi-
nally a large part of waste materials are formed as a result of destroying build-
ings. The amount of waste materials may account to millions of tons annually.
Recycling waste clay brick by incorporating them into building materials is a
practical solution for pollution problem. The aim of this study is to investigate
the use of waste brick as aggregates and as a partial replacement for Portland
cement in the production of heat-resistant mortar, which is used for the potential
fire structure, created by flammable materials stored or used in the building.
This mortar is capable of working in the heat range of 800°C and 1000°C. Me-
chanical properties of mortar are based on the finding a suitable content of addi-
tives from waste brick used for binder and a reasonable aggregate gradation.
The research results showed that the use of additives from waste brick in differ-
ent proportions by weight (20-45%) for the cement increase heat resistance of
the binder. In addition, aggregate particles from waste brick having sintered
structure can be able to improve the thermal stability of mortar at high tempera-
ture.

Keywords: Heat resistant mortar, Waste clay brick, Portland cement.

1 Introduction

During the last decade, the problem of waste accumulation, which is increasing year
by year, exists worldwide. Most of waste materials are left as a landfill material or
illegally dumped. Environmental impact can be reduced by making more sustainable
use of this waste [1]. The main principle of a quality waste management is finding
ways to recycle and reuse existing, and safe and ecologically acceptable depositing of
unused waste [2]. Recycling or re-using of bricks is an environmentally friendly way
of eliminating it from the waste stream. The waste materials of clay bricks are usually
come in different ways. Some are created in factories during and after the production
process as a result of human mistakes, inappropriate materials, or a mistake in produc-

© Springer Nature Singapore Pte Ltd. 2020 549

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_86
550 N. N. Lam

tion process, some others are formed in transportation and distribution stage and final-
ly a large part of waste materials are formed as a result of destroying buildings. The
amount of waste materials may account to millions of tons annually [3].
Recycled crushed clay brick is one of the best aggregates for mortar and
concrete that may have to resist fire, and it performs much better than similar concrete
containing granite aggregate [4,5]. Brick aggregate is a thermally stable aggregate. It
is probably why it performs well when used as an aggregate in mortar and concrete
subjected to high temperatures. Fire resistance of clay brickwork is an important char-
acteristic, since it has been recognized that brickwork masonry is a very effective
material for resisting and preventing the spread of fire [4].
Several studies have been conducted to investigate the potential of using
crushed clay brick as an alternate aggregate [6]. Poon and Chan investigated the pos-
sibility of using crushed clay brick as aggregates in sub-base materials [7]. Akhta-
ruzzaman et al. studied the use of crushed clay brick aggregate as a 100% replacement
of coarse natural aggregate in concrete [8]. The physical and mechanical properties of
solid cement brick manufactured with crushed clay brick as a recycled aggregate have
been reported by Sadek [9]. Ge et al. studied the effect of clay brick powder on con-
crete mechanical properties [10].
This paper presents a comprehensive study on the use of recycled aggregates
and powder produced from clay brick production wastes in heat resistant mortar in-
dustry. The main focus of the research is to present an additional information in the
field of recycling clay brick waste in order to explore the possible uses of these recy-
clable materials in structural applications. The aim of this study is to investigate the
use of waste brick as aggregates and as a partial replacement for Portland cement in
the production of heat-resistant mortar.

2 Materials and Experimental methods

2.1 Materials

Cement used in this research is an ordinary Portland cement PC30 Hoang Thach. The
physical properties of cement are presented in Table 1. Recycled fine aggregates of
0.15mm to 2.5mm and brick powder (BP) from clay bricks wastes were obtained
through crushing process.

Table 1. Physical properties of cement PC30 Hoang Thach

Compressive
Standard Setting time (minutes)
Blaine strength, MPa
consistency,
surface, cm2/g
% Initial Final 3 days 28 days
3760 28.0 110 185 18.8 39.0
Heat resistant mortar using Portland cement … 551

2.2 Experimental method

Preparation of the hardened binder pastes
Powder from clay brick waste is introduced in binder as a partial replacement of ce-
ment by weight in all mixtures from 20 to 50%. Heat-treated binder pastes were pre-
pared by using the standard water of consistency of cement paste. Each paste was
prepared by mixing the dry mix with the required amount of water for about 3 min.
After complete mixing, the resultant paste was molded into cubic specimens by using
2.0 cm cube moulds. The specimens were demoulded after 24 h, cured for 28 days
under tap water, dried for 24 h at 105°C, then subjected to thermal treatment for 2 h at
100, 200, 400, 600, 800°C and 1000°C with rate of heating 10°C/min, afterward
cooled to room temperature in the furnace to determine the compressive strength of
treated binder pastes.
Preparation of the hardened mortar
Binder (cement + clay brick powder) was mixed with sand (fine aggregate from clay
brick waste) by the ratio of binder and sand was 1:3. The water/binder ratio used for
the preparation of the specimen was 0.65. The fresh mortar was casted into the 7.07
cm cube moulds. After 24 hours, the mortar samples were removed from the mould
and placed in water for 28 days at room temperature. Later those cubes were dried at
105°C for 24 h in the oven (except control sample at 25°C). Afterward, these cubes
were heated for two hours at six different temperatures (100°C, 200°C, 400°C, 600°C,
800°C and 1000°C), and then remained in the room temperature for 24 hours for cool-
ing and testing for compressive strength.

3 Results and discussion

3.1 Results on heat resistant binder

The properties of heat resistant binders such as standard consistency, setting time and
compressive strength after curing 28 days at 25°C are presented in Table 2:

Table 2. Properties of heat resistant binder with different contents of clay brick powder

Ratio Setting time, Standard Compressive

(Cement/brick minutes consistency; strength
powder) Initial Final % (28days), MPa
100:0 110 185 28.25 39.0
80:20 135 205 29.25 35.0
75:25 140 215 29.75 33.4
70:30 150 230 30.25 31.8
65:35 165 245 30.75 29.2
60:40 180 260 31.25 27.1
55:45 190 275 31.75 24.7
50:50 205 280 32.25 23.1
552 N. N. Lam

With respect to the control sample, when increasing the clay brick powder additive in
binder (from 20% to 50%), the standard consistency is increased; consequently the
setting time is prolonged. The compressive strength is inversely proportional to
amount of clay brick powder in binder. By the experimental results, influence of tem-
perature and clay brick powder content on compressive strength of binder was deter-
mined and provided in Table 3 and in Figure 1:

Table 3. Compressive strength of binder using clay brick powder at different temperatures

Compressive strength at different temperatures, MPa

Binder
25°C 100°C 200°C 400°C 600°C 800°C 1000°C
0% BP 39 58.6 71.7 56.9 35.8 12.7 8.0
20% BP 35 52.9 59.1 46.5 35 31.7 11.2
25% BP 33.4 49.7 55.5 42.2 33.7 29.2 12.6
30% BP 31.8 47.4 52.7 39 30 27.4 13.4
35% BP 29.2 44.6 50.1 37 28 26.2 15.0
40% BP 27.1 38.8 47.1 34 27 25.1 18.7
45% BP 24.7 34.2 43 30.7 25.6 24.1 23.0
50% BP 23.1 31.5 39 27.5 25 23 21.8

Fig. 1. Compressive strength of binder using clay brick powder at different temperatures

In fact, the reduction in binder compressive strength at 28 days is observed for binder
modified with 20% - 50% of clay brick powder as compared with control sample
(100% cement). From the results in table 3 and figure 1, it is prominent that the com-
pressive strength has increased up to 200°C for all percentage of clay brick powder in
binder as a replacement of cement and after that compressive strength of all mixture
Heat resistant mortar using Portland cement … 553

begin to decrease gradually as the temperature is further increased. These results can
be explained by the fact that the hydration process of binders is accelerated faster and
stronger when the temperature rises from 20-200°C. But when the temperature is over
200°C, the decomposition of calcium silicate hydrate (C-S-H), which degrades the
compressive strength of binders, occurs. Contradictorily, slightly beneficial effect in
compressive strength is noticeable when clay brick powder is incorporated as an addi-
tive to cement, that is to say that the decline rate of compressive strength will de-
crease at high temperature, specially at 800°C and 1000°C in the presence of clay
brick powder. The largest value of compressive strength at 800°C is belong to compo-
sition containing 20% clay brick powder in binder; For the temperature at 1000°C, the
optimum amount of clay brick powder additive used is 45%.
3.2 Results on heat resistant mortar

The heat resistant mortar specimens were prepared by the ratio of binder and fine
aggregate was 1:3. The optimum water/binder (W/B) ratio selected was 0.65. The
optimum amount of clay brick powder in binder is 20% at 800°C and 45% at 1000°C
with respect to the control binder. Compressive strength of mortar using these opti-
mum values of clay brick powder was determined on the specimens of 7.07 cm x 7.07
x 7.07 cm at different temperatures. The results are shown in the table 4 and graph-
ically shown in Figure 2 as follows:

Table 4. Compressive strength of mortars at different temperatures

Mortar compressive strength at different temperature (MPa)

N° % BP
25°C 100°C 200°C 400°C 600°C 800°C 1000°C
1 20 31.7 33.5 36.4 30.8 24.1 20.3 14.1
2 45 24.6 28.1 31.3 26.2 21.5 18.2 16.2

Fig. 2. Compressive strength of mortars using optimum content clay brick powder
at different temperature
554 N. N. Lam

From the results in table 4 and figure 2, it is obvious that the mortar exhibits more
resistance to high temperature (1000°C) as the percentage of clay brick powder in-
creased up to 45%. These results may be due to the fact that crushed clay brick pow-
der improves binder microstructure at high temperature. However, the compressive
strength of mortar containing less quantity of clay brick powder is more important at
lower temperature (800°C).

4 Conclusion

From the experimental results, the following conclusion may be drawn:

The compressive strength of binder containing clay brick powder initially increas-
es up to 200°C and after that it decreases with the further increase of temperature.
The compressive strength of binder decreases with the increase of percentage of
clay brick powder as a replacement of cement at temperature less than 800°C but it
increases with the further increase of percentage of clay brick powder at temperature
more than 800°C.
The strength development of heat resistant mortar is in the same manner of the
binder. The use of crushed clay brick powder as a cement addition also improves
mortar compressive strength at high temperature.

References
1. Batayneh, M., Marie I., and Asi I.: Use of selected waste materials in concrete mixes.
Waste Management, 27(12), 1870-1876 (2007).
2. Bjegovi , D.: Sustainability as a Condition for Development in Croatia. In: Proceedings of
International Conference on Sustainability in the Cement and Concrete Industry,
Lillehammer - Norway, pp. 2-16 (2007).
3. Tavakoli D., Heidari A. and Pilehrood S.H.: Properties of Concrete made with Waste Clay
Brick as Sand Incorporating Nano SiO2. Indian Journal of Science and Technology, 7 (12),
1899 - 1905 (2014).
4. Kesegi , I., Netinger I., and Bjegovi D.: Recycled clay brick as an aggregate for concrete:
Overview. Technical Gazette, 15 (3), 35-40 (2008).
5. Khalaf, F.M., and De Venny A.S.: Performance of Brick Aggregate Concrete at High
Tempera-tures. Journal of Materials in Civil Engineering, 16, pp. 556-565 (2004).
6. Aliabdo, A., Abd-Elmoaty M., and Hani Hassan, H.: Utilization of crushed clay brick in
concrete industry. Alexandria Engineering Journal, 53 (1), 151–168 (2014).
7. Poon, C.S., and Chan D.: Feasible use of recycled concrete aggregates and crushed clay
brick as unbound road sub-base. Construction and Building Materials, 20(8), 578-585,
(2006).
8. Akhtaruzzaman, A., and Hasnat A.: Properties of concrete using crushed brick as aggre-
gate. Concrete International, 5 (2), 58-63 (1983).
9. Sadek, D.M.: Physico-mechanical properties of solid cement bricks containing recycled
aggre-gates. Journal of Advanced Research, 3(3), 253-260 (2012).
10. Ge, Z., Gao Z., Sun R., and Zheng L.: Mix design of concrete with recycled clay-brick-
powder using the orthogonal design method. Construction and Building Materials, 31,
289-293 (2012).
 Performance of Recycled Coarse Aggregate Concrete
with Different Nylon Fiber Content

Seungtae Lee1, Youngkyu Chu1, Sangwook Ha2, Yong Kim1 and Jaehong Jeong1
1
Kunsan National University, Kunsan, Jeonbuk 54150, South Korea
2
Buildchem Co. Ltd., Gongju, Choongnam 32533, South Korea
stlee@kunsan.ac.kr

Abstract. In the present study, the effect of nylon fiber (NF) on the permeabil-
ity as well as the mechanical properties of concrete incorporating 100% recy-
cled coarse aggregate (RA) was experimentally investigated. Concrete was pro-
duced by adding 0, 0.6 and 1.2 kg/m3 of NF and then cured in water for the pre-
determined period. Measurements of compressive strengths, ultrasonic pulse
velocity and total charge passed through concrete were carried out, and the cor-
responding test results were compared to those of concrete incorporating natural
coarse aggregate (NA). Test results indicated that recycled coarse aggregate
concrete (RAC) showed lower performance than natural coarse aggregate con-
crete (NAC) because of the adhered mortars in RA. However, it was obvious
that the addition of NF in RAC mixtures was much effective in enhancing the
performance of the concretes due to the crack bridging effect from NF. In par-
ticular, high content of NF (1.2 kg/m3) led to a beneficial effect on concrete
properties compared to low content of NF (0.6 kg/m3) with respect to mechani-
cal properties and permeability, especially for RAC mixtures.

Keywords: Recycled Coarse Aggregate, Nylon Fiber, Mechanical Properties,

Permeability.

1 Introduction

In Korea, it is advised to use recycled aggregate in concrete structures, since a huge

amount of construction wastes were generated, amounting for 67,000,000 tons yearly.
Furthermore, development in aggregate production technologies, and innovations in
the recycling process have led to reliable and enhanced properties of the recycled
aggregates. However, some doubts on the mechanical properties and durability of
concrete using recycled aggregates may arise. In general, performance of recycled
aggregate concrete heavily depends on the properties of the original concrete. A num-
ber of researchers have reported on the properties of concrete made with recycled
aggregate [1,2], but most of them resulted in a lower level of concrete strengths. This
is mainly due to the residual impurities on the surface of the recycled aggregates,
which blocked the strong bond between cement paste and aggregate. Moreover, it is
well known that there are two interfacial zones in recycled aggregate, which affect
negatively concrete properties.

© Springer Nature Singapore Pte Ltd. 2020 555

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_87
556 S. Lee et al.

In order to enhance the properties of recycled aggregate concrete, researchers have

come up with diverse modifying techniques. These modifying techniques include the
surface treatment, the use of mineral admixtures and the addition of fibers, which can
improve the performance of recycled aggregate concrete.
In particular, although nylon fiber (NF) shows a rising acceptance in the literatures
[3], it remains unpopular compared to steel fiber and polypropylene fiber. In compari-
son with steel and polypropylene fibers, while there is limited literature available on
the use of NF, some authors [3,4] reported that the use of NF stepped up the perfor-
mance after the presence of cracks in concrete, and sustained high stresses.
This study, thus, is aimed to investigate usability of NF in recycled aggregate con-
crete in order to be used in structural concrete, since the use of NF in field concrete is
gaining popularity nowadays. In order to achieve this goal, mechanical properties of
recycled coarse aggregate concretes with or without NF were experimentally exam-
ined, and the corresponding test results were compared to those of concretes incorpo-
rating natural coarse aggregate.

2 Experimental Procedure

2.1 Materials

ASTM C 150 Type I Portland cement was used in preparing the concrete speci-
mens. The density and specific surface area of the cement used were 3.15 g/cm3 and
328 m2/kg. Natural coarse aggregate (NA) and recycled coarse aggregate (RA) were
used as coarse aggregates for concrete production. RA was produced by crushing of
the waste concrete with a jaw crusher and an impact crusher. Both NA and RA have
continuous grains from 5 to 25 mm. The main properties of coarse aggregates are
presented in Table 1. Natural river sand was used as fine aggregate with a fineness
modulus of 2.80, a water absorption of 0.98 % and a density of 2.65 g/cm3.
The nylon fibers (NF) used in the present study had been supplied from Kolon
Global Co., Ltd, South Korea, and they are now commercially available in domestic
market. The properties of NF used in the present study are depicted in Table 2. In
addition, a polycarboxylate-based high performance superplasticizer (SP) was used to
improve the initially low workability of fresh concrete.

Table 1. Physical properties of coarse aggregate.

Properties NA RA
Density (g/cm3) 2.64 2.37
F.M. 7.17 7.43
Absorption (%) 0.66 4.31
Abrasion rate (%) 21.7 47.5
Adhered mortar (%) 5.56
Performance of Recycled Coarse Aggregate … 557

Table 2. Physical properties of nylon fiber.

Properties NF
Diameter (μm) 23
Length (mm) 19
3
Density (g/cm ) 1.16
Tensile strength (MPa) 919
Elastic modulus (GPa) 5.3

2.2 Mix Proportions

Four concrete mixes of 24 MPa grade concrete, which is a conventional mix pro-
portion for structural concrete, have been prepared. For all specimens, w/c ratio was
chosen as 0.50. Desired workability which is the range of 130 – 170 mm collapse was
obtained by means of SP. The RAC was prepared with 100% replacement of RA with
NA. Concrete mix proportions are shown in Table 3. The NF was added at the con-
centrations of 0, 0.6, and 1.2 kg/m3 for RAC mixtures, respectively. The SP was ini-
tially mixed with water to achieve a uniform dispersion. Concrete materials were
mixed and then cast into the 100 ⅹ 200 mm cylinder molds for compressive strength,
ultrasonic pulse velocity, and rapid chloride ion penetration tests. The concrete spec-
imens were cured under moisture condition for 24 h, after which they were demolded
and moved into a plastic tank for water curing until the time of testing.

Table 3. Mix proportions of concrete.

Concrete Mix proportion (kg/m3)

mixes water cement sand NA RA NF SP
NAC 170 340 745 1015 - - 2.45
RAC-A 170 340 745 - 912 - 2.72
RAC-B 170 340 745 - 912 0.6 2.72
RAC-C 170 340 745 - 912 1.2 2.72

2.3 Testing Methods

The compressive strengths of concrete cylinders with dimension of 100 mm in di-

ameter and 200 mm in length after 7, 28, and 91 days of curing were tested according
to ASTM C39. For evaluating the compressive strength, the cylinders were placed
under compression testing machine of 2000 kN capacity. The non-destructive test like
ultrasonic pulse velocity (UPV) was conducted according to ASTM C597-16. The
used transducers are 50 mm in diameter, and have a maximum resonant frequency of
54 kHz. Rapid chloride ion penetrability test was used to evaluate the permeability of
concrete specimens. This test was based on the standard test method of ASTM
C1202-18. The higher level of total charge passed represents the higher permeability
of concrete.
558 S. Lee et al.

3 Results and Discussion

3.1 Compressive Strength

Fig. 1 shows the compressive strength values of concrete mixtures with different
NF content. Due to the adhered mortars in recycled aggregates, the compressive
strength values of RAC-A is much lower than NAC. It can be seen that the compres-
sive strength value of RAC-A without NF decreased by 31% at the age of 28 days,
compared to that of NAC. At 91 days, NAC also exhibited higher compressive
strength value, showing 41.2 MPa, than the RAC-A (30.1 MPa). It is presumed that
the higher absorption and the larger pores, due to the adhered mortar in RA contribut-
ed to a decreased compressive strength in RAC-A.
Compressive strength generally increased when NF is added to the mixtures.
Among concrete mixtures with RA, the highest compressive strength was obtained
from RAC-C with 52.6 MPa at 91 days of curing. It was, thus, obvious that the addi-
tion of NF led to the increase of compressive strength in RAC mixtures. It is worth
noting that despite of the use of RA, RAC-C with 1.2 kg/m3 of NF showed much
higher compressive strength values, ranging from 27 ~ 41% compared to NAC at the
all stages of curing.

3.2 Ultrasonic pulse velocity

The ultrasonic pulse velocity (UPV) measurements as non-destructive tests are a

suitable tool to evaluate the quality of concrete. UPV values for concrete measured at
7, 28, and 91 days are shown in Figure 2. According to the classification criterion for
concrete based on ultrasonic pulse measurements by Leslie and Cheeseman [5], UPV
values for concrete mixes observed in this work can be classified as shown in Table 4.
It was observed from Figure 2 that the UPV values increased with curing time. The
increase of UPV with time is a natural development due to the increase of stiffness by
the hydration reaction [6]. It was found that UPV values of NAC were higher than
those of RAC. This can be attributed to the much higher water absorption capacity of
RA (4.31%) in comparison to NA (0.66%) which is due to the adhered mortar in RA,
resulting in the increased porosity in the RAC mixtures.
In case of RAC mixtures, with further incorporation of NF, the general increases in
UPV were examined, which shows similar trends with compressive strength results
by the increase of NF (Fig.1). At 7 days, the UPV values of RAC mixtures were com-
parable with NAC ranging from 2984 ~ 3354 m/s. However, relative small increases
in UPV were observed in RAC mixtures at later ages of curing. Similar to the results
of compressive strength, the UPV results indicated that RAC mixtures with NF
(RAC-B and RAC-C) showed higher values in UPV than RAC-A without NF. This
may be partially attributed to the bridge effect of NF which leads to the reduction of
micro-cracks in cement matrix by the addition of NF.
Performance of Recycled Coarse Aggregate … 559

60 5000
7 days
28 days Excellent
50
Compressive stregnth, MPa

91 days

40 4000

UPV, m/s
Good
30

Questionable
20 NAC 3000
RAC-A
RAC-B Poor
10 RAC-C

0 2000
0 20 40 60 80 100 NAC RAC-A RAC-B RAC-C
Age, days Concrete mixes

Fig. 1. Compressive strength of concrete. Fig. 2. Ultrasonic pulse velocity of concrete.

Table 4. Concrete classification based on UPV.

Pulse velocity (m/s) Concrete classification

4500 < V Excellent
3600 < V < 4500 Good
3000 < V < 3600 Questionable
2100 < V < 3000 Poor
V < 2100 Very poor

3.3 Rapid Chloride Ion Penetrability

Results of the rapid chloride ion penetration test (RCPT) based on ASTM C1202-
18 standard were presented in Figure 3. It was observed that the charge passed for 28-
day RAC mixtures were 2872, 2294, and 2050 coulombs for RAC-A, RAC-B, and
RAC-C, respectively. According to the ASTM C1202-18 criterion, the charges passed
for all RAC mixtures equated to the level of ‘Moderate’ while ‘Low’ for NAC. This
suggests that RA in concrete may impose a high risk to the durability of concrete
structure with respect to chloride-induced steel corrosion. However, the usage of NF
in concrete made with RA can mitigate the possibility of steel corrosion oriented from
external chlorides due to the reduction of micro-cracks in the cement matrix.

4 Conclusions

(1) Due to the adhered mortar in RA, the compressive strength values of RAC –
A were significantly lower than those of NAC. However, it was found that
the addition of NF led to the increase in compressive strength of RAC mix-
tures. In particular, this trend was more remarkable in RAC mixtures with a
high content of NF.
(2) From the test results of UPV, it was seen that in case of RAC mixtures, the
increase in UPV with further incorporation of NF was examined. This may
be due to the increase of materials density and the bridge effect of NF.
560 S. Lee et al.

(3) Based on the results of RCPT, it was observed that the reduction of total
charge was clearly examined in RAC mixtures with increasing NF content.
This implies that the usage of NF in RAC mixtures can mitigate the possibil-
ity of steel corrosion oriented from external chlorides due to the reduction of
micro-cracks in the cement matrix.
5000

High
Total charge passed, coulomb

4000

3000
Moderate

2000
Low

1000
Very low

0
NAC RAC-A RAC-B RAC-C
Concrete mixes
Fig. 3. Total charge passed through concrete.

Acknowledgements

This work was financially supported by the Korea Agency for Infrastructure Tech-
nology Advancement to carry out the R&D project, which is entitled as “Develop-
ment of organic and inorganic high-elasticity hybrid repair materials for the extended
service life of road facilities”. The authors gratefully acknowledge the financial sup-
port.

References
1. Ahmadi, M.; Farzin, S.; Hassani, S.; Motamedi, M. Mechanical properties of the concrete
containing recycled fibers and aggregates. Construction and Building Materials 144, 392-
398(2017).
2. Akca, K.R.; Cakir, O.; Ipek, M. Properties of polypropylene fiber reinforced concrete us-
ing recycled aggregates. Construction and Building Materials 98, 620-630(2015).
3. Kurtz, S.; Balaguru, P. Postcrack creep of polymeric fiber-reinforced concrete in flexure.
Cement and Concrete Research 30, 183-190(2000).
4. Perez-Pena, M.; Mobasher, B. Mechanical properties of fiber reinforced lightweight con-
crete composites. Cement and Concrete Research 24, 1121-1132(1994).
5. Leslie, J.R.; Cheeseman, W.J. An ultrasonic method for studying deterioration and crack-
ing in concrete structures. ACI Materials Journal 46, 17-36(1949).
Shariq, M.; Prasad, J.; Masood, A. Studies in ultrasonic pulse velocity of concrete contain-
ing GGBFS. Construction and Building Materials 40, 944-950(2013).
 Air permeability coefficients of expansive concrete
confined by rebars

Le Anh Van1, Kenichiro Nakarai1, Nguyen Huu May1,2, Yasutaka Kubori1, Toshikazu
Matsuyama1, Hajime Kawakane3, Shintaro Tani3
1
Department of Civil and Environmental Engineering, Graduate School of Engineering, Hiro-
shima University, 1-4-1, Kagamiyama, Higashi-Hiroshima, Hiroshima 739-8527, Japan
2
Department of Bridge and Tunnel, Faculty of Civil Engineering, University of Transport
Technology, 54 Trieu Khuc, Thanh Xuan, Hanoi 100000, Vietnam
3
Kyokuto Kowa Corporation, 2 Chome-6-31 Hikarimachi, Higashi Ward, Hiroshima 732-
0052, Japan

E-mail addresses: m173648-u.ac.jp (V.A. Le), nakarai@hiroshima-u.ac.jp

(K. Nakarai), maynh@utt.edu.vn (M.H. Nguyen), m174898@hiroshima-
u.ac.jp (Y. Kubori), mtoshi@hiroshima-u.ac.jp (T. Matsuyama)
kawakane@kkn.co.jp (H. Kawakane), tani@kkn.co.jp (S. Tani)

Abstract.
Torrent air permeability tester has been widely used as one the most reliable
non-destructive tests for evaluating the quality of the concrete cover. In this
study, the Torrent tester was employed to evaluate the influence of rebar con-
finement on the surface layer quality of expansive concrete. Blast furnace slag
cement was used with the expansive agent to prepare two types of specimens
with/without embedded rebars. The results comparing the variation of the
measured air permeability at the ages of 28 and 56 days indicated that the con-
crete cover confined by rebars obtained a better quality than that of the plain
one without rebars.

Keywords: Air permeability, Concrete cover, Expansive concrete, Rebars.

1 Introduction

Deterioration of concrete structure mainly involves in the movement of detrimental

substances through the near surface layers [1]. These aggressive agents will then
physiochemically interact with concrete constituents. As a result, the embedded steels
are corroded inducing concrete to crack and eventually to spall. Concrete cover plays
as an important barrier against the immigration of harmful substances from the sur-

© Springer Nature Singapore Pte Ltd. 2020 561

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_88
562 L. A. Van et al.

rounding environment into the insight of concrete body, which in turn, threatens the
concrete durability and its service life [2]. Abundant non-destructive methods for
evaluating the concrete cover quality have been introduced and developed basing on
the three fundamental transports mechanism (diffusion, absorption, permeability) of
the near surface layers [3-5]. Permeability which is considered to be one of the key
indicators for concrete cover quality assessment [6-7]. It is defined as the characteris-
tic that governs the flow rate of a fluid through a porous solid under a pressure gradi-
ent [8]. Torrent air permeability tester has been widely used as one the most reliable
non-destructive tests for evaluating the quality of the concrete cover through the kT
index that automatically calculated by the device [3,6]. The concrete permeability can
be largely reduced by the utilization of supplementary cementitious material such as
ground granulated blast furnace slag (GGBFS). Extensive studies have been conduct-
ed and pronounced its technical benefits for the refinement of the pores, as a conse-
quence, improving the permeability [9-12].
Concrete itself vulnerable to crack induced by shrinkage. Many previous studies
have been debating on how much the difference of cracks behaviours between con-
crete containing slag and concrete without slag [13-17]. According to these works,
they concluded that GGBFS concrete had lower cracking resistance (from 1.5 to 3%)
than the normal concrete without slag. The introduction of expansive concrete has
greatly positive effect on this problem [18-20]. Basically, expansive agent alone can
not definitely prevent concrete from shrinking [21], and sometimes it may cause crack
during the first 7 days after casting if concrete is free to expand [20,22]. In addition,
some studies have pointed out the advantages of reinforcement in reducing cracks
generated by expansion [20], and cracks induced by shrinkage [23-25].
Therefore, this current work aimed to investigate the beneficial contribution of re-
inforcement to inhibit crack due to expansion at early stage and shrinkage subsequent-
ly by comparing the coefficients of air permeability kT of two expansive concretes
with/without embedded rebars. The measured results at the age of 28 and 56 days
indicated that the kT of reinforced concrete obtained lower values than that of con-
crete without rebars.

2 Experimental program

2.1 Materials and mix proportions

The expansive concrete mixture was proportioned using W/B ratio of 0.53. This
ready-mix concrete was purchased directly from the supplier. Prior to casting, the
concrete slump and air content were checked to ensure that they meet the require-
ments. The details will be shown in Table 1.
Air permeability coefficients of expansive … 563

Table 1. Material mix proportions and measured parameters of the specimens

Slump Air s/a 2 Unit mass (kg/m3)

W/B 1
(cm) (%) (vol%) W C Ex S G
0.53 12±2.5 4.5±1.5 46.3 175 310 20 796 943

1
B: Binder includes cement (C) and expansive agent (Ex).
2
a: total volume of aggregates, s: volume of fine aggregate

2.2 Specimens preparation

Two types of expansive concretes with/without reinforcement were prepared with a

dimension of 600x600x900 in mm as shown in Fig. 1.

123
4Ø19
600
2@450

4Ø16
600
600

Fig.1. Dimension of specimens (Note: Fig.2. Arrangement of rebars - Cross

Unit of the dimensions are millimeters) section of specimen (Note: Unit of the
dimensions are millimeters)
Regarding the reinforcing specimen, the placement of rebars was indicated in Fig.
2. There were four longitudinal reinforcement bars surrounded by four stirrups with
diameters of 19 and 16 mm, respectively.
In relation to curing condition, the two specimens were cast at the same time and
cured for 7 days in accordance with JIS A 6202 - standard curing for expansive con-
crete. They were then exposed in a room with ambient relative humidity and tempera-
ture.
564 L. A. Van et al.

2.3 Experimental methods

A representative surface of each specimen was

taken for measurement. A total number of 8 test-
ing points was determined and selected to conduct
the air permeability test as shown in Fig. 3. This
selection was to avoid abnormal results might be
due to bleeding on the top of specimen and segre-
gation between the two casting layers. Torrent air
permeability tester was employed to measure the
coefficient of air permeability kT at the age of 28
and 56 days. kT was automatically calculated
from the devices [26].
Fig. 3. Sketch test locations
on measured surfaces.

3 Results and discussion

The air permeability results of the two concretes as shown in Fig. 4 obtained the
same linear tendency. Regarding the plain concrete, kT increased from 0.34 to 1.34
(10-16m2), whilst, in the case of reinforced one, it varied slightly from 0.16 to 0.32 (10-
16 2
m ). These increments can be explained by the establishment of pore system due to
evaporation of water [27-28].
Air permeability coefficient,

10
kT (10-16m2)

0.1 Plain concrete

Reinforced concrete
0.01
28d 56d
Age of concrete (days)

Fig. 4. Air permeability coefficient, kT at the ages of 28 and 56 days

Moreover, the obtained air permeability of two concrete types showed a large dif-
ference for both cases 28 and 56 days measurements. Reinforcing concrete obtained
lower results than that of the plain one. In other words, the concrete with embedded
rebar achieved a better quality compared to the one without reinforcement. This dif-
ference might be caused by the establishment of microcracks due to expansion (early
stage), and microcracks induced by shrinkage. Association with expansion phase, the
Air permeability coefficients of expansive … 565

reinforcement restrained concrete to not freely expand, which in turn, might reduce
cracks [20]. In term of shrinkage due to drying period, as a result of volume change
caused by water loss, concrete shrank. The bonding property between concrete and
reinforcement provided to restrain shrinkage [23-25].

4 Conclusion

Torrent air permeability tester was used in this work to investigate the different
quality of concretes with/without rebars. A conclusion is withdrawn as follows:
- The concrete cover confined by rebars obtained lower air permeability coeffi-
cients than that of the plain one without rebars which was indicated elaborately from
previous explanation. In other words, the reinforcing concrete cover achieved a better
quality than that of the concrete without reinforcement.
Further studies should pay attention to varied concrete cover thicknesses, dimen-
sions of rebars, and reinforcement arrangement.

References
1. P. Paulini, Fachiri Nasution, Air permeability of near-surface concrete. Conference:
CONSEC 07, June 2007.
2. A. Meyer, Importance of the Surface Layer for the Durability of Concrete Structures, ACI,
SP 100-5.
3. R. J. Torrent, A two-chamber vacuum cell for measuring the coefficient of permeability to
air of the concrete cover on site, Mat. Struct, 1992, 25, 358-365.
4. S. Nishio, Simple evaluation of water permeability of cover-concrete using a Water Spray
Method, Quarterly Report of RTRI, 2017.
5. W. Elkey, E. J. Sellevold, Electrical resistivity of concrete, 1995.
6. R. Torrent, A method for the rapid determination of the coefficient of permeability of the
coverconcrete, NDT-CE, 26. 28. 09. 1995.
7. Lulu Basheer, Assessment of the durability of concrete from its permeation properties: a
review, Constr. Build. Mats, 2001, 93-103.
8. P. Kummar Mehta Paulo J.M. Monteiro, Concrete structure, properties, and materials,
Third ed, 2006:394-395.
9. A.A. Ramezanianpour, Effect of curing on the compressive strength, resistance to chlo-
ride-ion penetration and porosity of concretes incorporation slag, fly ash or silica fume,
Cem. Concr. Compos. 17 (1995) 125–133
10. G.J. Osborne, Durability of Portland blast-furnace slag cements concrete, Cem. Concr.
Compos. 21 (1999) 11–21
11. Ha-Won Song, Studies on the corrosion resistance of reinforced steel in concrete with
ground granulated blast-furnace slag—An overview, Jour. Haz. Mats. B138, 2006, 226-
233.
12. Erdog˘an Özbay, Utilization and efficiency of ground granulated blast furnace slag on
concrete properties – A review, Const. Build, Mats. 105 (2016), 423-434.
13. J. Yuan, Effect of Slag Cement on Drying Shrinkage of Concrete, ACI Material Journal,
March-April 2015.
566 L. A. Van et al.

14. D. Hooton, The effect of ground granulated blast furnace slag (slag cement) on the drying
shrinkage of concrete – A critical review of the literature, Jan 2004.
15. Total shrinkage, oxygen permeability, and chloride ion penetration in concrete made with
white Portland cement and blast furnace slag.
16. Slag Cement in Concrete and Mortar, ACI 233R-03.
17. T. Kanda, Shrinkage cracking resistance of Blast furnace slag – blended cement concrete.
Influencing factors and enhancing measures, Journal of advanced concrete technology,
2015, vol. 13, 1-14.
18. M. Kokubu, Use of expansive components for concrete in Japan, ACI SP 38 (1973) 353–
378.
19. C.E. Kesler, D.W. Pfeifer, Report of ACI Committee 223, Expansive cement concrete –
present state of knowledge, ACI J. Proc. 67 (8) (1970) 563–610.
20. S. NAGATKI, Recommended practice for expansive concrete, concrete library of JSCE
No. 23, June 1994.
21. M. S. Meddah, Influence of a combination of expansive and shrinkage-reducing admixture
on autogenous deformation and self-stress of silica fume high-performance concrete, Con-
struction and Building Materials 25 (2011) 239–250.
22. N. V. Hoang, Applicability of a simplified estimation method to steam-cured expansive
concrete, Cement and Concrete Composites Volume 95, January 2019, Pages 217-227.222.
23. J. Zhang, Influence of reinforcing bars on shrinkage stresses in concrete slabs, Journal of
Engineering mechanics, 2000.
24. C. Raymond Nowacki, Effect of concrete shrinkage on stresses in reinforcing steel, master
thesis, 1958.
25. Hatt, W. K., Researches in concrete, Bulletin No. 24 Engineering Experiment Station, Pur-
due University, 1925, p.8.
26. R. J. Torrent, A two-chamber vacuum cell for measuring the coefficient of permeability to
air of the concrete cover on site, Mat. Struct, 1992, 25, 358-365.
27. M. Romer, Effect of moisture and concrete composition on the torrent permeability meas-
urement, Mater. Struct., 38 (5) (2005), p. 541.
28. P.A.M. Basheer, É. NolanNear-surface moisture gradients and in situ permeation tests,
Constr. Build. Mater., 15 (2–3) (2001), pp. 105-114.
 Application of Six Sigma on METRO Rail Construction
Project

Viraj Parekh1 , Kushal Solanki1 and Nishit Prajapati1

1 Civil Engineering Department, Institute of Technology, Nirma University, Ahmedabad, India

Abstract. Six Sigma is a Quality Improvement Technique widely used in Man-

ufacturing Industry. Application of Six Sigma is yet to be explored exhaustively
in the field of Construction. This study explores the possibility of application of
Six Sigma in METRO Rail Construction. This paper defines and analyses Con-
struction Performance of Pier Construction in METRO Rail Project of Ahmed-
abad using Six Sigma Technique. A stretch of 4 km from North-South Corridor
is selected for the study from APMC to Shreyas. The data such as Work Break-
down Structure, List of Activities, Time of Activities, Dependency of Activi-
ties, Construction sequence etc. is collected and analysed using software Pri-
mavera and Minitab. Process Capability Analysis is used to determine how well
an actual process meets a set of specification limits in a scheduled plan. Data is
represented and analysed by preparing different charts and graphs such as I –
Chart, Moving Range Chart, Capability Histogram and Capability Plot. A com-
parison was done between the actual works done at the site and the schedule
planned for the progress to calculate the Sigma value. It is found from the result
that the calculated Sigma value is near to the ultimate Sigma Level. This work
provides valuable insights for the implementation of Six Sigma Technique in
the Construction Industry. Six Sigma Technique can evaluate the quality of cur-
rent construction activity and quantify the improvement goals so as to control
succeeding critical activities for the project. Improvement in Quality of Con-
struction can be observed at large extent with the application of Six Sigma
Technique. Six Sigma Technique can be linked with Lean Principle.

Keywords: Six Sigma, Construction Project Management, Quality Manage-

ment, Lean Construction

1 Introduction

The Six Sigma concept is derived from bell-shaped curve. Theoretically, this bell-
shaped curve has been proven very useful as numerous natural continuous phenomena
seem to follow it or can be approximated by it. [1] Six Sigma is used in Manufactur-
ing Industry since last two decades to monitor and improve the quality aspect in the
process. The Six Sigma concept emphasis on improving the Process Performance
rather than focusing on final output only. Process Performance can be evaluated by
DPMO (defects per million opportunities). To achieve the sigma level at six, the value
of DPMO is 3.4. [5]

© Springer Nature Singapore Pte Ltd. 2020 567

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_89
568 V. Parekh et al.

The application of Six Sigma in Manufacturing Industry has been proven quite use-
ful. So it was applied in Construction Industry as well, but at small scale projects such
as a residential building to improve the quality of internal finishing work [4]. Six
Sigma principle was applied in lightweight partition walls also to determine and im-
prove the key input variables affecting the cracks [2]. Application of this technique is
yet to be explored in huge Infrastructure Projects like Metro Rail Construction. There-
fore, to assess the application of Six Sigma at this level, Ahmedabad Metro Rail Pro-
ject was selected for the research purpose.
Ahmedabad Metro is an under construction mass transit system being built for the
Ahmedabad City in the state of Gujarat, India. It is being built by Gujarat Metro Rail
Corporation (GMRC) Limited; a Special Purpose Vehicle constituted to implement all
the Metro Projects in the state of Gujarat. The Ahmedabad Metro Rail Project is di-
vided in Phase-I and Phase-II. Phase-I have the total span of 39.259 km which is the
combination of underground tunnel construction (nearly 6km length) and elevated
construction (nearly 32km). The North-South Corridor and East-West Corridor of the
Metro Rail will connect four corners of the Ahmedabad City with 32 stations. For the
research purpose, 4 km long stretch in the East-West Corridor was selected from
APMC to Shreyas.

2 Methodology

DMAIC i.e. Define, Measure, Analyse, Improve and Control is an integral part of
Six Sigma Process for continuous improvement. DMAIC is a five-step process where
one has to define the key variables in process at first place, measure the process per-
formance at second place, analyse the data collected and find out the root- cause at
third place, improve the process after rectifying the root-cause at fourth place and
finally to control the improved process. Application of Six Sigma starts with DMAIC
methodology to find out the value of DPMO.[3]
In contrast to the manufacturing industry, construction work is characterised by its
fragmented and project orientated work process. Manufacturing Industry observes
repetitive work but Construction is unique from Project to Project. To assess the qual-
ity of construction project, the triangle of project management highlights on three
aspects: Time, Cost and Scope of the project. Time overrun leads to cost overrun and
overall the quality of entire project is suffered. We have consider time as an important
aspect and study of critical path of this project was done. Based on this study, three
important activities were identified having float value zero. These activities are Piling,
Construction of Pier and Construction of Expansion Joint.
Piles constructed in Metro Projects are about 25-30m in depth. Thus, it is difficult
to find defects after the construction of a pile is done. Construction of Expansion
Joints are scheduled after the launching of girders which was yet to be done. Pier
construction was the only activity on which this study could be done. Pier defects can
be easily diagnosed as they are visible. In this study, original schedule and actual time
of construction of pier is taken to measure the process performance. Planned data was
available and actual data was collected by onsite visits. To analyse the data, Process
Application of Six Sigma on METRO Rail … 569

Capability Analysis was carried out using Minitab Software. Process Capability
Analysis is an important technique used to determine how well a process meets a set
of specification Limits.

3 Data Analysis

A Capability Analysis is done on the collected data and it produces an estimate of

the DPMO (defects per million opportunities), one or more capability indices and an
estimate of the Sigma Quality Level at which the process operates.
Process capability index (e.g., Cpk and Cp) or Process performance index (e.g.,
Ppk and Pp) can be found out from the process capability analysis. The output of this
measurement is usually illustrated by a histogram and calculations of the overall per-
formance of the process will predict how many parts will be produced within the
specification of the upper limit and lower limit.
The data of original duration and actual duration was analyzed for 129 piers in the
selected stretch of APMC to Shreyas. The histogram represents the distribution of
sample data. The frequency of data within an interval is shown by each bar on the
Histogram. From Fig.1 It is observed that Planned Construction Days of various piers
varied from 3 to 7 days having median as 4.62 Days. Fig.2 gives information about
actual days required to construct a pier. It is observed that median of actual days re-
quired to construct a pier is 4.853 which is marginally higher than the planned sched-
ule. It is mainly because of the operational problem faced while construction.

Fig. 1. Histogram of Original Duration Fig. 2. Histogram of Actual Duration

It is important to have I-chart which plots individual observations. The estimate of
the process average is represented as the center line. Certain amount of variation in
the individual sample value is expected. That is presented by the Control Limits. Con-
trol Limits are set at three standard deviations above and below the center line. Fig.3
Shows results for scheduled Days required for an individual pier and Fig.4 shows
actual days taken for the construction of an individual pier.
570 V. Parekh et al.

Fig. 3. I–MR Chart of Original Duration

Fig. 4. I–MR Chart of Actual Duration

The MR chart represents the moving ranges for the consecutive observations. Fig.3
and Fig.4 represents moving range for an original duration and actual duration respec-
tively.
In Fig.5 comparison of original days and actual days of construction is shown. A
capability analysis is based on a sample of data taken from a process and it produces:
An estimate of the DPMO, capability indices such as Cpk, Cp and an estimate of the
Sigma Quality Level at which the process operates. Fig.6 shows the Process capabil-
ity analysis. Lower Specification Limit (LSL) and Upper Specification Limit (USL)
is decided on original days required and tolerance of 0.5 days is provided for the
problems that may arise due to the operational problem while construction. As shown
in Fig. 6, DPMO for LSL is 5.33 and USL is 253.57. It suggests 5 sigma level of con-
struction which is considered above expectation. Cp value is found out to be 1.85 and
Cpk value is 1.63. Cpk is a better estimate of the potential for the process. It repre-
sents the best a process can do and measures how close a process is running to its
specification limits. A Cpk of 1.33 (4 sigma) or higher is most satisfactory for the
entire process.
Application of Six Sigma on METRO Rail … 571

Fig. 5. Comparison of Original Days and Actual Days

Fig. 6. Process capability analysis

4 Results and Discussion

This work comprehend the application of six sigma principle in construction indus-
try. The Principle can be applied to huge construction projects such as metro rail con-
struction to evaluate the quality of the current operation and to find out the scope of
improvement. Improvement through the six sigma principle can be large if it is ap-
plied to the more complicated projects.
On the basis of analysis of data in which the six sigma technique is applied in met-
ro construction project, the sigma level of construction activity for pier is calculated
by using Capability Analysis. Various charts regarding scheduled activity and actual
work done at site is compared. The sigma value is found out to be 5 which is near to
the ultimate sigma level, so the pier construction activity as per planning and quality
assurance checklists are satisfied up to higher level. The DPMO (defects per million
572 V. Parekh et al.

objectives) can be find out from the sigma table which is near to 258 which means
there can be 258 defects in one million opportunities. The Histogram, capability plot
and moving range charts are also found out.

5 Conclusion

Compared to manufacturing industry, construction industry is different because it

is less repetitive and outdoor in nature. Most of the time, materials are being counted
responsible for the poor quality of construction but focus should be drawn towards the
work flows and overall process. In present study, methodological and statistical ap-
proach is used to determine work flow in terms of time duration. As of now no re-
search is available in the field of application of six sigma in metro rail construction so
this study helped us to apply well known concept in comparatively new field of work.
The Six Sigma methodology not only helps in identifying the defects but it pro-
vides practical ways to control the quality variances. This technique can be applied
for each construction activity at micro level of major project for project improvement
and process improvement. In future it can be very useful statistical tool. By combin-
ing all the data and applying it as integrated system, helps to nullify defects by con-
trolling each activity of whole project. By doing this six sigma level can be achieved.

Acknowledgement

Authors of the paper are grateful to Nirma University for providing encouragement
and all kind of support for this research work.

References
1. Aboelmaged, M. G. (2010). “Six sigma quality: A structured review and impli-
cations for future research.” International Journal of Quality & Reliability
Management. 27(3), 268–317.
2. Kuo-Liang L., Yang S. (2013)“Applying Six Sigma to Quality Improvement in
Construction”, Journal of Management in Engineering, 29(4): 464-470
3. Sarathkumar K., Loganathan R. (2016) “Evaluation of Six Sigma Concepts in
Construction Industry”, International Journal of Scientific & Engineering Re-
search,7(4): 202-207
4. Sawant S., Pataskar S. (2014) “Applying Six Sigma Principles in Construction In-
dustry for Quality Improvement”, International Journal of Scientific & Engineering
Research, 117(4): 129-139
5. Seung H., Myung C., Keon S., Ho D. (2008) “Six Sigma-Based Approach to Im-
prove Performance in Construction Operations”, Journal of Management in Engi-
neering, 24(1): 21-31
 Experimental investigation of loading rate effects on the
shear capacity of reinforced concrete deep beams

La Vanny1, Kenichiro Nakarai1, Halwan Alfisa Saifullah1, 2, Asuka Mizobe 1

1
Department of Civil and Environmental Engineering, Graduate School of Engineering
Hiroshima University, 1-4-1, Kagamiyama, Higashi-Hiroshima, Hiroshima 739-527, Japan
2
Department of Civil Engineering, Faculty of Engineering, Universitas Sebelas Maret, Sura-
karta, Indonesia
m185478@hiroshima-u.ac.jp (L. Vanny), nakarai@hiroshima-u.ac.jp (K.
Nakarai), halwan@ft.uns.ac.id(H.Alfisa Saifullah), m184098@hiroshima-
u.ac.jp (A. Mizobe).

Abstract. Unlike slender RC beams, shear failure mechanism of RC deep beams

is mainly governed by shear compression; a considerable amount of the load is
carried by the compression struts of concrete. For this reason, when subjected to
a long-term sustained load, the shear strength of RC deep beam could be reduced
as the sustained load might have negative impacts on the compressive strength of
concrete. The purpose of this study is to investigate the influence of sustained
loads on the shear capacity of RC deep beams. In this experiment, different load-
ing rates were chosen as an alternative method to clarify the effect of sustained
loads. Two main different types of loading rate, a normal loading rate of 10
mm/hr and a very slow loading rate of 0.01 mm/hr, were applied. 100x320x1200
mm RC beams with a ratio of shear span to the effective depth a/d=1 were tested
under two-point loading. The test results showed that under very slow loading
rate, the ultimate shear capacity was moderately reduced.

Keywords: Sustained load, Loading rate, RC deep beam, Shear capacity.

1 Introduction

Some underground structures such as cut-and-cover duct, tunnels, and retaining walls
are generally subjected to a sustained loading. During service life, their capacity to re-
sist the external loads could be affected. The study of this topic was credited to the
pioneer researcher Rüsch who studied the effect of sustained loads and slow loading
rates on compressive strength of concrete and flexural capacity of RC beams and the
results showed that sustained loads and slower loading rates decreased the ultimate
compressive strength of concrete and increase concrete strain [1]. Since then, extensive
research has been done to investigate the influence of sustained loads on concrete and
concrete members [2,3]. However, only few studies were examined this effect on the
shear performance of reinforced concrete beams.

© Springer Nature Singapore Pte Ltd. 2020 573

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_90
574 L. Vanny et al.

In fact, there was a case of one tunnel constructed in 1982 that was in deteriorating
condition with cracks. Those cracks were suspected to be due to creep shear failure
under high levels of sustained loads [4]. The creep shear failure has been further inves-
tigated for RC slender beam without shear reinforcement [5-7]. Sarkosh et al.,(2013)
conducted experiments on RC slender beam with shear span to effective depth ratio a/d
>2.5 and revealed that the sustained load has no significant effect on the shear capacity.
Similarly, Safiulah et al.,(2016, 2017) performed experiments of slender beam (a/d
>2.5) under different loading rates instead of constant sustained loads and reported that
creep strain might or might not affect the shear capacity depending on the time available
for structure to respond to the loads and the negative effect of shear creep might com-
pensate with the increase contribution of aggregate interlock.
While most studies focused on RC slender beams (a/d >2.5), one exceptional study
was carried on effects of sustained loads on large scale RC deep beams. The effect of
creep on shear capacity was numerically analyzed from the creep damage of diagonal
compression strut over the web concrete. The study suggested the design safety factor
for shear should be 1.2-1.3 for avoiding the risk of creep failure against long-term sus-
tained loading [8]. Shear failure mechanism of RC deep beams is complex and failure
mode of RC deep beam is mainly governed by shear compression. For this reason, when
subjected to a long-term sustained load, the shear strength of RC deep beam could be
reduced.
Therefore, this study aimed to experimentally investigate the influence of sustained
loads on the capacity of RC deep beams by applying slow loading rates.

2 Experimental Program

2.1 Materials and mix proportions

In this study, a normal-strength concrete with water-cement ratio (W/C) of 0.65 was
adopted. A chemical admixture was used to enhance the workability of the fresh con-
crete. The detail of mix proportion is listed in the Table 1.

Table 1. Mix proportions

s/a Unit mass (kg/m3)
Air content Slum W/C
(%) W C S G AD
(%) (cm)
4.5±1.5 8±2 0.65 48.9 170 262 894 933 SV10L*
a: total weight of aggregate, AD: admixture. *SV10L: AE water-reducing admixture.

2.2 Test specimens

The details of the RC beams are shown in Fig.1. The ratio of shear span to effective
depth (a/d) of RC beam equals to 1. To induce the shear failure before yielding of rein-
forcement, high-strength deformed steel bar (fy =1062 N/mm2) with tension reinforce-
ment ratio of 1.96% was used. The RC beams and cylindrical specimens were prepared
and sealed using aluminum tape then were cured under normal room temperature. The
Experimental investigation of loading rate effects … 575

seal was removed after 28 days such that the drying shrinkage was accelerated and
remained stable before loading tests. This could minimize the effect of drying shrinkage
during testing at loading age.

2.3 Test setup and loading condition of RC beams

Test setup. The details of the RC specimens and measurement system installations are
illustrated in Fig. 1. The beams were tested under two-point loads acting at 100 mm
from center line on each side. Under both applied loads, 10 mm thick steel plates with
the dimension of 100mm x 150mm were used to distribute the load to the beam. Each
beam was instrumented to measure the vertical deflections at mid-span, at point loads,
and at the supports using the displacement transducer.
A A-A’

Stirrup, D10

D25, tension
reinforcement A’ Reinforcement strain gauge
ρ=1.96%

Fig. 1. Detail of RC beams and measurement set up

Shear displacement transducer, namely the shear frame, made up of vertical, hori-
zontal and diagonal transducers were installed in the shear span of the RC beam to
measure the diagonal crack widths. Due to possibility of multiple cracks along the beam
span, the total of 7 π-shaped displacement transducers were placed at every 100 mm
interval to measure the crack widths. Reinforcement strain gauges were installed at 3
predetermined locations along reinforcement bar.

Loading condition. In this experiment, slow loading rate was chosen as an alternative
method to clarify the effect of sustained loads. This method was adopted from the study
performed by Safiulah et al., [6, 7]. A loading rate of 10mm/hr was selected to represent
the beam specimens under normal loading condition while the loading rate of
0.01mm/hr was represented the specimens under very slow loading. Prior to application
of a very slow loading rate, transition path was introduced with loading speed of 0.1
mm/hr from 400kN to 420kN. The detail of the loading condition for each RC beam is
shown in the following Table 2.

Table 2. Loading condition for each RC beam

Displacement control Load control
No. Normal Rate Slow Rate Very Slow Rate Very Slow Rate
(10mm/1hr) (0.1mm/hr) (0.01mm/hr) (0.60kN/hr)
576 L. Vanny et al.

RC 1 0kN to failure - - -
RC 2 0kN to 400kN 400kN to 420kN 420kN to failure -
RC 3 0kN to 400kN 400kN to 420kN 420kN to failure

3 Results and Discussion

3.1 Concrete properties

The test results of the mechanical properties of concrete are shown in Table 3.

Table 3. Mechanical properties of concrete

Age (days) 1 3 7 28 91 160 (loading )
Compressive Strength (MPa) 3.9 11.3 24.1 33.6 35.7 34.6
Young's Modulus (kPa) 12.3 21.6 30.3 34.1 29.2 28.6
Splitting Tensile Strength (MPa) - 1.30 2.34 2.90 3.18 2.98

3.2 Relationship between load and mid-span deflection

The load versus average mid-span deflection of all RC beam specimens is shown in
Fig. 2, and it is observed that all RC beams had similar behavior prior to the application
of the very slow loading rate.
RC1-NL beam represented the specimen loaded with normal loading rate while
RC2-VS.D420 beam and RC3-VS.L420 beam represented the specimens loaded with
a very slow loading rate.
600
Applied
500 very slow rate
400
Load (kN)

300
RC1-NL
200
RC2-VS.D420
100 RC3-VS.L420

0
0 1 2 3 4 5 6
Avg.mid-span deflection (mm)
Fig. 2. Load and average mid-span deflection curve

Under a very slow loading rate, RC2-VS.D420 beam and RC3-VS.L420 beam
showed a change in the load and mid-span deflection curve after a very slow loading
rate was applied at 420 kN. In addition, the increase in deflection at the same load level
was clearly showed the creep effect due to a very slow loading rate. The reduction of
ultimate failure loads was also observed in both RC beams.
Experimental investigation of loading rate effects … 577

3.3 Shear capacity of RC beams

Comparison between measured load values and calculated load values. The diago-
nal cracking load is referring to the load at which the significant diagonal crack oc-
curred in the shear span of the RC beam. The diagonal cracking load was determined
from a combination of visual crack observations, the load and mid-span deflection re-
lationship, the load-diagonal displacement relationship, and the load-vertical displace-
ment relationship.
The flexural cracking load (Pcr_cal ) was calculated from the cracking moment at mid-
span whereas the diagonal cracking load (Pdiag_cal) and ultimate failure load (Pu_cal) were
calculated by Niwa’s equations (1) and (2), respectively.
The results indicated that for all RC deep beams, the measured load values of the
flexural cracking loads were smaller than calculated load values and this reduction
might be the effect of shrinkage after seal removal from the specimen. While the diag-
onal cracking loads and ultimate failure loads showed higher load values than those
given by Niwa’s equation as listed in the Table 4.

Table 4. Comparison of measured load values with calculated load values.

Flexural cracking load Diagonal cracking load Ultimate failure load
Specimen Pcr/Pcr_cal Pdiag/Pdiag_cal
Pcr (kN) Pdiag (kN) Pu (kN) Pu/Pu_cal
(kN) (kN)
RC1-NL 34.67 0.54 272.44 2.17 571.87 1.54
RC2-VS.D420 31.07 0.48 251.84 2.01 515.29 1.39
RC3-VS.L420 27.31 0.42 225.84 1.80 554.54 1.49
* Pcr_cal : 64.60 kN; Pdiag_cal : 125.40 kN ; Pu_cal : 371.18 kN.

Pdiag , cal  0.20 f c'1/3 (d / 1000)1/ 4 (100 s )1/3 (0.75  1.4 / (a / d ))b d (1)
Pu ,cal  0.24 f c'2/3 (1  3.33r / d ) (1  (100 s )1/ 2 (1 / (1  (a / d )2 ))bd (2)
Where: fc’: concrete compressive strength (N/mm2); ρs: tension reinforcement ratio
a: shear span length (m); b: width of beam (m); d: effective depth (m)
r: width of steel plate (mm)

Effect of a very slow loading rate on shear capacity of RC deep beam. The compar-
ison of the ultimate failure load between RC beam under a normal loading rate and
under a very slow loading rate is presented in the Table 5. The result showed that the
average ultimate failure load of the RC beams under a very slow loading was reduced
with ratio of 0.94 comparing to that of RC beam under a normal loading rate.

Table 5. Decrease in ultimate loads under very slow loading rate

Decrease in the ultimate loads
Loading rate Specimen avg.Pvu/avg.PvuN
Pvu (kN) avgPvu (kN)
Normal (10mm/hr) RC1-NL 571.87 571.87 1.00
Very slow RC2-VS.D420 515.29
534.91 0.94
(0.01mm/hr) RC3-VS.L420 554.54
578 L. Vanny et al.

Under very slow loading rate, RC beams exhibited the average increase in value of mid-
span deflection, shear displacement at failure side, total crack widths and stain of mid-
span steel rebar with a ratio of 1.35, 2.53, 1.42 and 1.07 respectively. This calculation
was performed for a load interval between 420 kN to 515 kN.
This result confirmed the study conducted by Safiulah et al., [6, 7] which reported
that the effect of a very slow loading rate (1000 times slower than a static loading rate)
on shear capacity seemed to be more pronounced with decreasing ratio a/d of the beams.
Unlike RC slender beams (a/d>2.5) which have failure mode mainly governed by shear
tension, RC deep beams (a/d<2) showed a different behavior as they failed in shear
compression. For this reason, the negative impact on compressive strength of concrete
under very slow loading rate affected the capacity of RC deep beams.

4 Conclusion

This experimental program was performed to investigate the influence of sustained

loads on the shear performance of RC deep beams by applying a very slow loading
rates. Based on the results of this study, the following conclusions can be drawn:
-Under a very slow loading rate, the compressive strength of concrete was af-
fected. As the failure mechanism of a RC deep beam mainly governed by shear com-
pression, its ultimate shear capacity was slightly reduced.
-The increase in shear displacement, mid-span deflection, total crack width and
the strain of mid-span steel bar clearly showed the creep effect on RC deep beams.
- Displacement control system was critical in case of RC deep beam loaded with
a very slow loading rate.

References
1. Rüsch, H.: Research toward a general flexural theory for structural concrete. Journal of ACI,
57(7), 1-27 (1960).
2. Higgins, L., Forth, J. P., Neville, A., Jones, R. and Hodgson, T.: Behavior of cracked rein-
forced concrete beams under repeated and sustained load types. Engineering Structure, 56,
457-465.
3. Tasevski, D., Ruiz, M., F. and Muttoni, A.: Behavior of concrete in compression and shear
under varying strain rates: from rapid to long-term action. In: 11th Feb International PhD
Symposium in Civil Engineering, Tokyo, 881-886 (2016).
4. Chijiwa, N, Zhu, X., Ohno, H., Tanabe, S., Nakarai, K. and Maekawa, K.: Delayed shear
crack formation of shallow RC box culverts in service. CONCREEP 10, 1579-1586 (2015).
5. Sarkhosh, R., Walraven, J. C. and den Uijl, J. A.: Time-dependent behavior of cracked con-
crete beams under sustained loading. In: FraMCoS-8 (2013).
6. Safiullah, H. A., Nakarai, K., Piseth, V., Chijiwa, N. and Maekawa, K.: Effect of loading
conditions on shear creep failure of RC beams. In: 11th Feb International PhD Symposium
in Civil Engineering, Tokyo, 125-132 (2016).
7. Safiullah, H. A., Nakarai, K., Piseth, V., Chijiwa, N. and Maekawa, K.: Shear creep failures
of reinforced concrete slender beams without shear reinforcement. Journal of ACI Structure,
114(6) 1581-1590 (2017).
8. Bugalia, N. and Maekawa, K.: Tune-dependent capacity of large-scale deep beams under
sustained loads. Journal of Advanced Concrete Technology, 15, 314-327 (2017).
Fatigue characterization of conventional and high rutting
resistance asphalt mixtures using the cyclic indirect
tensile test

H.T. Tai Nguyen1,2, Anh Thang Le1, Vu Tu Tran1 and Duy Liem Nguyen1
1
Ho Chi Minh City University of Technology and Education, Ho Chi Minh City, Vietnam
2
BMT Construction Investment J.S.C., Ho Chi Minh City, Vietnam
tainht@hcmute.edu.vn; thangla@hcmute.edu.vn; tutv@hcmute.edu.vn;
liemnd@hcmute.edu.vn

Abstract. In this paper, the fatigue behavior of several asphalt mixtures that are
widely used in Vietnam such as the dense graded asphalt concrete (AC) pro-
duced with 60/70 pen bitumen and high rutting resistance asphalt mixtures
(HRRA) are investigated using the cyclic indirect tensile (IDT) testing device
equipped at Ho Chi Minh City University of Technology and Education
(HCMUTE). Several level of stress magnitudes were applied to determine the
characteristic fatigue line (CFL) of mixtures. The results show that the widely
used power law of CFL is applicable for all investigated mixtures. Although the
stiffness of the HRRA is twice that of conventional AC, its resistance to fatigue
is much higher. The effect of strain magnitude on the stiffness of mixtures or
the nonlinearity effect was also recognized. It was observed that the impact of
binder type on the nonlinearity effect is insignificant compared to that due to
the volumetric design of mixture.

Keywords: Asphalt mixtures, Indirect tensile fatigue test, Characteristic fatigue

line, Nonlinearity effect.

1 Introduction

In recent years, the system of roadway in Vietnam has to serve a large number of
heavy vehicles like tandem, tridem and 18-wheel trucks due to the increasing demand
of the economy. Consequently, pavement distresses, e.g. rutting and cracking, have
occurred in many highways and principal roads of Vietnam. Since rutting distress
normally appears prematurely after opening to traffic, it has attracted much attention
of local authorities and researchers [1,2,3,4,5]. While fatigue cracking of asphalt
pavement is considered as a serious distress in many countries, it has not been proper-
ly considered yet in the national standard for design and construction of asphalt mix-
tures (e.g. TCVN 8819-2011), giving rise to unreliable resistance to fatigue of pave-
ment structures. Furthermore, the use of coarser asphalt mixture with very little binder
content in an attempt to prevent rutting distress may result in a significant reduction
of fatigue life. The use of HRRA like SBS-modified, plastic-modified crumb rubber

© Springer Nature Singapore Pte Ltd. 2020 579

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_91
580 H. T. T. Nguyen et al.

modified asphalt mixtures without controlling its resistance to fatigue also has poten-
tial risks of cracking in the near future.
The fatigue characterization of asphalt concretes is usually performed using cyclic
bending or indirect tensile tests with the aid of dynamic universal testing machine
(UTM). There are only two UTMs for testing asphalt mixtures in Vietnam―namely,
the Cooper pneumatic UTM at the University of Transport and Communication
(UTC), and the Pavetest hydraulic UTM at the HCMUTE. Owing to the lack of test-
ing instrument, there is not much research on the fatigue behavior of asphalt mixtures.
Attempts have been made to predict the fatigue life of 60/70 pen AC pavement struc-
ture based on the results of 4-point bending fatigue test using either the analytical or
numerical method [6,7,8]. However, the prediction results have not been yet com-
pared to onsite measurements of fatigue cracking to validate the proposed models.
Much more attempts should be performed to prevent fatigue cracking in the near fu-
ture by examining the correlation between laboratory fatigue testing results and onsite
cracking of asphalt pavement, controlling the resistance to fatigue of asphalt mixture
and developing a reliable pavement thickness design.
This paper reports a very preliminary study on the fatigue behavior of asphalt mix-
tures that are widely used in Vietnam, especially the HRRA mixture, using the IDT
testing method. The results of this work will contribute to the understanding of fatigue
behavior of asphalt mixtures as well as the numerical prediction of fatigue life of
pavement structure which is in progress and will be presented in future works.

2 Materials and methods

2.1 Materials

In this experimental research, we will investigate the indirect tensile fatigue behavior
of asphalt mixtures that are widely used in Vietnam such as the dense graded AC
produced with 60/70 pen and high rutting resistance bitumen. The dense gradation
with nominal maximum particle size of 12.5 mm was chosen. All aggregate, filler and
binder are provided by BMT Construction Investment J.S.C. The physical properties
of aggregate and 60/70 bitumen are the same as those reported in [3]. The high rutting
resistance bitumen was produced by modifying the neat bitumen with some polymers
such that the performance grade (PG) of the modified bitumen attains at least PG88.
The design of asphalt mixtures was performed based on the Marshall method and
followed the instructions of [1] and TCVN 8820-2011. The fundamental properties of
these mixtures are summarized in Table 1. For the sake of simplicity, the sample’s
names are hereafter coded as D8S, D9S standing for AC produced with 60/70 pen
bitumen and using gradation curves of [1] and TCVN 8820-2011, respectively.
HRRA samples are coded as D9H and has the gradation curve following TCVN 8820-
2011.
Fatigue characterization of conventional … 581

2.2 Methods

Cylindrical samples of asphalt mixtures are prepared by either the Marshall com-
pactor or cored from Roller Sector Compactor slabs. The dimensions of the compact-
ed samples are 101-mm diameter and 63.5-mm height while those of the cored sam-
ples are 98.5-mm diameter and 62.5-mm height. Before evaluating the resistance to
fatigue, indirect tensile strength of the samples was measured based on ASTM
D6931-17. The resistance to fatigue of asphalt mixture samples was evaluated by the
Pavetest hydraulic UTM equipped at HCMUTE and the method used is EN 12697-24,
Annex E (Indirect tensile test on cylindrical shaped specimens).
The temperature and frequency are fixed at 25˚C and 10Hz, respectively. Before
testing, the mixtures were held in the conditioning chamber for at least 4 hours to
maintain a homogenous temperature field inside the sample. Several level of stress
magnitudes were applied while a small constant stress is maintained during the test to
keep a positive contact between the loading frame and the sample. Only one replicate
of test is performed in this study.

Table 1. Fundamental properties of asphalt mixtures used in this research.

Properties | Asphalt mixtures D8S D9S D9H

Maximum specific gravity 2.52 2.56 2.54
Bulk specific gravity 2.37 2.36 2.42
Air void (%) 5.99 7.76 4.68
Void in aggregate (%) 16.12 16.33 14.87
Marshall stability (kN) 11.1 12.2 25.02
Marshall flow (mm) 3.7 3.3 2.51
Indirect tensile strength (IDTS) (MPa) 1.318 1.186 2.23

3 Results and discussions

The indirect tensile (IDT) stiffness is recorded during testing and the number of cy-
cles at which the sample loses 50% of its initial stiffness ( E0 ) is considered as the
fatigue life ( N f 50 ) of the sample. The curve of logarithm of the fatigue life versus
logarithm of the initial stress or strain represents the fatigue behavior of the sample
and is called the characteristic fatigue line (CFL). The experimental testing results
will be presented and discussed in next Subsections.

3.1 Evolutions of the nominal damage

Let us define the nominal damage of the sample as Dnom  1  En E0 , where E 0 , E n
is the stiffness of the sample at initial state (cycle 100) and at cycle n  100 . To re-
duce the time of testing, the test was forced to stop when the slope of the nominal
damage attains a stable value. The fatigue life of the sample was extrapolated using
linear regression method. Figures 1 and 2 show the evolutions of the nominal damage
582 H. T. T. Nguyen et al.

of D8S, D9S and D9H samples. Due to the oscillation of the captured data during the
test and the imperfect contact between the testing frame and the sample, the value of
stiffness at a cycle oscillates around its average value, which gives rise to small nega-
tive value of Dnom at the beginning of the test. The results verify that a larger initial
strain or stress in the sample will give rise to a shorter fatigue life of the sample. A
non-dimensional quantity defined as the ratio of stress to tensile strength was used to
plot the characteristic line of the fatigue behavior of mixtures and to rank its re-
sistance to fatigue. The fatigue life of asphalt mixture in function of the ratio of stress
to IDTS follows the power function law with high value of R 2 coefficient (see Fig.
3). It can also be observed that the exponent of the CFL of D9S and D8S mixtures is
almost identical and is larger than that of D9H mixture. This means that the resistance
to fatigue of D9H mixture is more sensitive to the ratio of stress to strength
( 0 / IDTS ) than that of D8S and D9S mixtures. When the stress is equal to the
IDTS, an extrapolation of the CFL shows that the fatigue life of the 3 mixtures are at
the same level of magnitude (~102 cycles). In case of  0 / IDTS  1 , the CFL of D9H
is always above that of D8S and D9S. This means that, for a loading condition at
which the applied stress inside the sample is still lower than the tensile strength, the
resistance to fatigue of D9H is better than that of D8S and D9S. It is worth to mention
that the stiffness of D9H and the tensile strength are almost twice those of D9S and
D8S, which means that, at the same thickness, the HRRA pavement will have better
resistance to fatigue than conventional AC pavement. This result shows that stiffer
asphalt mixtures may have better resistance to fatigue than softer ones. Consequently,
well designed HRRA mixtures should be used for preventing rutting distress of high
trafficked roadway without concerns about insufficient resistance to fatigue cracking.

0.7
61.0 46.1
Nominal damage

0.5 32.6 125.7

0.3

0.1

-0.1 0 10000 20000 30000

Number of cycles

Fig. 1. Evolution of the nominal damage of D8S samples at different level of initial strain.

3.2 Nonlinearity of asphalt mixtures

Based on testing results, it can be found that the initial IDT stiffness decreases linear-
ly with an increase of initial strain as presented in Fig. 4. This observation is in good
agreement with that observed by other researchers, e.g. [9]. Although the IDT stiff-
ness of D9H mixture is higher than that of D9S and D8S mixtures due to the use of
stiffer binder, the rate of the decrease of stiffness with an increase of initial strain is
almost similar—namely, 27.025 MPa/ for D9H mixture, 26.257 and 28.195 for
Fatigue characterization of conventional … 583

D9S and D8S mixtures, respectively. Because all mixtures are designed using the
dense gradation and are compacted at similar air void (4.68 to 7.76%) while the type
of binder used is totally different, it seems to be that the impact of binder type on the
nonlinearity effect is insignificant compared to that due to volumetric design of mix-
ture.

0.5 0.5
54.0 86.3 77.5 94.8

Nominal damage
37.3 46.2
176.6 42.54
Nominal damage

147.9 0.3
0.3

0.1
0.1
-0.1
-0.1 0 5000 10000 15000 0 5000 10000 15000 20000 25000
Number of cycles Number of cycles

Fig. 2. Evolution of the nominal damage of D9S and D9H samples.

1.E+08
y = 1.11E+02x-3.66E+00 D9H
R² = 9.81E-01 D9S
1.E+06 D8S
y = 4.51E+01x-6.77E+00
Nf50

1.E+04 R² = 8.72E-01
y = 7.84E+01x-3.45E+00
1.E+02 R² = 9.75E-01

1.E+00
0.1 1
0/IDTS

Fig. 3. Fatigue life versus stress level of D9S, D8S and D9H samples at 20˚C,10Hz.

30000 D9H
D9S
Initial stiffness (MPa)

D8S y = -27.025x + 22237

20000 R² = 0.8367
y = -28.195x + 11542
R² = 0.9968
10000
y = -26.257x + 10730
R² = 0.9831
0
0 50 100 150 200
Initial strain ()

Fig. 4. Effect of initial strain on initial IDT stiffness at 20˚C,10Hz.

584 H. T. T. Nguyen et al.

4 Results and discussions

The resistance to fatigue of conventional and HRRA mixtures is successfully charac-

terized using the cyclic indirect tensile test. Based on the results obtained in this
study, the following conclusions could be drawn:
 The widely used power law of CFL is applicable for all investigated mix-
tures including the conventional AC and HRRA.
 The use of a non-dimensional quantity defined as the ratio of stress to tensile
strength in CFL allow us to compare and rank the resistance to fatigue of
mixtures possessing different stiffness and tensile strength.
 The resistance to fatigue of HRRA mixture is much higher than that of con-
ventional mixtures when it is well designed. This result shows that softer as-
phalt mixture does not necessarily have better resistance to fatigue than stiff-
er one. Consequently, HRRA mixtures should be used for high trafficked
roadway to prevent both rutting and cracking distress.
 The effect of initial strain magnitude on the initial stiffness is significant and
should be considered in the modelling of the fatigue behavior of asphalt mix-
tures; and the impact of binder type on the nonlinearity effect is insignificant
compared to that due to the volumetric design of mixture.

References
1. Decision 858-MOT: Applying current technical standards to strengthen the quality of
management of design and construction of hot mix asphalt pavement of high traffic road-
way. Ministry of Transport, Hanoi. (2014).
2. Nguyen, H.T.T.: Modelling the mechanical behaviour of asphalt concrete using the
Perzyna viscoplastic theory and Drucker–Prager yield surface. Road Mater Pavement
18(S2), 264-280 (2017).
3. Nguyen, H.T.T., Tran, T.N.: Effects of crumb rubber content and curing time on the prop-
erties of asphalt concrete and stone mastic asphalt using dry process. Int. J. Pavement Res.
and Technol. 13(3), 238-244 (2018).
4. Tran, T.N. et al.: Semi-flexible Material: The Sustainable Alternative for the Use of Con-
ventional Road Materials in Heavy-Duty Pavement. Congrès International de Géotech-
nique – Ouvrages – Structures, pp. 17-24. Springer, Singapore (2018).
5. Le, A.T. et al.: Development of a permanent deformation model of asphalt mixtures for
South Korean pavement design guide. Transport Res Rec. (2095)1, 45-52 (2009).
6. Tran, T.L. et al.: Experimental study on the resistance to fatigue of asphalt concrete used
for pavement surface in Vietnam (in Vietnamese). Journal of Transportation (no 4) 2015.
7. Tran, T.L., Nguyen, M.L.: The studying of the resistance to fatigue of asphalt concrete in
Vietnam. Journal of Transportation (in Vietnamese) (no 6), 23 – 24 (2012).
8. Nguyen, H.T.T, Nguyen, N.H.: Using a Non-local Elastic Damage Model to Predict the
Fatigue Life of Asphalt Pavement Structure. Proceedings of the International Conference
on Advances in Computational Mechanics 2017. Springer, Singapore (2018).
9. Di Benedetto et al.: Nonlinearity, Heating, Fatigue and Thixotropy during Cyclic Loading
of Asphalt Mixtures. Road Mater Pavement 12(1), 129-158 (2011).
 Sustainable Measures to Achieve Better River-Crossing
Access in the Rural Areas of Developing Countries

Yoshinori Fukubayashi1[0000-1111-2222-3333] and Makoto Kimura2[1111-2222-3333-4444]

1
University of Miyazaki, Gakuen Kibanadai Nishi, 8892192 Miyazaki, Japan
2
Kyoto University, Katsura Campus, Nishikyo-ku, 6158540 Kyoto, Japan
fukubayashi@cc.miyazaki-u.ac.jp

Abstract. Due to financial and technical constraints of the governments of de-

veloping countries, rural road networks generally offer poor access. At stream-
crossing sections on rural roads, in particular, access may be cut off during the
rainy season because of high water flooding the bridge. Considering the socio-
economic situation and limited resources in rural areas, the authors proposed to
build road-stream crossings through community initiatives by mobilizing local
resources. Since then, several projects have been conducted. In order to assess
the practicability of improving access at road-stream crossings and the perfor-
mance of constructed structures, this paper presents case studies on projects in
which low-water bridges and vented fords were constructed. Then, we discuss
the performance of the structure and application criteria for construction, either
by the community themselves or in collaboration with local government and
community. It is demonstrated that road-stream crossing construction through
community initiatives provides better access to roadside communities and en-
hances feelings of ownership, such that locals feel responsible for performing
continuous maintenance and bridge safety.

Keywords: Access, Road-stream Crossing, Developing countries

1 Introduction

In developing countries, especially in rural areas, road networks are unreliable and not
always accessible due to their poor conditions. This limits rural access to hospitals,
schools, markets, and so forth, thereby hampering development of their livelihoods.
Due to the limited financial and technical resources of governments, rural roads have
left behind from development. Most rural roads are not well engineered, making rural
roads impassable during rainy seasons.
In order to improve the trafficability of earth or gravel roads, the authors have tak-
en a local resource-based approach that enhances community initiatives toward self-
development [1], [2]. With this approach, design and construction work need to be
practical for implementation by the community or through collaboration between
community and the local government.
To improve rural-road-network accessibility, measures for crossing streams are al-
so required. By taking the local resource-based approach, several projects for improv-

© Springer Nature Singapore Pte Ltd. 2020 585

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_92
586

ing access at road-stream crossings have been implemented by the authors as engi-
neers of an international NGO. The practicability and performance of road-stream
crossings constructed through the local resource-based approach must be assessed so
that this approach may be replicated and disseminated in other areas to extend access
to rural people in developing countries.
In this paper, projects for improving access at road-stream crossings in two coun-
tries are presented as case studies. Firstly, the design, construction methods, and im-
plementation structure are summarized. Secondly, the findings of attempts to monitor
the conditions of the crossings are reported. Then, the performance of the structure
and the application criteria for construction through collaborations with local gov-
ernments and communities are summarized.

2 Access-improvement projects at road-stream crossings based

on a local resource-based approach

2.1 Low-water bridge designed to be overtopped by high flows in Zambia

In Zambia, villagers are often hampered in their access to markets and hospitals due
to high water flows at road-stream crossing (Fig. 1 a)). Through field surveys and
discussions with local communities familiar with the condition of the stream during
rainy seasons, the crossing location was decided. Then, the design conditions were
decided, as shown in Table 1. A low-water bridge to be overtopped by high flows
(Fig.1b)) was designed and constructed by the communities with technical and finan-
cial support from the NGO, that the authors are belong to. The construction record is
shown in Table 2.
The abutments and pier were constructed with the embankment covered by a riprap

Table 1. Design condition for road-stream crossings

Items for the project description Zambia Philippines
Maximum Traffic 8ton trucks 4ton trucks
Traffic volume Less than 50 vehicle per day
Height of high-water level from the channel bottom 2.0 m 2.0 m
Height of driving surface from the channel bottom 1.5 m 1.5 m
Length of constructed road-stream crossing 10 m 20 m
Width of constructed road-stream crossing 3.5 m 3.0 m

Table 2. Record of the construction of the road-stream crossing

Items for the project description Zambia Philippines
Type of crossing Bridge with two spans Vented Ford
Material of bridge beam Reinforced concrete -
Construction period 72 days 37 days
Completion year, month 2015, June 2017, June
Construction cost (USD) 8,500 18,000
Cost per square meter (USD) 243 300
Sustainable Measures to Achieve Better River-Crossing … 587

Monitoring year, month 2017, October 2018, October

a) b)

c) d)
Fig. 1. Construction of the low-water bridge: a) before construction; b) after construction; c)
abutment and pier covered with riprap; d) casting concrete into the form of the beam

revetment (Fig.1c)). Then, the reinforced concrete beam and floor slab were con-
structed (Fig. 1d)).
No support had been obtained from the local government. The design and con-
struction were very basic, but the absence of experienced workers for mixing con-
crete, setting form work, and assembling iron bars caused slower construction pro-
gress than expected.

2.2 Vented ford constructed in collaboration with local governments,

communities and NGOs in the Philippines
In the Philippines, as shown in Fig.2, the provincial road is located in the vicinity of
the Magat river, with some sections often cut due to flooding during rainy seasons.
The community living on the east side of the Magat river had faced difficulty access-
ing high schools and the provincial capital during the rainy season. The communities
had listed several crossing sections including the crossing at the Magat river (which
was obviously beyond the scope of community initiative to build) to be improved
through a local resource-based approach. Through a field survey of the community
that considered both practicality and capability of extending access to basic human
588

needs, the village road-stream crossing near the high school (Fig.3 a)) was identified
as the target of the project.
Girder bridge(Maddiangat Bridge）
Bayombong:
Provincial capital
1km

Queson

Vented Ford

High school ：National road

：Provincial road
: Village road
Girder bridge The other : Stream
(Batu bridge) village : Concrete pavement

Fig. 2. Road network and the location at which the vented ford was constructed

a) b)

c) d)
Fig. 3. Construction of the vented ford: a) before construction; b) after construction; c) excava-
tion of the foundation by the equipment; d) side wall consisting of Do-nou with soil-cement

Based on the conditions summarized in Table 1, the vented ford was designed. In
collaboration with the local government, community, and the NGO, the vented ford
Sustainable Measures to Achieve Better River-Crossing … 589

was constructed as shown in Fig.3 b). The construction record is summarized in Table
2. The local government allocated their own budget to rent the excavator (Fig.3 c))
and concrete mixer, dispatching engineers for supervision and aggregates for concrete
works, which accounted for 45% of the total construction cost. The other cost was
provided by the NGO. Upon demonstration of the durability of the side wall consist-
ing of Do-nou filled with soil-cement reinforced with iron bars against high-flow
water pressure, a vented ford of the same structure as used in a project in 2015 [2]
was adopted (Fig.3d)).

3 Performance of the road-stream crossing and maintenance by

the communities

3.1 Low-water bridge condition in Zambia at an emergent flood in October,

2017
In October, 2017, a dam owned by the commercial firm upstream of the bridge broke
and an emergent flood occurred. The low-water bridge was overtopped, but the bridge
itself withstood the flood. Some of the safety poles were washed away or bent due to
the overtopping water flow (Fig. 4 a); however, the backfill of the abutments was
washed out and settled (Fig. 4 b)). These sections needed to be filled, and then the
slope of the embankment was reinforced with riprap. The community who participat-
ed in the construction project was eager to fix the section by extending the bridge to
both sides.

3.2 Performance of the vented ford during the rainy season and further
initiatives of the local government and community following the project in
the Philippines

In 2018 October, the site was visited to observe the conditions and interview the vil-
lagers concerning their use of the vented ford. According to the villagers, the vented
ford has been overtopped 5–6 times per year and withstood the high flow (Fig.5 a)).
With local government and community initiatives, the debris stuck inside the pipes
was removed, especially after the high flow.

a) b)
590

Fig. 4. Conditions hit by the emergent flood: a) withstanding bridge; b) eroded section behind
the abutments

a) b)
Fig. 5. Condition of the vented ford after the construction; a) vented ford during rainy season;
b) concrete pavement on the sloped access road to the vented ford

To make the village road into a detour of the national road as shown in Fig.2, the
sloped access road at both sides of the vented ford was paved with concrete (Fig.5 b)).
The safety poles were not placed because they might be damaged by overtopping
flow. So far, no road users, nor anyone in the community or local government have
complained about the risk of falling down from the edge of the driving surface. They
consider people using the vented ford to be responsible for their own safety.

4 Conclusion

The followings results were found by the case studies.

1. Building low-water crossings though community initiatives and local-resource mo-

bilization may be practical.
2. Experienced concrete workers from local communities could facilitate the work.
3. Measures to avoid erosion of the backfill behind the abutment are required.
4. Sidewalls consisting of Do-nou filled with soil-cement and reinforced with iron
bars could be alternatives to riprap.
5. Side-safety poles may be damaged by overtopped water flow.
6. The community involved takes ownership of the crossings to maintain them con-
tinuously and take self-responsibility for passing the bridge safely.

References
1. Fukubayashi, Y. and Kimura M.: Improvement of Rural Access Roads in Developing
Countries with Initiative for self-reliance of communities. Soils and Foundations 54(1),
23–35(2014).

2. Fukubayashi, Y. and Kimura, M.: Spot-Improvement of Rural Roads using a Local Re-
source-Based Approach: Case Studies from Asia and Africa, In Seth Appiah-Opoku (eds),
International Development, pp. 89–117, Intech, Croatia (2017).
 Assessment of Design Guidelines for Fiber-Reinforced
Polymer Shear Contribution of Prestressed Concrete
Beams Strengthened by Fiber-Reinforced Polymer Sheets

Dien Ngoc Vo-Le12, Chinh Ho-Huu1, and Long Nguyen-Minh1

1
Department of Structural Design, Faculty of Civil Engineering, HCMC University of
Technology
268 Ly Thuong Kiet, District 10, Ho Chi Minh City, Viet Nam
2 Faculty of Civil Engineering, Tien Giang University

119 Ap Bac, My Tho city, Tien Giang Province, Viet Nam

1585034@hcmut.edu.vn

Abstract. This research is mainly focussed on statistical assessment and analysis

of the accuracy of predicting the shear resistance of the fiber-reinforced polymer
(FRP) sheets for prestressed concrete beams strengthened by FRP sheets pre-
sented in the current design guidelines. The evaluation of the current prediction
models is based on a database of experimental results from the previous and cur-
rent author's research. The specifications of the beams are diverse and wide
enough such as beam types (prestressed concrete beams using bonded tendons -
BPC beams and unbonded tendons - UPC beams), cross-section shape, concrete
strength, effective pre-stress stress, and shear span to depth ratio - a/d …. The
results of the evaluation have shown that the formulas in recent design guidelines
overestimated the shear contribution of FRP sheets for prestressed concrete
beams.

Keywords: Shear, strengthened, FRP sheets, prestressed concrete beam, design

guidelines.

1. Introduction

In recent decades, the technique of shear strengthening using fiber reinforced polymer
(FRP) sheets have been quite popular and mainly focused on reinforced concrete (RC)
structures. However, the study of this technical solution on prestressed concrete (PC)
members is really scarce and limited because there are only a few studies available in
the literature [1-7]. Previous studies mainly focussed on studying the effect of several
major parameters, which influenced the shear contribution of FRP to shear resistance,
for instance, the stirrups ratio, the FRP shear reinforcement ratio, the strengthening
scheme, the beam geometry, the concrete strength, the effective prestress in the strands
and the ratio of the shear span to effective depth. The current design guidelines have
introduced specific provisions to determine the shear resistance for PC beams strength-
ened with FRP sheets [8-11].

© Springer Nature Singapore Pte Ltd. 2020 591

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_93
592 D. N. Vo-Le et al.

These calculation terms in design guidelines were built mainly based on experi-
mental studies of RC beams strengthened with FRP sheets. Besides, the superposition
theorem with separately shear resistance contributions of concrete, stirrups and FRP
sheets was used. However, the interaction between these components and the effect of
them were ignored. In fact, the shear behavior of RC beams and PC beams have many
differences. Therefore, the use of the terms in the design guidelines to analyze and pre-
dict the shear resistance of FRP sheets reinforcement on PC beams may lead to irra-
tional results.
This paper makes analysis reviews to evaluate the formulas of the prediction shear
resistance of FRP sheets in the current design guidelines [8-11]. Moreover, the study
compares the shear resistance of FRP sheets calculated in the current design guidelines
with the corresponding experimental results. The evaluation the accuracy of the predic-
tive standard shear resistance of FRP sheets for PC beams is really necessary to see
rationality and ensure the safety of the standards to apply in actual designs.

2. Contribution of FRP sheets to shear resistance

Describing the shear resistance mechanisms of PC beams strengthened FRP sheets is

not simple. Therefore, the common method to calculate the shear resistance of PC
beams strengthened FRP sheets is mainly based on the superposition theorem with sep-
arately shear resistance contributions of concrete, stirrups, prestressing force cable, and
FRP sheets. However, this superposition model does not explicitly describe the shear
resistance mechanisms of PC beams and the interaction between these mechanisms.

Table 1. Status of influencing factors to shear resistance of FRP sheets in the current design
guidelines.

Design guidelines
Influencing factors ACI 4402R CNR-DT 200 TR55 CIDAR
1. Bond model   x 
2. Effective FRP strain    
3. Configuration    
4. FRP sheet ratio    x
5. Concrete strength    
6. Strip width to spacing ratio x  x 
7. Cracking angel x   
8. Anchorage length    
9. Transverse steel ratio x x x x
Note: : included; x: not included.

Many models predicted the shear resistance of externally bonded FRP sheets. The re-
searchers calculated the shear resistance of FRP sheets similar to stirrups and mentioned
Assessment of Design Guidelines for Fiber-Reinforced … 593

the influences of the effective FRP stress. Some authors indicated the unsuitability of
the predictive formulas shear resistance of FRP sheets in the design guidelines [2,5,6,7].
Several influencing parameters were mentioned in design guidelines such as bond
model, FRP effective strain, concrete strength, anchorage length, cracking angel, strip
width to spacing ratio. While a few key factors haven’t yet been captured by design
guidelines and codes such as shear span to effective depth ratio, transverse steel ratio,
anchorage systems, prestressing force cable. Therefore, the scope of using standards is
limited. However, due to its simplicity and sufficient accuracy in predicting the shear
resistance of the beam, it is still used to calculate in the current design guidelines. Table
1. lists the major influencing factors to the shear resistance of FRP sheets in the current
design guidelines.

3. Comparison of experimental results and predictions by design

guidelines.
In order to validate the rationality and accuracy of the formulas predicting the shear
resistance of FRP sheets in the design guidelines, test results from other research studies
were used. A database representing 33 prestressed concrete beams strengthened in shear
with FRP U-jacket was collected from published literature [1,4,5,6,7]. These PC beams
used unbonded and bonded tendons. The investigated factors include: the cross-sec-
tional beam (rectangular, I and T), the concrete strength (f’c) from 30.6 to 71.2 MPa,
the effective concrete stress (fpc) from 3.32 to 9.56 MPa, the transverse steel ratio (ρsw)
from 0.0 to 0.56%, the shear span to effective depth (a/d e) from 1.5 to 3.1, the strength-
ening scheme (U-strips, U-wrap and full-wrap), the FRP shear reinforcement ratio (ρf)
from 0.06 to 2.17%, the trajectory cable (curved cable and straight cable). All beams
were tested under monotonic loading (four-point bending) up to failure.
The results of statistical analysis evaluate the accuracy of the current design guide-
lines with the mean value (Mean) and the coefficient of variation (COV) of the ratio
between the design shear resistance of FRP sheets and the corresponding experimental
values (Vf,theor/Vf,exp) in Fig. 1. In general, all design guidelines are overestimated as
compared to experimental results. So, they are not safe for purpose design. The pre-
dicted value of CIDAR and ACI 440.2R are unsafe with Mean = 2.05 and COV = 0.50
for CIDAR; Mean = 1.98 and COV = 0.49 for ACI 440. The predicted value of TR 55
and CNRDT 200 are closer to the experimental values, but they have large deviations
with Mean = 1.50 and COV = 0.59 for TR55 and Mean = 1.32 and COV = 0.44 for
CNRDT 200.
There were a number of causes for differences between the predictions and the ex-
perimental results. First, all design guidelines have been built based on data of RC
beams strengthened FRP sheets. Second, the effective work of FRP sheets is evaluated
unproperly with the interactive bonding surface between concrete and FRP sheets. Nev-
ertheless, all design guidelines also ignore the interaction of other important parameters
such as shear span to effective depth ratio, transverse steel, anchorage systems, pre-
stressing force cable, number layers.
To further investigate the accuracy shear resistance of FRP sheets between predictive
design values and the corresponding experimental values, the variation (Vf,theor/V,f,exp) is
594 D. N. Vo-Le et al.

examined against factors including: the compressive strength concrete f’c, the effective
concrete prestress fpc, the strip width to spacing ratio wf /sf, the shear span to effective
depth ratio a/de, the transverse steel ratio ρsw, and the FRP shear reinforcement ratio ρf
in Fig. 2. There is a large dispersion of the ratio Vf,theor/V,f,exp regarding the six factors,
so the design guidelines are unreliable for the prediction shear resistance of FRP rein-
forcement. Therefore, it is necessary to consider adjusting these formulas or establish-
ing a new formula for PC beams strengthened by FRP shear resistance

500 500
(a) (b)
400 400

Vf, theor (kN)

Vf,theor (kN)

300 300

200 200

TR55 CIDAR
100 100
Mean = 1.50 Mean = 2.05
COV = 0.59 COV = 0.50
0 0
0 100 200 300 400 500 0 100 200 300 400 500
Vf,exp (kN) Vf,exp (kN)

500 500
(c) (d)
400 400
Vf, theor (kN)

Vf, theor (kN)

300 300

200 200

CNRDT 200 ACI 440-2R

100 100
Mean = 1.32 Mean = 1.98
COV = 0.44 COV = 0.49
0 0
0 100 200 300 400 500 0 100 200 300 400 500
Vf,exp (kN) Vf,exp (kN)

Fig. 1. Comparison between design guidelines and experimental shear strength: (a)
TR55; (b) CIDAR; (c) CNRDT 200; (d) ACI 440-22R.
Assessment of Design Guidelines for Fiber-Reinforced … 595

8 8
TR55 CNRDT-200 ACI 440 CIDAR TR55 CNRDT-200 ACI 440 CIDAR

6 6
Vf,theor/ Vf,exp

Vf,theor/ Vf,exp
4 4

2 2

0 0
20 30 40 50 60 70 80 0 2 4 6 8 10
(a) f'c (MPa) (b) fpc (MPa)
8 8
TR55 CNRDT-200 ACI 440 CIDAR TR55 CNRDT-200 ACI 440 CIDAR

6 6
Vf,theor/ Vf,exp
Vf,theor/ Vf,exp

4 4

2 2

0 0
0.0 1.0 2.0 3.0 4.0 0.0 0.5 1.0 1.5 2.0
(c) a/de (d) w f/sf
8 8
TR55 CNRDT-200 ACI 440 CIDAR TR55 CNRDT-200 ACI 440 CIDAR

6 6
Vf,theor/ Vf,exp

Vf,theor/ Vf,exp

4 4

2 2

0 0
0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.5 1.0 1.5 2.0 2.5 3.0
(e) psw (%) (f) pf (%)

Fig. 2. Evaluation of predictive design guidelines for various parameters: (a) f’c; (b) fpc;
(c) a/de; (d) wf /sf; (e) ρsw; (f) ρf

4. Conclusions
This study assesses the accuracy of predicting the shear resistance of the FRP sheets
for strengthened PC beams in the design guidelines based on 33 experimental datasets.
596 D. N. Vo-Le et al.

All current design guidelines are overestimated the shear resistances of FRP sheets as
compared to experimental results. So they are unsafe for purpose design. Therefore, it
is necessary to consider adjusting these formulas or establishing a new formula for PC
beams strengthened by FRP shear resistance with mention the interaction of key factors
including shear span to effective depth ratio, transverse steel ratio, anchorage systems,
prestressing force cable.
Because of the lack of the experimental results, there is a need for further studies to
properly evaluate the efficiency of using FRP sheets strengthening PC beams.

References
1. Kang, T.H.K. and Ary, M.I.: Shear-strengthening of reinforced & prestressed concrete
beams using FRP: Part II—Experimental investigation. International Journal of Concrete
Structures and Materials, 6(1), 49-57 (2012).
2. Kang, T.H.K. and Ary, M.I: Shear-strengthening of reinforced & prestressed concrete beams
using FRP: Part I-Review of previous research. International Journal of Concrete Structures
and Materials, 6(1), 41-47 (2012).
3. Kim, Y., Quinn, K., Satrom N.: Shear strengthening of reinforced and prestressed concrete
beams using carbon fiber reinforced polymer (CFRP) sheets and anchors. Technical Rep.
No. FHWA/TX-12/0-6306-1, Center for Transportation Research, The University of Texas
at Austin, Austin, Texas (2012).
4. Murphy, M., Belarbi, A., Bae, S.W.: Behavior of prestressed concrete I-girders strengthened
in shear with externally bonded fiber-reinforced-polymer sheets. PCI Journal, 57(3), 63-82
(2012).
5. Nguyen, T.T.D., Matsumoto, K., Yamada, M., Niwa, M.: Behaviors of pretensioned PC
beams strengthened in shear using externally bonded CFRP sheets. JCI Concrete Journal,
37(2), 1183-1188 (2015).
6. Nguyen, T.T.D., Matsumoto, K., Sato, Y.: Shear-resisting mechanisms of pre-tensioned PC
beams without shear reinforcement strengthened by CFRP sheets. Journal of JSCE, 4(1), 59-
71(2016).
7. Nguyen-Minh, L., Vo-Le, D., Tran-Thanh, D.: Shear capacity of unbonded post-tensioned
concrete T-beams strengthened with CFRP and GFRP U-Wraps. Composite Structures, 184,
1011-1029 (2018).
8. ACI Committee 440.: Guide for the Design and Construction of Externally Bonded FRP
Systems (ACI 440.2R-17). Farmington Hills, MI: American Concrete Institute (2017).
9. CIDAR.: Design guideline for RC structures retrofitted with FRP and metal plates: Beams
and slabs. Draft 3 – Submitted to Standards Australia, The University of Adelaide, Autralia
(2006).
10. CNRDT200.: Guide for the Design and Construction of Externally Bonded FRP Systems
for Strengthening Existing Structures. Advisory Committee on Technical recommendations
for construction, Rome, Italy (2013).
11. TR55.: Design Guidance for Strengthening Concrete Structures using Fibre Composite Ma-
terials. Second Edition. Report of a Concrete Society Committee (2012).
 Development of Disaster Resilient and Sustainable
University Framework : Case of Bandung Institute of
Technology (ITB)

Krishna Suryanto Pribadi 1[1]

, Rahma Hanifa†1[2], Eliya Hanafi‡1[3], and Giovanni Pra-
dipta§1[4]
1
Research Center for Disaster Mitigation, Bandung Institute of Technology (RCDM ITB) –
RCDM ITB, PAU Building 8th Floor, Bandung Institute of Technology, Jalan Ganesa 10,
40132, West Java, Indonesia
2
University Forum for Disaster Risk Reduction (UF-DRR) Indonesia
ppmb@lppm.itb.ac.id

Abstract. Indonesia is an archipelagic and seismic active country, home to 241

million people. It also has 3,276 higher education institutions (in 2017), with a
total of more than 6.9 million students and more than 247 thousands of academ-
ic staffs. Following the Sendai Framework for Action (SFDRR 2015-2045),
Sustainable Development Goals (2015-2030), the Association of Pacific Rim
Universities (APRU) has declared that universities need to ensure the safety of
students, faculties, staffs and assets during natural disaster, which led to the
concept of Disaster Resilient University (DRU). Bandung Institute of Technol-
ogy (ITB) is the oldest engineering higher education institution in Indonesia. Its
main campus is located in Bandung city and home to about 21 thousand stu-
dents and 2600 permanent faculties and staffs. Bandung is vulnerable to flood,
rain storm, fire, earthquake, landslide and volcano. Since 2017 the Research
Center for Disaster Mitigation (PPMB), working together with ITB Unit for Se-
curity, Safety, Health and Environment (K3L) is promoting the concept of Dis-
aster Resilient University. The ultimate goal is that ITB to achieve a Disaster
Resilient University status by 2025, through the achievement of the following
objectives: strong partnerships within the campus community as well as with
government, built-in awareness and safety culture in ITB, and improved resili-
ency by building capacity for reducing disaster risk, managing emergency, in-
cluding continuity of operation. The activities consist of 4 phases with the fol-
lowing outputs : ITB DRU policy and action plan; organized campus resources,
disaster risk assessment of ITB Campuses; annual emergency response exercis-
es and drill; trained campus staffs and students in the emergency response and
continuity of operation. The DRU best practices will be shared with other uni-
versities and higher education institutions in Indonesia.

Keywords: Resilient University, Sustainable City, Risk Assessment.

© Springer Nature Singapore Pte Ltd. 2020 597

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_94
598 K. S. Pribadi et al.

1 Background

Indonesia is an archipelago state, home to 241 million people. It is a seismic prone

country, experiencing an average of 2000 earthquakes of M4.5 or more per year. Ex-
ample of recent catastrophic earthquakes are the M9.2 Sumatera-Andaman Meg-
athrust Tsunamigenic Earthquake in 2004, M6.3 Yogyakarta 2006 earthquake, M6.5
Pidie Jaya Aceh 2016 earthquake.
Indonesia registered a total of 3,276 higher education institutions, comprising of
121 state universities and 3,104 private universities, with a total of more than 6.9
million students and 247 thousands academic staffs (MoRTHE, 2017). Following the
Sendai Framework for Action (SFDRR 2015-2045), Sustainable Development Goals
(2015-2030), the Association of Pacific Rim Universities (APRU, 2018) have de-
clared that Universities need to ensure the safety of students, faculties, staffs and as-
sets during natural disaster. The concept of Disaster Resilient University is estab-
lished under this principle. The Ministry of Research, Technology and Higher Educa-
tion (MoRTHE) together with the National Disaster Management Agency (BNPB)
and the National University Forum for Disaster Risk Reduction (NUF-DRR) has
launched the Indonesian Disaster Resilient Campus Initiative on January 25, 2019.
The Bandung Institute of Technology (ITB) main campus, located in Bandung
City, West Java Province, Indonesia was established in 1920 as the oldest technology-
oriented university in Indonesia. It has annually 15.000 undergraduate students in
average, and 6.000 graduate students, with 2600 permanent faculties and staffs. Most
of earthquake experts in the country come from this university.
Bandung is vulnerable to flood, rain storm, fire, earthquake, landslide and volcano.
Several old big trees fell down in ITB campus during storms, causing several damag-
es to vehicles and campus facilities. Fires damaging university facilities have been
experienced. While fire drill and fire prevention have been well practiced at the uni-
versity, earthquake was not much considered until 2016 when researchers found out
that the Lembang Fault, located 7 km to the north of the city, was capable of generat-
ing M6.6 earthquake with a VI-VIII MMI intensity (PuSGeN, 2017, BMKG 2017).
In line with Sendai Framework for Action (SFDRR 2015-2045) and Sustainable
Development Goals (2015-2030), since 2017 the ITB Research Center for Disaster
Mitigation (PPMB-ITB) collaborates with ITB Unit for Security, Health, Safety and
Environment (K3L) to promote Disaster Resilient University concept at ITB. As a
Disaster Resilient University (DRU), a university needs to maximize the process of
communication, collaboration, and coordination in the face of disaster emergency and
its recovery process. Bandung Institute of Technology, as a prominent university in
Indonesia, promotes the Disaster Resilient University concept, that will contribute
significantly to the effort of Bandung City to become a Resilient City.

2 Disaster Resilient University Concept

ITB together with the University Forum for Disaster Risk Reduction (UF-DRR)
developed the concept of Disaster Resilient University. UF-DRR defines DRU as a
Development of Disaster Resilient and Sustainable University … 599

university that has the ability to manage and minimize disaster risks, able to recover
and return to academic activity immediately when affected by disasters, and contrib-
ute the benefit of science and technology for disaster risk reduction and resilience to
the society, through the implementation of Tridharma of Universities (three pillars of
university mission) : education, research and community outreach.

The APRU workshop on Disaster Resistant University in Sendai (2018) has

proposed five minimum aspects of DRU and their elements:

Governance or policy : University have disaster emergency plans (communication,

electricity backup); Staff and teaching staff know the plans and understand how to
act in emergency; Disaster emergency plans are reviewed and updated annually;
University prepares disaster management tools once recovered from a disaster;
University has an emergency structure; University has communication plans;
Disaster emergency training is conducted at least once a year; A plan for the safety
of students, staff and faculty has been prepared; University has developed an
evacuation plan to accommodate students and staff; University has backup
generator on location for emergency purpose.
Understanding of Risk : Risk assessment is carried out and updated regularly;
University develops risk preparedness and mitigation / preparedness plans; Based
on the plan, mitigation measures are implemented; Emergency supplies (food,
water, blankets, etc.) are always available; All actions are checked regularly to
determine their effectiveness.
Physical and digital infrastructure : Buildings have earthquake resistant structures;
Buildings are facilitated by drainage, electricity, fire alarm systems, sprinklers, fire
extinguishers, etc.; Maintenance work is carried out regularly in buildings;
Building inspections are conducted regularly; Important information if an
emergency occurs such as an evacuation route and emergency assembly points are
shared with students, faculty, and staff. This facility should have signage; IT
recovery plans are developed; Technology tools such as satellite / cellular
emergency phones, alarms, and drones put in place.
Preparedness Education and Training : An orientation is carried out among new
students about emergencies and disaster preparation; Special guides for foreign
students are available; Safety protocols for students abroad are shared before their
departure; Disaster emergency training is carried out; Programs are carried out to
increase awareness; Training such as CPR for students, teaching staff and leaders
on campus is done; Material information about past disasters, contingency plans,
etc. distributed.
Physical and Psychological Assistance : There is a hotline in place for students,
staff, and faculty when they need physical or psychosocial support during and after
an emergency; There is a prior agreement with local governments, organizations
and other Universities on collaboration for reciprocal support in case of
emergencies including voluntary registration;
600 K. S. Pribadi et al.

3 Hazard Assessment at ITB Campus

Bandung Institute of Technology (ITB) three campuses are exposed to various nat-
ural and non-natural hazards, such as earthquake, flood, fire, volcanic eruption, land-
slide and extreme weather event (strong wind) as well as technological hazard

i. Earthquake : Several active faults may affect the three campuses of ITB (Ganesa
main campus, Jatinangor campus and ITB campus in Cirebon City), such as the
Lembang Fault, capable of generating M6.6 earthquake, the Java megathrust
earthquake which is capable of generating M8.7-9 earthquake (PuSGeN, 2017,
Hanifa et al, 2014), potentially causing VI-VIII MMI shaking (PuSGeN, 2017,
BMKG 2018), and the Baribis-Kendeng fault south of Cirebon ITB Campus.
Based on the probability of a 10% shock in 50 years (PuSGeN, 2017), the poten-
tial for earthquake shocks at the three ITB campuses is equivalent to the intensity
of VI-VIII which can have widespread damage effects.
ii. Flood : The flood threat also has a high potential, especially for the Cirebon cam-
pus ITB area.
iii. Fire Hazard : Several fire incidents have occurred in the ITB main campus which
caused quite significant losses. Besides fires caused by fire hazard from inside
the campus, there are potential hazard from land and forest (bush) fires
(Jatinangor Campus).
iv. Volcanic hazard : There are at least two active volcanoes near ITB campuses,
namely Mt.Tangkuban Perahu and Mt. Ciremai. The ITB campuses are actually
outside of the volcanic hazard areas, however the Cirebon Campus may need to
be prepared for possible volcanic activity of Mt Ciremai.
v. Landslide : The ITB Ganesha and Cirebon campuses are outside the Landslide
Hazard Zone, but the ITB campus in Jatinangor and Walini is in an area prone to
landslides.
vi. Extreme weather events : Several rain events accompanied by strong winds had
caused several trees to fall and hit vehicles parked in the ITB Ganesha campus.
vii. Technological Hazard : There are activities, hazardous material and equipments
located in some laboratories at ITB campus, such as chemistry laboratories, biol-
ogy laboratory, mechanical and electrical engineering laboratories and so on,
which have the potential to cause hazardous event when there is a technological
failure and human error. Beside that, security threat may come also from terror-
ism act and violence. In particular, cyber attack may cause great damage to digi-
tal assets of ITB.

Fig 1 show maps for earthquake, flood, forest fire, volcanic eruption and landslide
hazard that may affect ITB campuses.
Development of Disaster Resilient and Sustainable University … 601

Fig. 1. Maps of Earthquake, Flood, Forest Fire, Volcanic, and Landslide Hazards affecting ITB
Campuses

4 Building Disaster Resilient ITB

4.1 Objective

The program aims ITB to achieve a Disaster Resilient University status by 2025,
by building partnership with the whole campus community as well as external stake-
holders (government and non-government organizations) and promoting safety cul-
ture among the ITB community as well as strengthening campus preparedness for
emergency and post impact recovery. It has four missions, (1) building university
capacity in increasing its resilience, (2) understanding university disaster risk in a
dynamic and multidisciplinary manner, (3) prepare action plans for DRR and prepar-
edness (4) advocacy for awareness.

4.2 Activities

Following FEMA (2003) Guideline, activities are carried out through 4 phases:

Phase 1 - Organize Resources : ITB DRU Implementation Team is established, led

by ITB Unit for Security, Health, Safety and Environment (K3L) in collaboration
with Research Center for Disaster Mitigation (PPMB) of ITB, representatives from
different units and faculties from ITB, student union and relevant student units such
as Red Cross volunteers, Student Military Unit, etc.

Phase 2 - Hazard Identification and Risk Assessment: Identifying and analysis of

hazards, assessment of building and campus facility vulnerabilities, developing
disaster scenarios and impacts.

Phase 3 - Developing the Mitigation and Response Plan: Short and long term plans
for reducing vulnerabilities, preparedness planning and emergency response proce-
602 K. S. Pribadi et al.

dures including evacuation planning, university continuity of operation plan, devel-

oped through a series of workshops involving internal and external stakeholders.

Phase 4 - Adoption and Implementation: Adoption of the plan by university

management, implementing vulnerability reduction programs (retrofitting of building
and critical facilities and utilities, exercise and drills, review and improvement of
plans). Various preparedness training are organized, including evacuation and rescue
drills, fire fighting drills etc. collaborating with local and national relevant agencies.
On Disaster Preparedness Day of Indonesia, which falls on 26 April, annual campus
wide disaster emergency exercise is conducted. The 2018 and 2019 annual exercises
were conducted based on a 30 second earthquake scenario which trigger a fire inci-
dent, involved evacuation, rescue, medical, logistic and fire fighting actions as well as
emergency command center exercise .

5 Conclusions

A Disaster Resilient University (DRU) plan for Bandung Institute of Technology

has been developed and implemented. Disaster Resilient University is defined as a
university that has the ability to manage and minimize disaster risks, able to recover
and return to academic activity immediately when affected by disasters, and contrib-
utes the benefit of science and technology for disaster risk reduction and resilience to
the society. This status will be achieved through the implementation of the University
Tridharma (three pillars mission of universities) i.e. education, research and commu-
nity outreach. The initiative is also contributing to the Bandung City and West Java
Province efforts to become a resilient and sustainable city and province.

References
1. APRU-Multihazard Program. Report of the APRU-IRIDES Multi-Hazards Program Cam-
pus Safety Workshop 2018. http://aprumh.irides.tohoku.ac.jp/app-def/S-
102/apru/archives/1159
2. BMKG, 2017. Seismic Intensity Scenario for Lembang Earthquake of M6.8
3. BMKG, 2018. Seismic Intensity Scenario for Java Megathrust Earthquake of M.8.7
4. Director of Campus Resilience – University of San Francisco, 2018. Emergency Opera-
tions Plan. https://www.usfca.edu/public-safety/disaster-preparedness/plans/emergency-
operations-plan
5. FEMA, 2003. Building a Disaster Resistant University
6. Hanifa, N. R., Sagiya, T., Kimata, F., Efendi, J., Abidin, H. Z., & Meilano, I. (2014). In-
terplate coupling model off the southwestern coast of Java, Indonesia, based on continuous
GPS data in 2008–2010. Earth and Planetary Science Letters, 401, 159-171.
7. Map of Indonesia 2017”, Indonesia National center for earthquake Studies, Research and
Development Agency of Ministry of Public Work and Housing. ISBN 978-602-5489-01-3.
8. National Center of Earthquake Studies – PuSGeN, 2017. The Updating of National Source
and Hazard Map of Indonesia 2017
 Experimental study on the valorization of poplar by-
products in cement-based materials

C. Djelal1, J. Page1, Y. Vanhove1 and H. Kada1

1
Univ. Artois, EA 4515, Laboratoire de Génie Civil et géo-Environnement (LGCgE), 62408,
Béthune, France
jonathan.page@univ-artois.fr

Abstract. An experimental investigation was undertaken to evaluate the poten-

tial use of poplar by-products in cement mortars. Two by-products from poplar
processing were studied: sawdust (from sawing wood) and milled fiber (from
wood waste obtained during cutting). The fibers and sawdust were incorporated
into the mortars as a sand substitution with rates from 10 to 100%. The intro-
duction of these poplar by-products has a significant on the composite proper-
ties in both fresh and hardened state. It was noted that the workability of the
mixtures with the poplar wood varies with the substitution rate. Moreover, the
increase in the amount of fibers or sawdust causes a continuous increase in the
porosity of the material, which leads to a decrease in density. A significant de-
crease in the mechanical properties of bio-sourced mortars was noted, again re-
lated to the rate of fibers or sawdust introduced in mixtures. The high internal
porosity of the wood particles and their low density can explain these lower
mechanical strengths.

Keywords: Poplar sawdust; Plant fibers; Mix design; Workability; Mechanical

strength; Porosity.

1 Introduction

Economic issues related to the rising costs of non-renewable resources, their increas-
ing scarcity, and environmental impacts inherent to their production therefore lead the
construction industry to move towards the use of bio-based materials. Among agro-
based materials, wood stands out because of its low density and its environmental
assets. Indeed, a cubic meter of wood used in a construction represents about 1 ton
less of CO2 emitted into the atmosphere [1]. In France, the wood industry generates
over 12 million tons of waste annually, 60% of which stems primarily from sawmills
[2]. Thus, in an environmental approach, the Nord-Pas-de-Calais region wants to find
a new way of valorization of poplar by-products stem from sawmills. This choice is
justified by the massive production of this species in France [3].
Using these poplar-products in mortars as substitution of natural aggregates could
be an interesting way of valorization. Numerous studies dealt with the use of wood
waste as cementitious construction materials. It was highlighted that in fresh state,
plant particles affect the rheology of the material [4,5]. Moreover, because of their

© Springer Nature Singapore Pte Ltd. 2020 603

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_95
604 C. Djelal et al.

chemical composition, the inclusion of plant-based particles in a cementitious matrix

influences the cement hydration of the material, by inhibiting the reaction [6,7]. This
inhibition is generally attributed to the extractable wood substances that solubilize in
the cementitious matrix. These disturbances of cementitious hydration lead to the
reduction of the mechanical strengths of the wood-cement composite [8,9]. Many
authors highlight a decreased density in the composite which acts to depress the com-
pressive strength [6, 10,11]. Several studies have pointed out that the material proper-
ties depending on the wood specie [12].
This work aims to characterize the influence of by-products on the properties of
cementitious materials in both fresh and hardened state. These wood by-products are
used in two different forms: 2-mm fiber (crushed from wood waste) and sawdust.
This work is a preliminary mix-design study carried out before the industrial manu-
facture of concrete blocks made by vibrocompaction [13].

2 Characterization of poplar by-products

The wood by-products used in this study come from the processing of Robusta poplar.
They are used in two different shapes: sawdust and 2-mm fiber. The fiber is obtained
from a cutting mill combined with a 2-mm mesh sieve and the sawdust comes from
the wood sawing process. The observation under a scanning electron microscope
(SEM) of poplar fiber and sawdust (Fig. 1) shows an irregular shape and rough with
fibrils on their surface oriented with a preferential direction. This specific morphology
could influence the adhesion between wood particles and cementitious matrix.

Fig. 1. Observation of wood by-products with the naked eye and with a SEM.

The particle size distribution of wood by-products was determined by mechanical

sieving, according to standard NF P18-545, with the series of sieves relative to sand
(Fig. 2). The particle size distribution of sawdust is between 0 and 6.3 mm, with a
fineness modulus of 3.20, while that of fibers ranges from 0 to 5 mm, with a fineness
modulus of 3.72. The fineness modulus is an empirical figure obtained by adding the
total percentage of the sample of an aggregate retained on each of a normalized series
of sieves, and dividing the sum by 100. This modulus is even smaller than the aggre-
gate has fine elements. In particular, sawdust is characterized by a larger fraction of
fines. Indeed, the amount of sawdust particle passing the 2-mm sieve is greater than
90% of the total mass, while it is only about 50% for the fibers.
Experimental study on the valorization of poplar … 605

Fig. 2. Particle size distribution of poplar fiber and sawdust.

The water absorption coefficient of poplar by-products was characterized accord-

ing to the method for lightweight aggregates in EN 1097-6 standard. These poplar
fibers and sawdust therefore have absorption coefficients (WA24) of 199±15% and
174±6% respectively. Thus, both forms of wood particles have a very high-water
absorption capacity because of their porosity. These values are in agreement with
those found in the literature, with water absorption values for plant particles varying
between 34% and 407% [14]. These water absorption coefficients are relatively close
because fibers and sawdust come from the same species of poplar. This high absorp-
tion could be disruptive during the mixing process of mortars, by reducing the effec-
tive water amount available for cement. Thus, this characteristic will be taken into
account in the mortar mix-design. Absorption kinetics showed a saturation time
around 2 minutes.

3 Formulation of wood/cement composites

The study is based on different substitutions of sand by poplar sawdust and fibers into
the reference mortar mix, because of the grain size class similarity between these
materials. The compositions are given in Table 1. All mortars are prepared with a
constant Sand/Cement ratio fixed to 3 and an E/C of 0.5.
The binder used is a CEM I 52.5 N Portland cement, with a density of 3.11 g.cm -3.
A siliceous marine sand with grain size 0/4 mm was used. This sand presents a densi-
ty of 2.61 g.cm-3 and an absorption coefficient of 0.49%. The mixing protocol is given
in Fig. 3. Before introducing cement and sand in the mixer, poplar by-products were
saturated with water to avoid reducing the effective water content.

Fig. 3. Mixing process of mortar.

606 C. Djelal et al.

Table 1. Composition of wood mortars (kg.m-3).

Ref. Mortars with fibers (MF) Mortars with sawdust (MS)

R0 F10 F20 F50 F100 S10 S20 S30 S50 S80 S100
Cement 508 508 508 508 508 508 508 508 508 508 508
Eff. water 254 254 254 254 254 254 254 254 254 254 254
Sand 1525 1373 1220 763 0 1373 1220 1068 763 305 0
Fibers - 20 39 98 196 - - - - - -
Sawdust - - - - - 23 46 68 114 182 228
Wood vol.% 0 10 20 50 100 10 20 30 50 80 100

4 Characterization of wood/cement composites

4.1 Workability of fresh mortars

In order to estimate the rheological characteristics and the shaping ability of these
composites, it was chosen to measure the slump of fresh cement/wood composites
(with poplar fibers and sawdust) by using a mini cone of dimension Ø10 cm x Ø5 cm
x 15 cm. The slump of mortars varies from 0 to 50 mm depending on the incorpora-
tion rate of poplar by-products into mixtures (Fig. 4). There is a firm consistency for
the reference mortar, without fibers or sawdust. An increase in the workability of the
mortars is noted with the increase of the proportion of wood aggregates in the mix-
ture. Nevertheless, this improvement in fluidity presents a threshold, from which
workability then decreases. This limit could correspond to the rate at which there is
entanglement of the wood aggregates, thus increasing the viscosity of the mixtures,
and depending on the type of aggregates. For poplar sawdust, this threshold corre-
sponds to an incorporation rate of 30%, while it corresponds to 50% for poplar fibers.

Fig. 4. Evolution of slump of mortars with sawdust and fibers.

Experimental study on the valorization of poplar … 607

4.2 Apparent density and porosity

The water accessible porosity and the apparent density of hardened mortar specimens
were calculated through vacuum saturation and with a hydrostatic balance according
to the NF P18-459 standard. Fig. 5 gives the evolution of mortar’s density and porosi-
ty according to the sand substitution ratio by wood aggregates. A decrease in density
of about 10% to 70% is observed for mortars with poplar fibers and sawdust. This is
attributed to the low density of wood particles. In parallel, the porosity of these mor-
tars increases sharply with the rate of incorporation of wood aggregates. However,
there is no significant difference between both types of poplar by-products.

Fig. 5. Evolution of slump of mortars with sawdust and fibers.

4.3 Compressive strengths

Compression tests were implemented on wood/cement mortars in accordance with the
EN 196-1 standard. The compressive strengths of fiber and sawdust mortars for dif-
ferent curing time are summarized in Table 2. The reference mortar strength (M0)
increases over time from 32 MPa (3 days) to 54 MPa (28 days). The incorporation of
wood aggregates causes a significant decrease in the compressive strength of the mor-
tar. For substitution rates below 50%, the maximum strength decrease is about 43%
for MS mortars versus 66% for MF mortars. Beyond 50% of wood, the strength drop
is higher, around 80% for both types of poplar aggregates. According to literature,
this decrease in resistance is related to the inhibition of the cement setting by plant
particles, the low wood density and the increase in of the composite porosity [11].

Table 2. Compressive strengths of mortars depending on the curing time (MPa).

Mortar M0 F10 F20 F50 F100 S10 S20 S30 S50 S80 S100
3 days 32.0 21.1 10.9 3.0 2.1 26.4 18.1 16.4 6.3 3.2 2.8
7 days 42.5 25.1 14.3 3.8 2.4 34.4 23.5 19.6 7.1 5.0 4.6
14 days 45.5 32.6 18.1 6.0 3.1 37.6 27.8 21.1 8.6 6.1 5.0
28 days 53.8 38.1 22.3 6.6 3.4 42.6 31.8 27.1 8.9 6.7 5.8
608 C. Djelal et al.

5 Conclusions

The experimental work presented in this paper allowed to characterize the influence
of poplar wood by-products (sawdust and fibers) on the workability and the physical
properties of a cementitious material. It appeared that these wood aggregates absorb
very rapidly a large amount of water (up to 200%). A modification of the workability
of mortars is noted with the wood content. In hardened state, a decrease in density is
observed for mortars with poplar fibers and sawdust, accompanied by a significant
decrease in the compressive strength.
However, important issues remain to be investigated, as the effect of poplar wood
on cement hydration, or on the thermal properties and the flexural behavior of compo-
sites. These points are the subject of work in progress.

References
1. Trouy-Triboulot, M.C., Triboulot, P.: Matériau bois - Structure et caractéristiques. Tech-
niques de l’ingénieur, Saint-Denis (2012).
2. FCBA Homepage, https://www.fcba.fr/, last accessed 2015/02/12.
3. CRPF: Schéma Régional de Gestion Sylvicole. Centre Régional de la Propriété Forestière -
Nord Pas-de-Calais Picardie, Amiens (2006).
4. Dawood, E.T., Ramli, M.: Properties of high-strength flowable mortar reinforced with
palm fibers. ISRN Civil Engineering 2012, 1-5 (2012).
5. Ndong Engone, J.G., Djelal, C., Xing, X., Vanhove, Y., Kada, H.: Study of the influence
of wood aggregates on sand mortar extrudability. In : Proceedings of the International
Conference on Advances in Civil, Structural and Mechanical Engineering, CSM-2014,
pp.150-154, IRED, Santa Barbara (2014).
6. Sauvat, N., Sell, S., Mougel, E., Zoulalian, A.: A study of ordinary Portland cement hydra-
tion with wood by isothermal calorimetry. Holzforschung 53 (1), 104-108 (1999).
7. Wei, Y.M., Zhou, Y.G., Tomita, B.: Hydration behavior of wood cement-based composite
I: evaluation of wood species effects on compatibility and strength with ordinary Portland
cement. Journal of Wood Science 46 (4), 296-302 (2000).
8. Wei, Y.M., Tomita, B., Hiramatsu, Y., Miyatake, A., Fujii, T., Yoshinaga, S.: Hydration
behavior and compressive strength of cement mixed with exploded wood fiber strand ob-
tained by the water-vapor process. Journal of Wood Science 49 (4), 317-326 (2003).
9. Frybort, S., Mauritz, R., Teischlinger, A., Muller, U.: Cement bonded composites – A me-
chanical review. BioResources 3 (2), 602-626 (2008).
10. Ashori, A., Tabarsa, T., Amosi, F.: Evaluation of using waste timber railway sleepers in
wood-cement composite materials. Construction and Building Materials 27, 126-129 (2012).
11. Al Rim, K., Ledhem, A., Douzane, O., Dheilly, R.M., Quéneudec, M.: Influence of the
proportion of wood on the thermal and mechanical performances of clay-cement-wood
composites. Cement and Concrete Composites 21 (4), 269-276 (1999).
12. Fan, M.Z., Bonfield, P., Dinwoodie, J.: Nature and behaviour of cement bonded parti-
cleboard: structure, physical property and movement. Journal of Materials Science 41 (17),
5666-5678 (2006).
13. Xing, Z., Djelal, C., Vanhove, Y., Kada, H.: Wood waste in concrete blocks made by
vibrocompression. Environmental Processes 2 (1), 223-232 (2015).
14. Pacheco-Torgal, F., Jalali, S.: Cementitious building materials reinforced with vegetable
fibres: A review. Construction and Building Materials 25 (2), 575-581 (2011).
 Horizontal response of base-isolated buildings supported
to high damping rubber bearings

Nguyen Anh Dung1, Le Trung Phong1 and Tran Minh2

1
Thuyloi university, 175 Tay Son–Dong Da–Ha Noi, Viet Nam
2
State Audit Office of Vietnam, No. 116 Nguyen Chanh–Cau Giay–Hanoi, Vietnam
dung.kcct@tlu.edu.vn

Abstract. Seismic isolation is one of the most efficient techniques to protect

structures against earthquakes. Rubber bearings are suitable for low-rise and
medium-rise buildings due to its durability and easy fabrication. This paper pre-
sents the horizontal response of a six-storey base-isolated building using high
damping rubber bearings (HDRBs) under two ground motions of earthquakes as
types I and II in JRA (2002) by finite element analysis. In this analysis, these
bearings are modelled by the bilinear hysteretic model which is indicated in
JRA and AASHTO. Comparison of horizontal response including base shear
force and roof level acceleration between the two cases: base-isolated building
and fixed-base building is carried out to evaluate the effectiveness of the use of
HDRBs on the protection of buildings from earthquakes. The numerical results
show that the peak value of roof floor acceleration of the fixed-base building is
two times higher than that of the base-isolated building, and the floor accelera-
tions depend on the peak values of ground acceleration. In addition, the step-by-
step design procedure for determining the size of HDRBs used for buildings is
also presented in this paper.

Keywords: High damping rubber bearings; seismic responses; earthquakes;

buildings.

1 Introduction

A technique called seismic isolation has emerged as a practical and economical alter-
native to the conventional approach. The main concept of seismic isolation is to de-
couple a structure from the horizontal components of the ground motion by inserting
structural elements (e.g. isolation bearings) between the substructure and the super-
structure. Isolation bearings, which have low horizontal stiffness, can shift the natural
periods of structures to reduce the acceleration response, while its damping capability
can diminish the response of displacement. Various types of isolators have been man-
ufactured with a similar objective to provide a natural period shift and additional en-
ergy dissipation to structures. Recently, high damping rubber bearings (HDRBs) have
been widely used in Japan due to their high flexibility and high damping capability. In
the manufacturing process of high damping rubber material, a large number of fillers
(about 30%) including carbon black, silica, oils and some other particles are added

© Springer Nature Singapore Pte Ltd. 2020 609

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_96
610 N. A. Dung et al.

during the vulcanization process [1, 2] in order to improve the desirable material
properties, such as the strength and damping capabilities. They present high damping
capability, and bearings defined as HDRBs should provide an equivalent viscous
damping of at least 10% [3]. Gu and Itoh [4] estimated that the change of the equiva-
lent horizontal stiffness of HDRBs is only about 10-25% after 100 years. The purpose
of this paper is to investigate the effectiveness of base-isolated building using HDRBs
compared to the fixed-base building under ground motions of earthquakes. Further,
the effect of peak value and the time duration of ground accelerations on the horizon-
tal response of the building is also investigated in this paper.

2 Basic principles of base isolation

Fig. 1. Acceleration response spectrum in Fig. 2. Displacement response spectrum

function of the damping (as defined by EC8 in function of the damping (as defined by
[5]) EC8 [5])

The relationships between seismic response of structures and their periods are pre-
sented in Fig. 1&2 [5]. Based on these relationships, the principle of seismic isolation
is developed. The basic principle of base isolation is that the response of a structure is
modified so that the ground below is capable of moving without transmitting minimal
or no motion to the structure above. For this purpose, base isolation is inserted be-
tween the base of a structure and its foundation. The isolation is characterized by a
high flexibility in large earthquakes. This flexibility must be such as to shift the fun-
damental response frequencies of the structure to values that are below the range
where the seismic motion of soil is amplified. In this way, the soil motion is filtered,
because most energy is withdrawn from the frequency range that characterizes the
vibration modes of the superstructure. In principle, a structure may be isolated in both
the horizontal and the vertical directions. However, the application of seismic isola-
tion is usually limited to horizontal directions. The reasons are that seismic loads are
largely less critical in the vertical direction, compared to the horizontal; also, possible
rocking due to three-directional isolation must be avoided [3]. The use HDRBs has
proven, even in recent earthquakes, to be a very efficient technique to protect struc-
tures from earthquakes. HDRBs with their high flexibility and high damping capabil-
ity can achieve two effects: (1) To shift the own period of the structures reducing in
Horizontal response of base-isolated building … 611

such a way the seismic response and (2); To dissipate energy reducing again the re-
sponse and the displacement in Fig. 1&2.

3 Numerical analysis of a base-isolated building supported to

HDRBs

3.1 Prototype building and high damping rubber bearings

The case study building is a six – story reinforced concrete frame structure. The plan
dimensions are 21.6m x 18m and total height is 24.5m. The thickness of the floor is
100mm. Cross column section is 450x450mm. The layout of structural elements at
each floor is symmetrical in both plan directions. The critical damping of superstruc-
ture is assumed to be 5% in all the modes. HDRBs are placed between the column-
ends and substructures. One rubber bearing is provided under each column. In this
study, these bearings are modelled by bilinear hysteretic model as shown in Fig. 3.
The model is presented in some specifications [7, 8].

Fig. 3: Bilinear model for the bearings

[7, 8]

The step-by-step design procedure for determining the size of HDRBs and total thick-
ness of rubber in bearings are proposed as below:
Step 1: To determine the ground acceleration a g by Eq (1)
ag   I agR (1)
Where: agR is the peak value of ground acceleration on soil type A; I is the im-
portance coefficient of the building.
Step 2: Evaluation of the horizontal elastic acceleration spectrum at the period of 1
T 
second (S1) by Eq: TC  T  TD : S1  Se T   ag .S..2,5.  C  (2)
T 
Step 3: To determine the region coefficient F at the period of 1 (s) in Table 11.4-2 of
ASCE/SEI 7-10.
Step 4: To calculate dynamic coefficients SM1 and SD1 by Eq (3) & (4)
2
SM 1  F S1 (3); S D1  S M 1 (4)
3
Step 5: To apply the parameters of BD or BM based on the effective viscosity coeffi-
cient D.
Step 6: To select the period of the building at the horizontal displacement design
value, Td by Eq: 3T f  Td  3  s  (5)
Tf is the natural period of the superstructure.
612 N. A. Dung et al.

Step 7: Evaluate the effective horizontal stiffness of the bearings, K eff by Eq (6)
where W is the total vertical force caused by the dead load and live loads.
2
W W  2 
Td  2 => K eff   (6)
K d ,min g g  Td 
Step 8: To evaluate the design horizontal displacement of the bearings, DD by Eq (7)
 1 S T
DD   2  D1 d (7)
 4  BD
D D
Step 9: To determine the total rubber thickness (tr) by Eq:   D  t r  D (8)
tr 
Where  in the design displacement value following design standards
Keff tr
Step 10: To calculate the total cross-sectional area of the bearings, A  (9)
G
Where G is the shear modulus of HDRBs.
A
Step 11: To calculate the cross-sectional area of one bearing, Ai by Ai  (10)
n
Where n is the number of columns.
Step 12: To determine the design parameters of the bilinear model
HDRBs in this analysis are the same bearings in [5], in where, the bilinear parameters
C1 , C2 , and  cr are determined from the 1st cycle of sinusoidal loading tests and they
are presented in Table 1.
Table 1. Parameters in Nguyen (2017) Table 2. Design bilinear parameters
C1 (MPa) C2 (MPa) cr (MPa) K1 (N/mm) K2 (N/mm) Qy (N)
17.29 1.136 1.215 21613 1420 48600

These design parameters in Table 1 are converted into the above new design dimen-
A A
sions of HDRBs by Eqs: K1  b C1; Q y  Ab cr ; K 2  b C2 (11)
h h

Where Ab is cross-section area of the bearing, h is the total thickness of rubber.

Finally, the dimension of bearing is 200x200mm, the total thickness of each bearing is
32mm, and details of design parameters of the bearings are given in Table 2.

3.2 Ground motions of earthquakes and the analysis method of base-isolated

building supported to HDRBs
Fig. 4 (a) and (b) present the ground acceleration histories, which are designed for
earthquakes level-2 type-I and II by [8].
The time history analysis of both base-isolated and fixed-base buildings is carried out
using the computer program SAP2000 v14.2.2. A full three-dimensional representa-
tion of the structure is used in time history analysis of both base-isolated and fixed-
base buildings. Each isolator is modelled as a bilinear hysteretic element. The follow-
ing assumptions are made for the analysis of base-isolated building:
Horizontal response of base-isolated building … 613

• The superstructure is elastic at all time, the non-linear behaviour is restricted in

isolators only.
• All frame substructures are connected at each floor level by a diaphragm, which is
infinitely rigid in its own plane.
• Each floor has three degrees of freedom (two translations and one rotation) at-
tached to the center of mass of each floor.
• The isolation devices are rigid in the vertical direction and each has negligible
torsion resistance.
• Effect of infilled walls on stiffness of the structure is negligible. These are consid-
ered in computation of mass only.

(a) (b)
Fig. 4. Typical ground acceleration histories: (a) type-I, (b) type-II

3.3 Seismic response of the building

(a) (b)
Fig. 5. Base shear force at Joint 10: (a) type-I earthquake (b) type-II earthquake

(a) (b)
Fig. 6. Roof floor acceleration by (a) Earthquake type I (b) Earthquake type II
Table 2: Effectiveness of base isolation with HDRB in Y-axis

Peak value of roof floor accele-

Base shear at Joint 10 (KN)
Earthquakes ration (m/s2)
Fixed-base Base-isolated Fixed-base Base-isolated
Type-I (Kanto, 1923) 1374.16 582.57 10.39 4.91
Type-II (Kobe, 1995) 1229.04 470.83 13.91 6.18
614 N. A. Dung et al.

The comparison of seismic response in Y-axis of the fixed-base building to that of

base-isolated building is given in Fig. 5&6 and Table 2. The continuous lines present
the response of the fixed-base building and the hide lines describe the response of the
base-isolated building in these figures.

4 Conclusions

A seismic analysis of the six-storey building is carried out to investigate the effective-
ness of the use of HDRBs to protect buildings from earthquakes. This paper also pro-
poses a design procedure to determine the size of HDRBs. This procedure can support
to engineers to design base-isolated buildings using HDRBs.
The peak values of roof floor acceleration of the fixed-base building are 2.14 times
and 2.25 times higher than those of the base-isolated building under earthquakes type-
I and type-II, respectively. Moreover, the base shear forces of the fixed-base building
are 2.36 times and 2.61 times higher than the forces of the base-isolated building un-
der the earthquakes type-I and type-II, respectively. The base shear forces of the base-
isolated building depend on not only the peak values of ground acceleration, but also
the time duration of ground acceleration of earthquakes. These results show that using
HDRBs in buildings can improve the seismic capability of the structure against earth-
quakes.

References
1. Kelly, J.M.: Earthquake resistant design with rubber. 2nd edition, Springer-Verlag Berlin
Heidelberg, New York (1997).
2. Yoshida, J., Abe, M., Fujino, Y.: Constitutive model of high-damping rubber materials.
Journal of engineering mechanics, 130(2), pp 129-141 (2004).
3. Marioni, A.: The use of high damping rubber bearings for the protection of the structures
from the seismic risk. Jornadas portugesas de engenharia de estruturas, Lisbon, Portugal
(1998).
4. Gu, H., Itoh, Y.: Ageing effects on high damping bridge rubber bearing. Proceeding of the
6th asia-pacific structural engineering and construction conference, Kuala Lumpur, Malay-
sia (2006).
5. EC8.: Eurocode 8: Design of structures for earthquake resistance Part 1: General rules,
seismic actions and rules for buildings (EN 1998-1:2004). European Committee for Stand-
ardization (2004).
6. Nguyen Anh Dung (2017). “Effect of modeling on seismic response prediction of a pier
with high damping rubber bearings under earthquakes”. The 10th national Conference on
Mechanics, Hanoi, 8-9/12/2017, pp 191-197 (2017) (In Vietnamese).
7. American association of state highways and transportation officials-AASHTO.: 3rd Edi-
tion. Washington DC: Guide Specification for Seismic Isolation Design (2010).
8. Japan road association-JRA.: Specifications for highway bridges. Part V: seismic design.
Tokyo: Maruzen (2002).
 Evaluating the change in the properties of bitumen
during production, transport and construction of hot mix
asphalt in the conditions of Vietnam

K.H. Quang Trinh1, T. Nhan Tran4, and H.T. Tai Nguyen2,3

1 Ho Chi Minh City University of Transportation, Ho Chi Minh City, Vietnam
2
Ho Chi Minh City University of Technology and Education, Ho Chi Minh City, Vietnam
3 BMT Construction Investment J.S.C., Ho Chi Minh City, Vietnam
4
Ecole Centrale de Nantes, Nantes, France
tainht@hcmute.edu.vn; thien-nhan.tran@eleves.ec-nantes.fr

Abstract. In Vietnam, the total time of production, transport and construction of

hot mix asphalt (HMA) can last up to more than 10 hours depending on the dis-
tance from the asphalt plan to the construction site as well as the weather condi-
tions at that site. As a result, the properties of the onsite compacted asphalt mix-
ture may be far different from those obtained in the laboratory, giving rise to
unreliable predicted service life of the pavement. This study aims to evaluate the
change in the properties of bitumen during production, transport and construction
of hot mix asphalt. To simulate these processes, a 60/70 pen bitumen was heated
to 150˚C for 1h, 3h and 5h and stirred continuously to allows the interaction be-
tween bitumen and oxygen. Control samples were also prepared by maintaining
the bitumen at high temperature but not allowing the bitumen to interact with
oxygen. The properties of the conditioned samples were then evaluated using
viscosity test, penetration and softening point test, and Superpave performance
grading (PG) test. The results show that the properties of the control samples
almost do not change while those of the aged samples vary significantly with an
increase of ageing time. The more longtime the bitumen was exposed to high
temperature and interaction with air, the more stiffness gained by the bitumen.

Keywords: Hot mix asphalt, short-term ageing, hardening, performance grade.

1 Introduction

Asphalt binder is normally stored longtime in silo at a temperature of 135˚C before

heating up to approximately 150-160˚C for the production of asphalt mixture without
concern about quality degradation. This is because the surface area of bitumen exposed
to oxygen is very small when stored in bulk. However, during the production of hot
mix asphalt, the bitumen is spread into thin films of 5-10m during mixing with hot
aggregate, which enormously increase the surface area of the bitumen [1]. Under the
effect of high temperature and mixing of the pugmill, the oxidation of thin film bitumen
and the loss of volatile fractions within the bitumen is accelerated, giving rise to signif-
icant change in the properties of the bitumen [2]. After being aged in the pugmill, the
© Springer Nature Singapore Pte Ltd. 2020 615
C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_97
616 K.H. Q. Trinh et al.

bitumen will continue to interact with oxygen during storage in silo, transport in dump
truck and placement at the construction site. Since the total time from production to the
end of the placement of asphalt mixture commonly lasts more than 4 hours (sometimes
up to 10 hours) in Vietnam depending on the distance from the asphalt plan to the con-
struction site and the weather conditions at the site, the impact of short-term ageing
(ageing during production, storage and placement) to the change in the mechanical
properties of bitumen inside the asphalt mixture is significant and should be considered
in the design of asphalt mixtures and pavement thickness.
In order to simulate the short-term ageing of bitumen, the neat bitumen is condi-
tioned by high temperature and air interaction such that the loss of volatile fractions
within the bitumen and the oxidation are accelerated, and the ageing effect in the sim-
ulation test is equivalent to that caused by production, storage and placement of asphalt
mixture in the reality. Many methods have been proposed including the standardized
test, e.g. Thin film oven test (TFOT), Modified thin film oven test (MTFOT), Rolling
thin-film oven test (RTFOT), Rotating flask test (RFT), and non-standardized test. In-
terested readers are referred to the works [3, 4, 5, 6] and Refs. quoted therein for more
detailed discussions. Among standardized procedures, the RTFOT is widely used for
producing the short-term ageing of bitumen (AASHTO T240), during which the neat
bitumen is heated to 163˚C and exposed to an airflow of 4000ml/min for 85 minutes.
The aged bitumen after RTFOT will be used for checking the Superpave performance
grading of the bitumen [7]. In some project, specifications are provided for both neat
bitumen and RTFOT residue sample to control the change in properties of bitumen
during production, transport and placement of asphalt mixtures.
In recent years, rutting, the most common distresses of the system of roadway in
Vietnam, has attracted much attention of local authorities and researchers [8, 9, 10]
including the study of storage time on the change in properties of asphalt mixtures [11].
According to [11], the resistance to rutting of asphalt mixtures stored at high tempera-
ture increases with an increase of storing time. As a result, the mechanical properties
asphalt mixture placed at different sections of a project may be much different although
the job mix formula is the same. This effect when combined with wandering effect will
give rise to uneven rutting depths in reality. Although, short-term ageing is an important
effect and is widely known around the world, it has not been considered yet in Viet-
namese standards for design and construction of asphalt pavement.
Aims of this study is to investigate the change in properties of bitumen that was aged
by heating and interacting with oxygen in the air promoted by stirring action. The re-
sults of this work will contribute to the understanding of the change in mechanical
properties of asphalt mixtures observed in the reality—namely, the hardening effect
and decrease of adhesion between aggregate and bitumen of aged asphalt mixtures.

2 Materials and methods

2.1 Materials

The bitumen used in this study is one type of 60/70 pen bitumen, a widely used bitumen
in Vietnam, provided by BMT Construction Investment J.S.C. The engineering
Evaluating the change in the properties of bitumen … 617

properties of this bitumen are the same as those presented in [11]. Granite aggregate
from the quarry in the South of Vietnam was used for evaluating the affinity between
aggregate and bitumen.

2.2 Sample preparation

The neat bitumen was collected from silo and poured into tin cans then heated up to
150˚C by infrared cooker as illustrated in Figure 1 for the following 3 levels of heating
time: 1, 3, and 5 hours. In order to promote the bitumen-oxygen interaction, a stirrer
was installed and ran continuously during heating process. For comparison purpose,
control samples were also prepared by heating up the bitumen in an oven such that the
oxidation between the bitumen and oxygen is prevented. The heating time and temper-
ature was the same as those used for aged sample. Conditioned bitumen samples were
then used for evaluating the change in the properties compared to those of the original
neat bitumen.

Fig. 1. Test setup for ageing bitumen. (1) Infrared cooker. (2) Bitumen contained in a tin can. (3)
Stirrer.

2.3 Methods

The neat and conditioned bitumen samples were tested for fundamental properties such
as viscosity, penetration and softening point at 135˚C. The methods used are those de-
scribed in ASTM D5, ASTM D36 and ASTM D 4402, respectively. The affinity be-
tween aggregate and bitumen was also evaluated using the detachment in boiling water
method as described in EN-12697-11 and TCVN 7504. In order to well understand the
behavior of bitumen, Superpave performance grading (PG) test using the dynamic shear
rheometer (DSR) device was also performed according to the specification of SHRP-
A-379 [7]. At least three replicates of test were performed in this study except the PG
test.

3 Results and discussions

For the sake of simplicity, the sample’s names were coded as S-Axx and C-Axx where
the letters S and C denote aged sample and control sample, respectively; and Axx
618 K.H. Q. Trinh et al.

denotes the heating time of the bitumen on the cooker or in the oven. For example, S-
A01 is the bitumen sample that was heated and stirred for 1 hour; N-A05 is the bitumen
sample that heat heated for 5 hours in the oven. The experimental testing results will be
presented and discussed in next Subsections.

3.1 Viscosity

Figure 2-1 shows the changes in viscosity of aged and control samples. It can be seen
that both control and aged samples slightly increase its viscosities with increasing heat-
ing time, and the viscosity of aged samples is higher than those of control sample due
to the interaction of bitumen in aged samples with oxygen in the air. However, the
viscosity of aged and control samples is almost at the same level of magnitude (in the
range of 425.8 to 691.8 cP). Therefore, it seems to be that short-term ageing effect does
not affect much the workability of asphalt mixtures.

3.2 Affinity between aggregate and bitumen

Figure 2-2 shows the changes in affinity between aggregate and bitumen of aged bitu-
men and control samples. It can be clearly observed that the adhesion grade between
aggregate and bitumen of the control samples almost does not change while that of aged
sample degrades approximately 1 grade with increasing 2 h of heating time. Conse-
quently, the hardening of aged bitumen observed in Section 3.2 is compensated by the
degradation of its adhesion with aggregate.

6
1000
4 4
Adhesion grade

800 691.8 3.7

630.8
Viscosity (cP)

545.8 4
600 4
400 510 521.8 2 2.7
425.8 S-Axx S-Axx 2
200
C-Axx C-Axx
0 0
1 3 5 1 3 5
Heating time (hours) Heating time (hours)

Fig. 2. Change in viscosity and affinity between aggregate and bitumen.

3.3 Penetration and softening point

Figures 3 shows the changes in penetration and softening point of aged bitumen and
control samples. It can be clearly observed that the control samples slightly change its
penetration and softening point. In contrast, the aged samples harden with heating
time—namely, its penetration decreases approximately 4dmm and softening point in-
creases approximately 1.2˚ with increasing heating time by 1 hour.
Evaluating the change in the properties of bitumen … 619

3.4 Superpave performance grade (viscoelastic behavior at 10rad/s, 12%)

Since the mechanical properties of control samples do not change much with heating
time, only viscoelastic characterization of aged samples was performed. It was ob-
served that the shear modulus increases, and the phase angle decreases with increasing
ageing time. In other words, the elastic part of the behavior of aged bitumen increases
while the viscous part of the behavior decreases with an increase of ageing time. This
verifies that the more the bitumen (asphalt mixture) is exposed to ageing, the more the
resistance to rutting is gained by the bitumen (or asphalt mixture) as observed previ-
ously in [11]. Since the stiffness of bitumen is increased and the affinity between ag-
gregate and bitumen is degraded, the resistance to fatigue of aged bitumen is expected
to decrease with increasing ageing time, which will be verified in future works.

80 60 56.9
S-Axx
Penetration (10-1mm)

Softeing poin (˚C)

C-Axx 54
60 55
63.7 51
62.7 61.7
40 50
48.1

20 34.5 45 49
S-Axx 28.2 47.5 47.8
C-Axx
0 40
1 3 5 1 3 5
Heating time (hours) Heating time (hours)

Fig. 3. Change in penetration and softening point of S-Axx and C-Axx samples.

4 Conclusions

The change in the properties of bitumen when going through the production, transport
and placement was successfully reproduced by using a simple method in which the
bitumen was heated and stirred to promote the interaction with oxygen in the air. The
hardening of bitumen when stored in bulk was also reproduced by heating the bitumen
samples in an oven to prevent the interaction with oxygen, coded as control samples.
Based on the results obtained in this study, the following conclusions could be drawn:
(i) The properties of the control samples almost do not change with heating time; there-
fore, the storage of hot bitumen in large silo does not affect much the properties of the
bitumen. (ii) The properties of aged samples change significantly with heating time,
which verifies the effect of high temperature and oxidation during the production,
transport and construction phase on the properties of bitumen in the mixture. (iii) The
softening point of bitumen increases while the penetration decreases with increasing
ageing time; therefore, the aged bitumen is less susceptible to high temperature than
neat bitumen. (iv) The viscoelastic characterization of aged bitumen at angular fre-
quency 10 rad/s and shear strain 12% shows that the shear dynamic modulus increases,
and the phase lag angle decreases with increasing ageing time. As a result, the elastic
620 K.H. Q. Trinh et al.

part of the behavior of aged bitumen increases while the viscous part of the behavior
decreases with an increase of ageing time. This verifies that the more the bitumen (as-
phalt mixture) is exposed to ageing, the more the resistance to rutting is gained by the
bitumen (asphalt mixture). (v) The affinity between aggregate and bitumen degrades
due to ageing effect, approximately 1 grade lost after an addition of 2 h of heating time.
The stiffness gained during ageing is compensated by the degradation of its adhesion
with aggregate. Therefore, the resistance to fatigue of aged bitumen is expected to de-
crease with increasing ageing time, which will be verified in future works.
Consequently, the resistance to rutting of asphalt mixtures subjected to longer stor-
age time is better than those subjected to shorter storage time. Meanwhile, the resistance
to fatigue of aged asphalt mixtures are predicted to decrease with increasing ageing
time. In addition, the storage time during transport and construction cannot be con-
trolled, which means that one cannot control the quality of aged asphalt mixtures.
Therefore, the short-term ageing of asphalt mixture should be limited and is not recom-
mended to be used as a method for improving the resistance to rutting of asphalt pave-
ment.

References
1. Read, J., Whiteoak, D.: The Shell bitumen handbook. Thomas Telford publishing, London
(1990).
2. Hofko, B. et al.: Effect of short-term ageing temperature on bitumen properties. Road Ma-
terials and Pavement Design 18(S2), 108-117 (2017).
3. Airey, G.D.: State of the Art Report on Ageing Test Methods for Bituminous Pavement
Materials. International Journal of Pavement Engineering 4(3), 165-176 (2003).
4. Choquet, F.S.: Bitumen ageing. Centre de Recherches Routieres, Brussels (1993).
5. Khalid, H.A., Walsh, C.M.: A new approach for the accelerated ageing of porous asphalt
mixtures. ICE Transport 153(3), 171-181 (2002).
6. Lee, M., Tia, M., Ruth, B., Page, G.: Comparison between the aging processes for asphalt
mixtures and those for asphalt binders. Progress of superpave (superior performing asphalt
pavement): Evaluation and Implementation, West Conshohocken (1997).
7. Harrigan, E.T., Leahy, R.B., Youtcheff, J.S.: The Superpave Mix Design System Manual of
Specifications, Test Methods, and Practices. Strategic Highway Research Program, Wash-
ington, (1994).
8. Decision 858-MOT.: Applying current technical standards to strengthen the quality man-
agement of design and construction of hot mix asphalt pavement of high traffic roadway.
Ministry of Transport, Hanoi (2014).
9. Nguyen, H.T.T.: Modelling the mechanical behaviour of asphalt concrete using the Perzyna
viscoplastic theory and Drucker–Prager yield surface. Road Materials and Pavement Design
18(S2), 264-280 (2017).
10. Tran, T.N., Nguyen, H.T.T., Nguyen, K.S., and Nguyen, N.T.H.: Semi-flexible Material:
The Sustainable Alternative for the Use of Conventional Road Materials in Heavy-Duty
Pavement. Congrès International de Géotechnique – Ouvrages – Structures, Ho Chi Minh
City (2018).
11. Nguyen, H.T.T. and Tran, T.N.: Effects of crumb rubber content and curing time on the
properties of asphalt concrete and stone mastic asphalt using dry process. International Jour-
nal of Pavement Research and Technology 13(3), 238-244 (2018).
 Creep behavior and rutting resistance of asphalt
pavements by experimental testing and Finite Element
modelling

Van Bac Nguyen1, Chau Phuong Ngo2, Ngoc Bay Pham2, Thanh Phong Nguyen2, Van
Phuc Le2, Van Hung Nguyen2
1
University of Derby, Derby DE22 3AW, UK
2
University of Transport and Communications, Campus in Ho Chi Minh City, Vietnam
vb.nguyen@derby.ac.uk; ncphuong@utc2.edu.vn

Abstract. Asphalt pavement rutting is a major safety concern because it affects

the handling of vehicles. Research into effectively predicting of asphalt pavement
rutting is extremely necessary as it can help develop an optimal design of asphalt
mixture against rutting. The primary objectives of this study are to develop nu-
merical models to investigate the asphalt pavement rutting. Previous study’s ex-
perimental tests including triaxial repeated compression loading and wheel track
testing were used to evaluate the visco-elasto-plastic behavior and rutting re-
sistance of different pavement mixes. Based on the test results, sets of visco-
elasto-plastic creep parameters for the asphalt pavement mixes were character-
ized and related to their rutting resistances. Numerical simulations using three-
dimensional Finite Element models were developed to simulate both the triaxial
compression and wheel track testing. A ‘strain hardening’ creep material model
with the material parameters developed from experimental testing was employed
in the Finite Element modelling to the time-dependent characteristics of the as-
phalt concrete pavements. The Finite Element results were validated against the
experimental results to demonstrate that the Finite Element simulations can be
used to effectively evaluate the rutting behavior of asphalt concrete pavements.

Keywords: Viscoelastic, Creep, Asphalt Pavement Rutting, Wheel Track Test-

ing, Finite Element Modelling.

1 Introduction

Rutting is one of the most serious distresses in asphalt pavements affecting the pave-
ment service life and the handling of vehicles. In the last three decades, there have been
a lot of studies focused on investigating the material characteristics and prediction of
rutting resistance of asphalt pavements. A wide number of numerical studies for ana-
lyzing rutting of asphalt pavements with more realistic constitutive models have been
conducted that included more practical models in different Finite Element programs
such as ABAQUS and ANSYS [1-4]. In [1] a 3D nonlinear Finite Element analysis
with a viscoelastic model for asphalt concrete was used to analyze flexible pavements
under moving loads, however, due to computational difficulties, the analyses were

© Springer Nature Singapore Pte Ltd. 2020 621

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_98
622 V. B. Nguyen et al.

conducted for only ten loading cycles. The modeling of asphalt mixes using a visco-
plastic (creep) model available in ABAQUS was addressed in [2,3], and a 2D model to
analyze in-service pavements was developed. Recently, 3D analysis simulated the
HWTD testing of asphalt mixes using creep model in ANSYS was developed [4]. In
these studies, the viscoplastic (creep) model used was in the ‘time hardening’ form and
moving loads were treated as a static load, but due consideration was given to the du-
ration of loading. However, as the creep ‘time hardening’ model was only developed
for the assumption of the constant load, it might not be applicable for the moving load
situation [5]. In addition, from application consideration, it is important to examine
rutting potential under larger loading cycles.

The primary objectives of this study are to develop 3D Finite Element (FE) models to
simulate the laboratory testing of asphalt mixes in triaxial compression and wheel track
tests. The triaxial repeated load creep and creep recovery test results have been used to
obtain the associated visco-elasto-plastic (creep) parameters for asphalt concrete mixes.
A general multi purposes Finite Element program, ANSYS, was used for this study. A
creep model of the ‘strain hardening’ form was employed to represent the time-, stress-
and temperature-dependent behavior of asphalt concrete materials. A sensitivity study
was undertaken to examine the influence of some selected parameters on rutting.

2 Experimental Testing

Experimental testing data of a previous study [3] was used in this study. Three asphalt
mixes which were selected to represent a wide range of applications, from high traffic
freeways to low volume municipal roads. The mixes included HL 3: conventional dense
graded Marshall surface course mix; SMA L and SMA G: two Stone Mastic Asphalt
12.5 mm surface course mixes. The triaxial repeated load creep testing was carried out
in cylindrical samples 100 mm diameter × 150 mm height. The triaxial repeated load
creep and creep recovery tests were conducted. The testing was carried out at a constant
temperature of 50oC. The test results to investigate the visco-elasto-plastic behaviour
of the asphalt mixes and to develop the material parameters used in the Finite Element
modelling. The Hamburg Wheel Tracking Device (HWTD) testing general view and
diagram are shown in Fig. 1. The wheels are solid rubber (50 mm wide), and the load
applied to the wheels is 710 N. The test path is 230 mm long and the average speed of
each wheel is approximately 1.1 km/h. All the HWTD testing was carried out at a tem-
perature of 50oC. The number of wheel passes being used was typically 20,000.
Creep behavior and rutting resistance of asphalt … 623

Fig. 1. General view (a) and testing diagram (b) of the Hamburg Wheel Tracking Device.

3 Material Characterization

Visco-elasto-plastic model in the form of ‘strain hardening’ was used in this research
and it can be written in the form of creep strain rate which is expressed as a function of
stress, strain and temperature:

𝐶
𝜀̇𝑐𝑟 = 𝐶1 𝜎 𝐶2 𝜀𝑐𝑟3 𝑒 𝐶4/𝑇 (1)

Where 𝜀̇𝑐𝑟 is the creep strain rate, 𝜎 is the equivalent stress, 𝜀𝑐𝑟 is the creep strain at the
time t, and 𝐶1 , 𝐶2 , 𝐶3 , 𝐶4 are the creep parameters that are functions of state of stress,
material type and temperature. As the temperature in the test is fixed at 50ºC, the pa-
rameters 𝐶1 , 𝐶2 , 𝐶3 together with elastic parameters were first determined from the tri-
axial repeated load creep test and then calibrated based on the rutting measured in the
HWTD as shown in Table 1.

Table 1. Material parameters of asphalt mixes.

Mix Material parameter

Elastic Creep
𝐸𝑎𝑐 (kPa) 𝜇𝑎𝑐 𝐶1 𝐶2 𝐶3
HL 3 950000 0.41 2.96810-5 4 -1.703
-8
SMA L 800000 0.42 2.83710 4.727 -3.545
SMA G 800000 0.42 2.28410-7 6.238 -3.762
624 V. B. Nguyen et al.

4 Finite Element Modelling

Finite Element simulations were conducted using ANSYS (ANSYS software, version
2018). The samples were presented by 3D solid elements of the type SOLID45; this
element is defined by eight nodes having three degrees of freedom at each node trans-
lation in the nodal x, y, and z directions. The element has plasticity, creep, swelling,
stress stiffening, large deflection, and large strain capabilities. The element size is of
approximate size of 10 mm and 20mm in the triaxial and HWTD models, respectively
A full Newton-Raphson method was used for the iterative procedure and an implicit,
static analysis was employed.

Fig. 2 shows the FE model setup for the wheel track test HWTD which also shows the
mesh. The footprint of the HWTD solid rubber wheel on the surface of asphalt mix
sample, measured at a testing temperature of 50 oC, had a rectangular area of 50 mm in
width  28.5 mm in length, and this was used to calculate the loading time and contact
pressure. The test was done after 20,000 loading passes which was about 4,200 seconds.
The bottom and the four surrounding vertical boundaries were set to be confined with
restricting displacement in all directions. For the load of 710 N applied in this area, it
was equivalent to a uniform loading pressure of 0.5 MPa.

Fig. 2. FE model setup of the triaxial compression testing which shows loading, boundary con-
ditions and mesh.

5 Results and Discussion

The FE modelling was first verified by comparing the results obtained from the triaxial
repeated load creep testing model with the theoretical axial compression creep model
as presented in Eq. (1). In the FE model, the creep parameters 𝐶1 , 𝐶2 , 𝐶3 for each asphalt
mix as shown in Table 1 were used. The FE results in term of creep strain for a typical
sample, HL 3, are shown in Fig. 3 which also shows the comparison between FE and
theoretical results obtained from Eq. (1). It can be seen clearly that the FE model pre-
dicted accurately the axial compression and creep behavior of the asphalt mix under
axial constant loading.
Creep behavior and rutting resistance of asphalt … 625

Fig. 3. FE creep results for axial compression loading (a) and comparison with theoretical axial
compression creep model as presented in Eq. (1), demonstrating via the typical sample HL 3.

The measured rut depth versus time relationship with rutting predicted in FE models at
any time of loading up to about 4500 seconds for three samples are shown in the Fig.
4. It is observed that the difference between FE model prediction and measured rutting
depth is as high as 4% at around 2250 seconds and 4200 seconds. The maximum dif-
ferences between the predicted rutting values and the HWTD values are about 4%, 8%
and 16% for samples HL 3, SMA L and SML G, respectively. The FE predicted defor-
mation shape is illustrated in Fig. 5. It shows that the FE model is very capable of
capturing the rutting depth due to plastic flow and irrecoverable deformation in the
sample.
3.0

2.5
Experiment HL 3
2.0
Finite Element HL 3
Rutting depth (mm)

1.5 Experiment SMA L

1.0 Finite Element SMA L

Experiment SMA G
0.5
Finite Element SMA G
0.0
0 500 1000 1500 2000 2500 3000 3500 4000 4500
Time (s)
Fig. 4. Rutting depths predicted by FE modelling and measured in HWTD tests for the HL 3 mix.

A sensitivity study was undertaken and it was found that the rutting depth was very
sensitive with 𝐶1 and 𝐶3 , and increased with increasing 𝐶1 and 𝐶3 values. That could be
due to 𝐶1 and 𝐶3 were directly related to the amount of asphalt cement content, and
increased with increasing asphalt cement content in a mix. In addition, the rutting depth
decreased with increasing 𝐶2 values as that related to reducing stresses applied on the
sample.
626 V. B. Nguyen et al.

Fig. 5. FE predicted deformation shape of the HL 3 mix after 20,000 passes.

6 Conclusion

The Finite Element models were developed to simulate the laboratory testing of asphalt
mixes in triaxial compression and HWTD tests. The characterized material properties
determined from axial tests employed in the FE models and validated against the
HWTD test results which show very good agreement in rutting depths. The sensitivity
study conducted using the FE models gained more understanding on the influences of
creep parameters to the rutting resistance of asphalt mixes. It can be concluded that the
FE simulations can be used to effectively evaluate the rutting behavior of asphalt con-
crete pavements.

Acknowledgement

This study is part of the FIRST project no. 35/FIRST/1a/UTC. We acknowledge FIRST
Project, Ministry of Science and Technology, Vietnam, for sponsoring this project.

References
1. Zaghloul, S. M., White, T. D.: Use of three dimensional Finite Element program for flexible
pavement. In Transportation Research Record 1388, 60 – 69, TRB, National Research Coun-
cil, Washington D.C. (1993).
2. Huang, Y. H.: Pavement analysis and design. Prentice Hall, Inc., Englewood Cliffs, New
Jersey (1993).
3. Uzarowski, L.: The development of asphalt mix creep parameters and Finite Element mod-
elling of asphalt rutting. PhD thesis, University of Waterloo (2006).
4. Nahi, M.H., Kamaruddin, I., Ismail, A., Al-Mansob, R.A.: Finite Element model for rutting
prediction in asphalt mixes in various air void contents. Journal of Applied Sciences, (14),
2730 – 2737 (2014).
5. Betten, J.: Creep mechanics. 2nd edn. Springer, Germany (2005).
 Influence of High Temperature on Non-Silicate Based
Activated Blast Furnace Slag

Dr. Virendra Kumar1,3, Amit Kumar2,3, Dr. B.K. Prasad1,3

1 Associate Professor
2 Research Scholar
3 Department of Civil Engineering, National Institute of Technology, Jamshedpur,
Jharkhand, India
kumarvirendra57@gmail.com, amitsanajsr93@gmail.com, bkkpp@yahoo.co.in

Abstract. The energy consumption and the greenhouse gases emission in order to
manufacture Portland cement (PC) is very high. To find an alternative, the researchers
initially replace the PC partially by Fly Ash. Later on, the researchers introduce a new
binder which is being manufactured by the activation of ground granulated blast
furnace slag (GGBFS). The paper presents the detail of the binder, manufactured by
GGBFS and mild alkali as activator having the pH value of 10. The binder is
manufactured by simple blending in which 85% is GGBFS, and 15% is the chemical
activator. The test results of the alkali activated binder concrete (AABC) for
compressive strength, splitting-tensile strength and flexural strength were found after
exposure to elevated temperature 100 ℃, 200 ℃, 300 ℃, 400 ℃, 500 ℃, 600 ℃, 700
℃ and 800 ℃. It has been observed that these results of the AABC at elevated
temperature were very much comparable with that of the ordinary Portland cement
concrete (OPCC). The deterioration in strength starts from 200 ℃ onwards. No
spalling of the alkali activated binder concrete (AABC) has been observed at elevated
temperature even at 800 ℃.

Keywords: Ground granulated blast furnace slag, alkali, chemical, activator, elevated
temperature.

1 Introduction

Since Portland cement is an ancient commodity, hence it has become the governing
binder in concrete for construction. Industrial production and supply of the cement
had begun in the 19th century. The per year production of cement is 3 billion tonne
and expected to increase up to 4 billion tonne. This growth is majorly due to meet the
demand and the major part played by Republic of China, India, Middle East and
Northern Africa. In 19th century, the cement production was begun with shaft kilns
which was later transform into rotary kiln as a standard equipment worldwide [1]. The
cement production as well as the demand is rising every day, the cement industries
facing several issues such as energy supply, carbon dioxide emission, raw materials
etc. Cement production makes the cement manufacturing third largest man-made
carbon dioxide (CO2). The CO2 is the resultant of lime heating which is prominently
used in the cement manufacturing process. This required change accomplished by
means of two methods i) by bringing the change in manufacturing and production
© Springer Nature Singapore Pte Ltd. 2020 627
C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_99
628 V. Kumar et al.

process, ii) by bringing the change in the chemical composition of cement [2]. In
order to reduce the energy consumption, the conventional fuel is being substituted by
waste and biomass. In some of the European countries, the cement manufacturing
plant has substituted the conventional fuel by the high calorific waste material. Some
of the cement manufacturer achieved 100% replacement level of conventional fuel by
using suitable waste materials [2]. In the year 1996, the carbon dioxide emission is
around 1.12 tonne CO2/tonne of cement production, which reduces to the average of
0.719 tonne CO2/tonne cement in the year 2010 [3]. Another way to reduce the
emission of CO2 is the incorporation of supplementary cementitious material (SCM)
like volcanic ash, Ground Granulated Blast Furnace Slag (GGBFS), Fly Ash.
Volcanic ash, GGBFS and Fly Ash are also considered as non-hazard pollutant when
dumped directly in nature. These SCMs exabit similar property when mixed with
cement and also help to bring down the overall clinker consumption for
manufacturing cement.
Hence it is required to develop a binder or to develop an alternative system
in which the energy consumption and CO2 emission is very minimal. In the search of
alternative, one binder has been developed namely, alkali activated binder. This alkali
activated binder is first developed in 1908 by and later pioneering work has been done
by Glukhovsky, Krivenko, Davidovits, and Palomo et al [4]. Alkali activated binder
(AAB) are formed by chemical reactions between the combinations of concentrated
aqueous solutions of alkali hydroxide, silicate or carbonate and solid aluminosilicate
powders. In the present study the alkali activated binder is prepared by using 85% of
GGBFS and 15% of powder blend activator.

2 Materials

GGBFS used in the study for manufacturing the binder is collected from
Jindal Pvt. Ltd. Conforming Indian Standard code IS: 12089:1987 [5]. The chemical
constituent of GGBFS used in the study is provided in table 1. The locally available
fine and coarse aggregates have been used in the study which meet all the
requirements of Indian standard IS 383:2016[6].
Table 1 Chemical constituent of GGBFS
Chemical Constituent %
SiO2 37.7
Al2O3 14.4
Fe2O3 1.1
CaO 37.3
MgO 8.7
MnO 0.02
SO3 0.4
LOI 1.4
Glass Content 92
Influence of High Temperature on Non-Silicate … 629

3 Methodology

3.1 Mixing, Casting and Curing

The mix of concrete is prepared by using alkali activated binder, coarse
aggregate fine aggregate, and water. The binder is blend with activator in such a way
that it acts as a single part system just like convention cement. The properties of AAB
and conventional OPC have been given in table 2. In the present study, the used AAB
is having edge over other AAB, where while preparing other AABC, there is a need to
prepare a solution prior for mixing. The process of preparation of concrete
manufactured from OPC and AAB are similar and the design mix has been done for
medium workability as given in table 3. The standard specimens of cube, cylinder,
and prism have been cast and keep for water and ambient air curing after demolding.
A reference mix has also been prepared by using Ordinary Portland cement (OPC) for
comparative study.
Table 2 Physical property of AAB and OPC
Physical Property AAB OPC

Fineness (m2/kg) 340 225

Specific gravity 2.81 3.14

Soundness Le-chatelier (mm) 1.00 2.00

Standard Consistency 35% 33%

Initial Setting time (minutes) 50 65

Final Setting Time (minutes) 95 320

Table 3 Detail of concrete mix proportion per meter cube
Concrete ingredients Type of cementitiuos materials

AAB OPC

Type of Binder (kg/m3) 350.94 350.94

Fine Aggregate (kg/m3) 667.07 680.47

Coarse Aggregate (kg/m3) 1187.75 1210.37

Water (kg) 186 186

4 Analysis of Test Result

4.1 Result of Compressive strength, Flexural Strength and Split Tensile

Strength
Evaluation of compressive strength, the standard cube of size 150 mm have been
cast and cured in air and in water for 1, 3, 7, 14, 28 days of maturity. The specimens
630 V. Kumar et al.

prepared with OPC have been cured in water only as reference concrete mix. The test
results of compressive strength have been depicted in fig. 1. The trends of the
compressive strength development of the alkali activated binder concrete (AABC) at
different maturity days viz. 1, 3, 7, 14 and 28 days are being always higher than that
of the ordinary Portland cement concrete (OPCC). In similar process, the standard
cylindrical concrete specimens (height/length = 300 mm, diameter = 150 mm) were
cast for split tensile strength with AAB and OPC binder by using same mix
proportion of the ingredients. The test results of split tensile strength after 1, 3, 7, 14
and 28 days maturity have been shown in fig. 2. The prism specimen of cross section
(width & depth) 100 mm and length 500 mm have been cast for flexural strength test.
After 1, 3, 7, 14 and 28 days of maturity of the concrete specimens the flexural
strength test have been conducted and the test results are being shown in fig. 3.
Similar to the compressive strength development, it can be seen from fig. 2 and fig. 3
that the split tensile strength and flexural strength of the alkali activated binder
concrete (AABC) are always higher than that of the ordinary Portland cement
concrete (OPCC), respectively.

Fig. 1 Compressive strength vs age of air & water cure AAB and OPC concrete

Fig. 2 Split tensile strength vs age of air & water cure AAB and OPC concrete

Fig. 3 Flexural strength vs age of air & water cure AAB and OPC concrete
Influence of High Temperature on Non-Silicate … 631

4.2 Result of Compressive strength, Flexural Strength and Split Tensile

Strength at elevated Temperature
After proper curing and maturity for the desired days, the standards
specimens have been exposed to 100 ℃, 200 ℃, 300 ℃, 400 ℃, 500 ℃, 600 ℃, 700
℃ and 800 ℃ temperature. The rate of heating on the concrete specimens inside the
muffle furnace has been taken 10℃/minute and keeps the desired exposure
temperature for 4 hours and then finally allowed to cool naturally in the furnace only.
After natural cooling to the room temperature, these specimens have been tested for
compressive strength, split tensile strength and flexural strength to evaluate the
residual strengths. Additionally, the change is weight has also been recorded before
and after heating.

Fig. 4 Percentage variation in compressive strength vs temperature.

Fig. 5 Percentage variation in split tensile strength vs temperature

Fig. 6 Percentage variation in flexural strength vs temperature

The percentage variations in compressive strength, split-tensile strength and

flexural strength vs temperature have been depicted in fig. 4, 5 and 6, respectively.
From the figure, it is clear that that there is a rise in strengths at 100℃ for AABC
which are being higher than that of the OPCC. This rise in strengths for AABC may
632 V. Kumar et al.

be due to participation of the unreacted binder particle at 100℃. Considerable

percentage losses in residual strength have been observed above 100℃. The
downfalls in strengths get very severe beyond 400℃ because of loss of the free water
and bounded water escape from the specimens. Overall, it has been observed that the
residual strengths of the AABC are lower than that of the OPCC. The appearances of
the micro-cracks have also been observed on the surface of specimen after exposure
to temperature above 400 ℃. This may be attributed to the increase in volume of
aggregate at desired elevated temperature which leads to ultimately cracks and
downfall in residual strengths. It can be seen that the general trends of deterioration in
residual strengths are similar for both AABC and OPCC. Hence, it gives a clear
indication that the AAB may be used in place of OPC, even exposure at elevated
temperature.

5 Conclusion
The present binder, namely alkali activated binder (AAB) is being manufactured
by simple blending in which 85% is GGBFS, and 15% chemical activator. It is clear
that the AAB is having potential to replace OPC. The used CAB doesn’t contain
neither clinker nor chemicals that are highly corrosive in nature. From the study, it is
observed that the AAB could also be manufacture by using weak alkali chemicals. It
has been found that the hardened strengths viz. compressive strength, split tensile
strength and flexural strength of the alkali activated binder concrete (AABC) are
always higher than that of the ordinary Portland cement concrete (OPCC). Even, the
trends of reduction in strengths of AAB concrete at different elevated temperature are
found to be similar to that of the OPC concrete. No spalling of the concrete has been
observed at different elevated temperature. Additionally, the air cured concrete with
alkali activated binder (AAB) exhibits comparable strengths as compared to OPC
concrete, resulting in the additional benefit of the reduction in construction cost.

References

1. Schneider, M., Romer, M., Tschudin, M., Bolio. H.: Sustainable Cement Produc
tion – Present and Future. Cement and Concrete Research 41 (2011) 642-650.
2. https://www.climatetechwiki.org/technology/energy-saving-cement.
3. Kumar, V., Kumar. A., Prasad, B.: Mechanical Behaviour of Non Silicate based al
kali activated ground granulated blast furnace slag, Construction and building Mate
rial 198(2019) 494-500.
4. Glukovsky V.D. Soil silicates. Gosstroiizdat, Kevin, 1959.
5. IS:12089-1987, Specification of granulated slag for the manufacture of Portland
slag cement.
6. IS: 383-2016, Coarse and fine aggregate for concrete – specification.
Influence of Linz Donawitz slag aggregates on
strength development of concrete

Brajkishor Prasad1, Virendra Kumar1*, Prince Singh2

1. Associate Professor, Department of Civil Engineering, National Institute of
Technology Jamshedpur.
2. Research Scholar, Department of Civil Engineering, National Institute of
Technology Jamshedpur.
*Corresponding author (kumarvirendra57@gmail.com)

Abstract. This paper aims to present the feasibility of Linz-Donawitz (LD) slag as an
alternative for traditional natural aggregates in concrete. The LD slag possesses good physical
and mechanical properties as an aggregate and it has been incorporated in concrete as a
replacement material for natural fine and coarse aggregates. Both the natural fine and coarse
aggregates were partially replaced by LD slag aggregates at 20%, 40%, 60%, 80% and 100%
respectively. Concrete of grade M35 was cast, cured and tested with standard specimens
following Indian Standard codes to study its mechanical strengths. The test results of
compressive strength of concrete with LD slag aggregates are presented in this paper. The
results obtained, show that there is an increase in compressive strength of the various grades of
concrete with an increase in percentage replacement of natural fine and coarse aggregates with
LD slag aggregates.

Keywords. LD slag aggregate, fine aggregate, coarse aggregate, workability, strength

of concrete

1 Introduction

Construction industries all around the world are facing a massive problem; of
shortage of natural resources such as fine and coarse aggregates to be used in concrete
[1-2]. It is estimated that about 25 billion tons of concrete is manufactured globally
each year, which amounts to more than 3.8 tons per person in the world [3-4]. The
construction sector was the second largest material consuming sector in India in 2007,
accounting for around 20% of all material demand. It is also the fastest growing sector
with regard to increases in absolute material consumption. Between 1997 and 2007,
the amount of material consumption grew by more than one billion tonne [5]. With
such growth rates, the construction sector is expected to surpass the agricultural sector
before 2020 and become the highest material consuming sector in the next few years
[3-5]. The predominant materials used in the construction sector are sand, stones, soil
and limestone. Thus, researchers have been looking for different alternatives for the
traditional construction materials. Many alternative materials are being used by the
researchers to find out their suitability in concrete, but the need of the hour is to look
for a more localized solution as cost is also an important factor [1-2, 6].

© Springer Nature Singapore Pte Ltd. 2020 633

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_100
634 B. Prasad et al.

Rapid industrialization is one of the most important aspects for the economic
growth of a country. But the desire to achieve high economic benefits has
resulted into an unorganized waste disposal system [7]. Wastes such as fly ash,
blast furnace and steel slags, rice husk and its’ ash etc. pose a serious threat to
the health of the people and other animals inhabiting near by the dumping
zones [3, 6-7]. LD slag is the byproduct of Linz Donawitz process and is
produced when pig iron is processed into crude steel. Every year Tata Steel
produces approximately 1 million tonne of LD Slag which needs to be
disposed properly at suitable locations [8]. One of the solutions to tackle both
the problems at once is to use industrial wastes as a substitute for traditional
concrete materials [9]. Many researchers have studied the behavior of
concretes containing slag under different severe conditions and found that both
the mechanical properties and durability aspects of these concretes are
acceptable [7, 10-13]. But very limited work has been done worldwide to use
LD slag as an alternate civil engineering material. LD slag is a by-product of
steel making units resulting from pig iron refining process using LD converters
which is one of the important waste materials from all the integrated steel
plants [8]. Hence, an effort has been made in this research work to use LD slag
as a substitute for natural fine and coarse aggregates [14]. LD slag used in this
work was crushed into the specified sizes and then weathered for a period of 6
months before using it for making the concrete.

2. Material Properties

In this study, Portland slag cement (PSC) of Lafarge brand conforming to IS 455-
1989 has been used as binder as shown in table 1.
Table 1 Physical properties of cement
Sl. No. Particular of tests Results
1 Consistency 30.5%
2 Fineness 4%
3 Specific gravity 2.92
4 Setting time Initial 108 min
Final 258 min
5 Soundness(by Le-Chatelier apparatus) 1 mm (expansion)
6 Compressive 3 days 17.8 N/mm2
strength
7 days 24.85 N/mm2
28 days 36.75 N/mm2

Locally available natural fine aggregate (NFA), sand collected from Kharkhai river
has been used in the research work. Non-flaky, non-elongated and angular limestone
aggregates passing through 20 mm sieve and 10 mm sieve has been used as natural
course aggregate (NCA). LD slag used has been supplied by TATA Steel LTD.
Jamshedpur. Physical properties of the materials used in this work are shown in table
2 and table 3. The LD slag used was provided in two different sizes, LD coarse
Influence of Linz Donawitz slag aggregates … 635

aggregate (LDCA) varying between 4.75 mm to 20 mm and LD fine aggregate

(LDFA) below 4.75 mm in size. The LDCA provided were non-flaky, non-elongated
and had a smooth surface texture [2].

Table 2: Physical properties of NFA, LDFA, NCA and LDCA

Sl. Particular of tests NFA LDFA NCA LDCA
No.
1 Fineness modulus 2.83, Zone –II 1.93, Zone-III 6.60 6.94
2 Specific gravity 2.59 2.73 2.83 2.92
3 Free moisture 0.20% 3.9 % 0.10% 0.60 %
4 Water absorption 0.80% 6.46 % 0.35% 0.84 %
5 Bulk Compacted 1610 Kg/m3 1920 Kg/m3 1782 Kg/m3 2109 Kg/m3
Density Loose 1560 Kg/m3 1787 Kg/m3 1606 Kg/m3 1948 Kg/m3
6 Bulking of sand Maximum Maximum
bulking of bulking of LD
natural sand slag is 15.70 -------- --------
is 23.1 % at % at of 10 %
3% moisture moisture
6 Crushing Strength -------- -------- 17.6 % 13.35%
7 Impact Value -------- -------- 16.4 % 13.06%

The fineness modulus is an index number which represents the mean particle size of
the aggregates. Free moisture is the amount of water present in the pores of aggregate
when they are availablefreely in nature while water absorption is the total amount of
water present in aggregates in saturated surface dry (SSD) condition. The properties
of LD slag aggregates are well comparable to that of the natural aggregates, hence,
LD slag can potentially be used as aggregates in concrete.

3. Experimental Program

Conventional concrete of grade M35 was designed, cast and tested in the laboratory.
The concrete mixes were prepared and designed in order to obtain a slump value of
60-100 mm. Concrete cube, cylinders and prisms specimens of size 150 mm × 150
mm × 150 mm, 150 mm × 300 mm and 100 mm × 100 mm × 500 mm respectively
were cast at a temperature of (25±2)°C, cured under water and then tested for their
compressive, split tensile and flexural strength at the age of 28 days. The specimens
were cast at first using natural fine and coarse aggregates and then by replacing the
natural fine and coarse aggregates with LD slag aggregates in specified intervals of
various percentages. An average of 6 specimens was cast for testing each property of
the concrete mixes.

4. Results and Discussions

4.1 Workability
636 B. Prasad et al.

The workability of fresh concretes for different proportion of fine and coarse
aggregate replacement with LD slag was tested by using Slump Cone Test method.
Fig. 1 shows the relationship between the slump of concrete and the ratio of slag for
different concrete mixes.

0% LDFA 20% LDFA 40% LDFA

120 60% LDFA 80% LDFA 100% LDFA
100
Slump Value (mm)

80
60
40
20
0
0 20 40 60 80 100
NCA replaced with LDCA (%)
Fig.1 Workability vs aggregates replacement

The results indicate that the workability of concrete decreased significantly with the
increase in the percentage replacement of NFA with LDFA. The reduction in
workability of the mixes may be attributed to the fact that the LDFA used were much
finer than that of the NFA. The workability also decreased with the increase in the
percentage replacement of NCA with LDCA; however, this reduction has found to be
comparatively lower as depicted in fig. 1. This reduction in workability was may be
due to the porous nature of the slag aggregates.
4.2 Mechanical Strength
The compressive strength, split tensile strength and flexural strength of the concrete
specimens has been represented in the figures 1-3 respectively.

0% LDFA 20% LDFA 40% LDFA

44,00 60% LDFA 80% LDFA 100% LDFA
Compressive Strength (MPa)

42,00
40,00
38,00
36,00
34,00
32,00
30,00
0 20 40 60 80 100
NCA replaced with LDCA (%)

Fig.2 Compressive strength vs aggregates replacement

Influence of Linz Donawitz slag aggregates … 637

LDFA 0% LDFA 20% LDFA 40%

Split Tensile Strength (MPa) LDFA 60% LDFA 80% LDFA 100%
3,80
3,70
3,60
3,50
3,40
3,30
3,20
3,10
0 20 40 60 80 100
NCA replaced with LDCA (%)

Fig.3 Split Tensile strength vs aggregates replacement

LDFA 0% LDFA 20%" LDFA 40%"

LDFA 60%" LDFA 80%" LDFA 10 0%"
5,20
Flexural Strength (MPa)

5,10
5,00
4,90
4,80
4,70
4,60
4,50
0 20 40 60 80 100
NCA replaced with LDCA (%)

Fig.4 Flexural strength vs aggregates replacement

The maximum compressive strength of 42.20 MPa was achieved by the concrete mix
with 20% & 80% replacement, split tensile strength of 3.68 MPa was achieved at a
replacement of 40% replacement of both NFA & NCA and flexural strength of 5.10
MPa was achieved by the concrete mix with 60% & 80% replacement of NFA &
NCA with LD slag aggregates. Also, the strength of the concretes made with 100%
LDFA and 100% LDCA, separately was in accordance with the reference concrete.
LDFA being much finer than NFA reduces the number of voids in the concrete
making it less permeable. Also due to the shape and good mechanical properties of
LDCA, it makes better bonding with the matrix and hence increases the strength of
the concrete. However, LDFA is much finer than NFA due to which in the case of
100% replacement the gradation of particles in the concrete gets disturbed resulting in
production of weak concrete with a lot of voids.
638 B. Prasad et al.

Conclusions
The replacement of NFA and NCA with LDFA and LDCA decreases the workability
of concrete, especially, in higher grades; which can be taken care of by the use of
plasticizer. Also, the compressive strength, split tensile strength and flexural strength
of concrete made with partial replacement of natural aggregates with LDFA and
LDCA were found to be comparable with that of the referral concrete. As some of the
concrete mixes containing LDFA and LDCA possess higher compressive strength
than that of normal concrete; henceforth, LD slag aggregates can be adopted as a
substitute for the natural aggregates.

References
1. Md. Safiuddin, Mohd Zamin Jumaat, M. A. Salam, M. S. Islam and R. Hashim, “Utilization
of solid wastes in construction materials” International Journal of the Physical Sciences
Vol. 5(13), pp. 1952-1963, 18 October, 2010.
2. H. Motz, J. Geiseler, “Products of steel slags an opportunity to save natural resources”
Waste Management 21 (2001) 285-293.
3. Ivanka Netinger Grubeša, Ivana Barišić, Aleksandra Fucic and Samitinjay S. Bansode,
“Characteristics and Uses of Steel Slag in Building Construction” Woodhead Publishing,
Elsevier.
4. D.W. Lewis, Properties and uses of iron and steel slags. Presentation at symposium on slag,
South Africa. 1982.
5. A Baseline Study of the Automotive and Construction Sectors on, “Material Consumption
Patterns in India.”
6. George C. Wang, “The Utilization of Slag in Civil Infrastructure Construction” Woodhead
Publishing, Elsevier.
7. Yongjie Xue, Shaopeng Wu, Haobo Hou, Jin Zha, “Experimental investigation of basic
oxygen furnace slag used as aggregate in asphalt mixture” Journal of Hazardous Materials
B138 (2006) 261–268.
8. Sasmita Chand, Biswajit Paul, “An Overview on Steel Plant Waste Management in India”
Strategic Technologies of Complex, Environmental Issues- A Sustainable Approach.
9. George Wang, Yuhong Wang, Zhili Gaob, “Use of steel slag as a granular material:
Volume expansion prediction and usability criteria” Journal of Hazardous Materials 184
(2010) 555–560.
10. Perviz Ahmedzadea, Burak Sengoz, “Evaluation of steel slag coarse aggregate in hot mix
asphalt concrete” Journal of Hazardous Materials 165 (2009) 300–305
11. Hisham Qasrawi, Faisal Shalabi, Ibrahim Asi, “Use of low CaO unprocessed steel slag in
concrete as fine aggregate” Construction and Building Materials 23 (2009) 1118–1125.
12. Hisham Qasrawi, “The use of steel slag aggregate to enhance the mechanical properties of
recycled aggregate concrete and retain the environment” Construction and Building
Materials 54 (2014) 298–304.
13. M. Maslehuddin, Alfarabi M. Sharif, M. Shameem, M. Ibrahim, M.S. Barry, “Comparison
of properties of steel slag and crushed limestone aggregate Concretes” Construction and
Building Materials 17 (2003) 105–112.
14. Gozde Inan Sezder, Mert Gulderen, “Usage of steel slag in concrete as fine and/or coarse
aggregate.” Indian Journal of Engineering & Material Sciences, Vol.22, June 2015, pp 339-
344.
 Investigation of the effect of γ irradiations
on the mechanical and physico-chemical properties
of cementitious materials:
from the nanoscale to mortars’scale

Maxime Robira1,2,3, Benoît Hilloulin2[0000-0002-3668-6724], Ahmed Loukili2, Gildas Potin1,

Xavier Bourbon4, Abdessalam Abdelouas3
1
Tractebel Engineering – Le Delage, 5 rue du 19 Mars 1962 92622 Gennevilliers CEDEX
France
2
Laboratoire GeM – Ecole Centrale de Nantes, 1 rue de la Noë 44321 Nantes
3
Laboratoire SUBATECH – Ecole des Mines de Nantes, CNRS/IN2P3, Université de Nantes 4
rue Alfred Kastler 44307 Nantes cedex 3 France
4
Agence nationale pour la gestion des déchets radioactifs – 1-7, Rue jean-Monnet – 92298
Chatenay-Malabry Cedex

maxime.robira@ec-nantes.fr
benoit.hilloulin@ec-nantes.fr
ahmed.loukili@ec-nantes.fr
gildas.potin@tractebel.engie.com
xavier.bourbon@andra.fr
abdesselam.abdelouas@subatech.in2p3.fr

Abstract. Concrete is commonly used in nuclear construction for decades. Be-

cause low-dose γ irradiations may play an important role regarding the exten-
sion of service life of some structures and long term properties of waste storage
galleries, in-depth studies are necessary. A multiscale experimental campaign
on γ irradiated cementitious materials from mortars to pure phases (C-S-H) has
been carried out and the main results are presented herein. Macroscopic me-
chanical losses were measured after exposure to a total dose around 250 kGy.
They depend on the environmental and curing conditions. Viscoelastic proper-
ties were assessed using a novel method coupling indentation and 3D-imaging.
The structural and mineralogical modifications at microscopic scale due to irra-
diations were determined using a combination of experimental techniques (Ra-
man, TEM, indentation, MIP). It was found that C-S-H undergo nanoscale
structural modifications depending on their Ca/Si ratio (silica chains polymeri-
zation or network disorganization) which are probably related to the observed
macroscopic degradations.

Keywords: Radioactive waste, mechanical properties, creep, calcium silicate

hydrates.
© Springer Nature Singapore Pte Ltd. 2020 639
C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_101
640 M. Robira et al.e

1 Introduction

The project of the Industrial Geological Storage Center or CIGEO [1] carried by the
Andra (National Agency for the management of radioactive waste) has the objective,
if it is authorized, to store the MA / VL (Medium Activity Long Life) and HA (High
Activity) wastes 500 meters deep in a bed of argillite. This involves a network of
concrete galleries which will therefore be subjected to mechanical stresses and irra-
diations effects.
Recently, a compilation of data from the literature demonstrates a modification of
the mechanical behavior of cementitious materials during irradiation [2]. It is im-
portant to note, however, that these studies were conducted under extremely high dose
rates (neutron: 1020 n.cm-2, gamma: 108 Gy) [3]. Few studies deal with the effects of
gamma radiation on cementitious materials and even less with low dose rates while
the structures would be principally exposed to those more penetrating radiations. The-
se studies, however, tend to reveal the same decreases in the macroscopic mechanical
properties [4]. One of the first mechanisms concerning the degradation of concretes
under gamma radiation involves the production of hydrogen following the radiolysis
of the water present in the material motivating research focused on cement pastes and
calcium silicates hydrates (C-S-H) [5-7].
The main objective of this study was to analyze the effects of low dose rate gamma
irradiation on the physico-chemical and mechanical properties of cementitious mate-
rials, from pure hydrate mortars through innovative experimental techniques at differ-
ent scales. The unprecedented multi-scale analysis that was conducted then made it
possible to identify some of the main influencing parameters [8].

2 Materials and Methods

2.1 Samples preparation

A mortar sample formulation with a water-to-cement ratio of 0.43 was determined
using CEM I 52.5 and 0/4 mm calcareous sand to be as representative as possible of a
high-performance concrete used in nuclear waste storage facility with a 28-day com-
pressive strength close to 70 MPa [8]. Three different series of six 4×4×16 cm3 mor-
tar samples were prepared after a 28-day water curing: one of 14-day dried specimens
(‘MD’), one of 14-day sealed cured specimens (‘MH’ standing for ‘humid’) and one
of specimens dried during 14 days and then carbonated during 14 days (‘MC’). Some
cement pastes were prepared similarly.
Besides, C-S-H powders with Ca/Si ratios of 0.8 and 1.2 were synthetized mixing
ultra-pure water, CaO powder (calcined at 1000°C during 24 hours to eliminate all the
carbonates) and SiO2 powder. After mixing, the solution was agitated during one
month in polyethylene bottle in a glove box to prevent air contamination. The solution
was then filtered and the C-S-H gel obtained was dried in a dryer containing silica gel
for one month up to a constant mass. Powder was pelletized for some analysis.
Investigation of the effect of irradiations … 641

2.2 Irradiations conditions

Irradiations were performed in a gamma irradiator gamma with a 137Cs source (Fig.
1) at a dose rate of 8 Gy/min during 3 weeks for mortars and up to 15 days for C-S-H.

Fig. 1. : Photo and schematic representation of gamma irradiator

2.3 Multiscale analysis

Transmission electron microscopy (TEM)

The TEM analysis was conducted with an acceleration tension 300 kV and a point
resolution of 0.18 nm to determine the chemical and structural evolutions.

Raman
Raman spectra of irradiated and pristine C-S-H powders were recorded twice on a
range between 50 and 3550 cm-1 with a spectral resolution estimated at 3 cm-1.

Indentation
Indentation hardness HIT and creep modulus C were determined for the cement
pastes of mortars using a Vickers indenter with a 2000 mN force. A dedicated method
coupling micro-indentation and optical microscopy was developed (Fig. 2) [9].

Mercury Intrusion Porosimetry (MIP)

Mercury porosimetry was used to observe the evolution of porosity ranging from 5
nm to several hundred micrometers due to gamma irradiation.

Fig. 2. : Coupling method between indentation and optical microscopy.

642 M. Robira et al.e

3 Results and discussion

3.1 Results on cement pastes and mortars

The compression and the bending tests show different behaviors depending of the
states of the mortar (Fig. 3). Indeed a significant decrease of the compressive strength
(4% and 9%) and bending strength (15% and 20%) was observed for the humid and
the dried samples, respectively. For the carbonated samples no significant decrease
occurred, highlighting a beneficial role of the calcite precipitation. Mechanical chang-
es are linked to nanoporosity modifications observed using MIP. Indentation tests
confirm these results showing an increase of HIT and the creep modulus C which
means that irradiated cement pastes creep less (Fig. 4).

a) b)
Fig. 3. : Standardized compressive strength a) and bending strength b) and standard deviation
for mortar dried (violet), humid (blue) and carbonated (green), unirradiated (solid) and irradiat-
ed (hatched) in regards with the mean of unirradiated samples

Fig. 4. : Standardized hardness parameters of dried mortars (MD), humid mortars (MH) and
carbonated mortars (MC).
Investigation of the effect of irradiations … 643

3.2 Results on C-S-H

Different chemical analyses were performed on the C-S-H powder after irradiation,
especially Raman analyses showing an evolution of the chain length composing the
layers due to a variation of the quantities of the different silica tetrahedrons (Fig. 5 a).
This modification was confirmed by TEM analyses showing a short range organiza-
tion (Fig. 6) resulting from an increase of the silica chain length and a creation of
cross-linking between layers for the C-S-H 1.2.

a) b)
Fig. 5. : Evolution of the percentage of the 3 species Q1, Q2, Q3 composing a C-S-H with a
Ca/Si ratios of 1.2 (a) and 0.8 (b) in function of the dose received.

On the other hand, the C-S-H 0.8 show an increase of the layers disorder by a deg-
radation of the polymerization resulting in an amorphization of the sample (Fig. 5 b
and Fig. 7). These changes cause macroscopic modifications like a variation of the
porosity that could result in a variation of the mechanicals strength.

Fig. 6. : TEM analyses on a C-S-H 1.2 pristine (left) and irradiated (right)

Fig. 7. : TEM analyses on a C-S-H 0.8 pristine (left) and irradiated (right).
644 M. Robira et al.e

4 Conclusion

Physico-chemical and mechanical analyzes were carried out on mortar samples

(dry, wet and carbonated) as well as on two types of C-S-H after gamma irradiation of
low dose rate. The mechanical analyzes show a behavior of the mortars identical to
that observed in the literature during gamma irradiation with a high dose rate, i.e. a
decrease in compressive and flexural strengths after irradiation. This decrease seems
to be influenced by the water content within the material as well as the presence of
calcite. However, no physicochemical changes could be highlighted.
In order to explain these mechanical variations, a study of different pure phases
such as C-S-H was conducted. Two types of C-S-H with Ca/Si ratios of 0.8 and 1.2
were synthesized, irradiated and analyzed. It has thus been demonstrated a decrease in
hardness in both cases, correlated with a modification of the arrangement and the
length of the chains constituting the material. However, the behavior of a single phase
cannot be attributed to the behavior of this same phase in the presence of others as
shown by the tests on phase mixtures.
The results obtained lead to a better understanding of the phenomena of degrada-
tion of cementitious materials by ionizing radiation. Longer term studies would make
it possible to better dimension sensitive structures and even to estimate the occurrence
of possible disorders throughout their lifetime.

References
1. ANDRA. Projet Cigéo. p. 23 (2017).
2. Field, K.G., Remec, I. and Le Pape, Y.: Radiation effects in concrete for nuclear power
plants - Part I: Quantification of radiation exposure and radiation effects. Nucl. Eng. Des.
(28), 126–143 (2015).
3. Hilsdorf, H., Kropp, J. and Koch, H.: The Effects of Nuclear Radiation on Materials Prop-
erties of Concrete. ACI SP 55, 223–251 (1978).
4. Maruyama, I., Ishikawa, S., Yasukouchi, J., Sawada, S., Kurihara, R., Takizawa, M. and
Kontani, O.: Impact of gamma-ray irradiation on hardened white Portland cement pastes
exposed to atmosphere. Cem. Concr. Res. (108), 59–71 (2018).
5. Bouniol, P. and Bjergbakke, E.: A comprehensive model to describe radiolytic processes
in cement medium. J. Nucl. Mater.(372), 1–15 (2008).
6. Kontani, O., Sawada, S., Takizawa, M. and Sato, O.: Evaluation of irradiation effects on
concrete structure - Gamma-ray irradiation tests on cement paste. Proc. ASME 2013 Pow-
er Conf., pp. 1–8 (2013).
7. Tajuelo Rodriguez, E., Hunnicutt, W. A., Mondal, P., Le Pape, Y.: Examination of gam-
ma‐irradiated calcium silicate hydrates. Part I: Chemical‐structural properties.J. Am. Ce-
ram. Soc., 1–11 (2019).
8. Robira, M., Hilloulin, B., Loukili, A., Potin, G., Bourbon, X. and Abdelouas, A.: Multi-
scale investigation of the effect of c irradiations on the mechanical properties of cementi-
tious materials. Constr. Build. Mater. (186), 484–494 (2018).
9. Hilloulin, B., Robira, M. and Loukili, A.: Coupling statistical indentation and microscopy
to evaluate micromechanical properties of materials: Application to viscoelastic behavior
of irradiated mortars. Cem. Concr. Compos. (94), 153–165 (2018).
 Part IV
Geotechnics for Environment and Energy
 Evaluation of the At-Rest Lateral Earth Pressure
Coefficient of Fibre Reinforced Load Transfer Platform
and Columns Supported Embankments

Cong Chi Dang1 and Liet Chi Dang2

1
Faculty of Civil Engineering, Ho Chi Minh City University of Technology, VNU-HCM, 268
Ly Thuong Kiet Street, District 10, Ho Chi Minh City, Viet Nam
2
School of Civil and Environmental Engineering, University of Technology Sydney (UTS), 15
Broadway, Ultimo, NSW 2007, Australia
1870139@hcmut.edu.vn, dangchiliet@gmail.com

Abstract. The at-rest lateral earth pressure coefficient (K0) is an essential soil
property in design of geotechnical problems, but investigating its influence on
behaviour of embankments supported by load transfer platform and columns im-
proved soft soils has remained very limited. In this study, numerical modelling
of a novel ground improvement technique utilising fibre reinforced load transfer
platform (FRLTP) and columns supported embankment founded on top of mul-
tilayers of soft soils is proposed and investigated by finite element analysis
(FEA). This research aims to assess the influence of a new ground improvement
technique using FRLTP on the embankment behaviour supported by columns in
soft soils. Moreover, a numerical assessment by varying the K0 value of FRLTP
is performed through an extensive parametric study to investigate the K0 influ-
ence on the behaviour of FRLTP and column-supported embankments over soft
soils. Results of the numerical modelling show that the final settlement, the dif-
ference in settlement between columns and foundation soil, the lateral defor-
mation can significantly be reduced by the insertion of FRLTP into a column-
supported embankment system. The predicted results also indicate that the
changes in the K0 value were found to have no notable effects on the embankment
behaviour.

1 Introduction
The at-rest lateral earth pressure coefficient is an important soil property in practical
designs of many geotechnical problems. K0 representing the anisotropic geostatic stress
state is defined as the ratio of the horizontal effective stress to the vertical effective
stress. K0 is actually of essential input parameters for design and assessment of earth
retaining structures, calculations of friction between piles/columns and surrounding
soil, excavation supported systems as well as numerical simulations of boundary effects
of a number of geotechnical problems. Thus, determining a proper value of K0 is par-
ticularly important to accurately generate initial stress conditions in numerical model-
ling of geotechnical problems. Having several experimental and theoretical methods
has been proposed to calculate the K0 value in soils. However, the estimation method

© Springer Nature Singapore Pte Ltd. 2020 647

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_102
648 C. C. Dang and L. C. Dang

defined in the following equation proposed by Jaky [1] has widely been accepted for
normally consolidated soft soils.
K0= 1-sinI' (1)
where I' is effective internal friction angle of shear resistance of soils. Meanwhile, the
lateral earth pressure coefficient (‫ܭ‬଴ை஼ ) at rest for overconsolidated soils is expressed
by a function of OCR (overconsolidation ratio) in Eq. (2) presented by Schmidt [2] that
has commonly been used for estimating the K0 value for overconsolidated soils.
‫ܭ‬଴ை஼ = K0OCRsinI (2)
It can be noted that although K0 value was selected in a wide range of 0.5~1 for simu-
lations of soil-cement columns supported embankments in those previous investiga-
tions [3, 4], the K0 influence of reinforced lime-soil materials to be used as load transfer
platform on the overall behaviour of embankments over soft soils has not been investi-
gated thoroughly. In this study, a numerical assessment is undertaken on the full geom-
etry of FRLTP and column-supported embankment over soft soils to study the effect of
the FRLTP K0 value varying in a range of 0.5~1 on the overall embankment behaviour.
The calculated results of the numerical assessment consisting of final settlement, dif-
ference in settlement between columns and surrounding soil, lateral deformation during
the embankment construction and long-term service are presented and thoroughly com-
pared. The finding of this research is about to provide a better understanding of the
behaviour of columns supported embankment with FRLTP while K0 value of FRLTP
varies. Comprehensive comparisons between column-supported embankment with
FRLTP and without FRLTP are also made to comprehend the benefits of a column-
supported embankment system with FRLTP.
Centreline
10m 10.8m 59.2m

Embankment fill
1
H=6m 1.8
FRLTP 0.5m
0.0
1m GWT -1.0 Surface layer
Column A 3m Soft clay 1 -4.0

8m Soft clay 2

DCM columns I1.2m@1.9m -12.0

Square pattern
3m Stiff clay
-15.0

Sandy clay
15m

-30.0

Fig. 1. A typical profile of embankment supported by FRLTP and DCM columns in soft soils

2 Description of Case Investigation

In this numerical simulation, a 6 m high embankment supported by an FRLTP with 0.5

m height and DCM columns improved layered soft soils is considered (see Fig. 1). As
shown in Fig. 1, only the right half of the embankment domain is presented because of
the embankment symmetry around its centreline. The embankment is constructed on a
1 m thick fill material as a surface layer overlaying an 11 m thick deposit of soft clay.
Evaluation of the At-Rest Lateral Earth Pressure … 649

Stiff and sandy clay layers are assumed below the soft clay deposit. The ground-water
table is located at a depth of 1 m below the ground surface. Details of these soil layers
are summarised in Table 1 and 2. It is noted that a layer of lime-fibre reinforced soil
layer with a thickness of 0.5 m as adopted in the previous study by Dang et al. [5] is
used in this numerical simulation and serves as an FRLTP placed on the top of DCM
columns improved soft soils. DCM columns with 1.2 m diameter and 10 m length are
arranged in a square grid pattern with a centre-to-centre spacing of 1.9 m, which are
used to improve the thick soft soil strata of 12 m. The construction sequence of the
embankment is assumed to be in 0.5-1 m lifts at an average filling rate of 0.06 m/day
[3] to a total height of 6 m including the 0.5 m FRLTP height. The completion of em-
bankment construction is followed by a consolidation period of 2 years.
Table 1. Material properties of subgrade soil layers used in Modified Cam Clay model
Parameters Surface Soft clay 1 Soft clay 2 Stiff clay
Depth (m) 0-1 1-4 4-12 12-15
Material model MCC MCC MCC MCC
Unit weight, J (kN/m3) 16 13.4 14.3 18
Poisson's ratio, Q 0.15 0.15 0.15 0.15
Compression index, O 0.25 0.87 0.43 0.12
Swelling index, N 0.025 0.087 0.043 0.012
Over consolidation ratio, OCR 1.5 2.5 1.2 1.0
Slope of the critical state line, M 1.2 1.2 1.2 1.4
Initial void ratio, e0 1.5 3.1 2.49 0.8
Vertical permeability coeffi- 6×10-4 4.4×10-4 4.6×10-4 2.5×10-3
cient, kv (m/day)
Horizontal permeability coeffi- 9.1×10-4 6.6×10-4 6.9×10-4 2.5×10-3
cient, kh (m/day)
K0 0.6 0.6 0.5 0.5
Material behaviour undrained undrained undrained undrained

3 Numerical Modeling and Soil Parameters

For this study, 2D plane strain modelling was performed using geotechnical software
Plaxis 2D (2017) adopting the equivalent 2D numerical analysis method proposed by
previous researchers [3, 5, 6]. The DCM columns were simulated by continuous plane
strain walls of 0.6 m thickness for the entire columns length of 10 m and the centre-to-
centre spacing between two adjacent walls was remained the same as the 1.9 m centre-
to-centre spacing between two adjacent DCM columns; meanwhile, the equivalent nor-
mal stiffness (EA) was taken into account. For this 2D simulation, a half-fine mesh was
used because of geometrical symmetry of the embankment system. The foundation soil
was taken to 30 m depth from the ground surface to sandy clay stratum. Meanwhile,
the horizontal length of the FEA model was taken to be 80 m, which was almost three
times the half width of the embankment base in order to eliminate the boundary influ-
ence. In addition, the DCM columns, FRLTP, embankment and fill materials were sim-
ulated as a linear elastic-perfectly plastic material using Mohr-Coulomb (MC) model
650 C. C. Dang and L. C. Dang

[5]. Whereas, the surface, soft soil and stiff clay layers were assumed to follow Modi-
fied Cam Clay (MCC) model. A summary of the constitutive model parameters is pre-
sented in Table 1 and 2. The vertical and horizontal boundary conditions are described
in further detail in Dang et al. [5].

Table 2. Material properties adopted in Mohr-Coulomb model

Embankment DCM
Parameters Sandy clay FRLTP
fill columns
Depth (m) 15-30 - - -
Material model MC MC MC MC
Unit weight, J (kN/m3) 19 12.5 19 15
Young's modulus, E (MPa) 20 125.8 3 100
Poisson's ratio, Q 0.10 0.32 0.40 0.15
Effective cohesion, cc (kPa) 20 75 20 cu=500
Effective friction angle, Mc (q) 35 42 35 -
Initial void ratio, e0 0.7 - - -
Vertical permeability coefficient, 2.5×10-2 - - 4.6×10-4
kv (m/day)
Horizontal permeability coeffi- 2.5×10-2 - - 4.6×10-4
cient, kh (m/day)
K0 0.5 0.5 0.5 0.5
Material behaviour drained undrained undrained undrained

4 Analysis Results and Discussion

4.1 Effect of K0 value of FRLTP on the total and differential settlements

Fig. 2a shows the effect of K0 value of FRLTP ranging from 0.5 to 1 on the total settle-
ment at the centre of the embankment base during construction in stages. It should be
noted that a higher K0 value of FRLTP as compared with that for a reference case pre-
sented in Table 2 was selected for this simulation. This is because previous investiga-
tions by researchers [7-10] reported that chemically treated soils using cement, lime
without or with fibre reinforcement were found to have higher effective yield stress
(preconsolidation pressure) due to cementation bond when compared to untreated soils.
Since OCR for a soil defined in Eq. (2) is a relationship between preconsolidation pres-
sure and present effective vertical pressure, the K0 value has an increasing tendency as
OCR increases. As Fig. 2a illustrates, the changes in K0 value are observed to have a
negligible effect on the total settlement when the embankment height increases from 0
m to 6 m. This is because a similar pattern and almost a same value of total settlement
can be observed in Fig. 2a as the height of the embankment fill increases in the inves-
tigated range. Moreover, observation of the predicted results of the total settlement
notes that the column-supported embankment with FRLTP was found to have about
20% reduction of the total settlement as compared with that without FRLTP reinforce-
ment. This finding indicates that although the vertical settlement on the ground surface
of the embankment with FRLTP was unnoticeably influenced by the changes in K0
value of FRLTP, introducing FRLTP into the embankment compared with the embank-
ment without FRLTP was predicted to have an effective influence on the improved total
settlement. Additionally, the effect of K0 value of FRLTP on the differential settlement
Evaluation of the At-Rest Lateral Earth Pressure … 651

on the ground surface under the embankment base centre is presented in Fig. 2b. It
should be noted that the differential settlement is defined as the maximum settlement
difference between the DCM column top and its surrounding soil at the embankment
base centre. As plotted in Fig. 2b, when the K0 value of FRLTP increased in a range of
0.5~1, no significant difference in the settlement between the column and the founda-
tion soil was observed for the embankment with FRLTP. However, as Fig. 2b shows,
the differential settlement of the embankment reinforced with an FRLTP was approxi-
mately 75% less than the corresponding value for the embankment without FRLTP re-
inforcement when the embankment fill increased from 0.5 m to its final height of 6 m.
The significant reduction of the differential settlement observed for the embankment
with FRLTP indicates that the FRLTP introduction could be an effective solution to
minimising the embankment settlement difference. From the calculated result presented
in Fig. 2b, it is possible to conclude that the effect of K0 value of FRLTP in the exam-
ined range on the embankment differential settlement can be insignificant.

(a) (b)
Fig. 2 Variation of (a) the total settlement and (b) the differential settlement on the ground surface
under the embankment base centre
4.2 Effect of K0 value of FRLTP on the lateral displacement
Fig. 3 presents the 2 years post-construction lateral settlement along a DCM column
under the embankment toe with various K0 values of FRLTP ranging from 0.5 to 1. It
is observed that similar to the trend of the total and differential settlements plotted in
Fig. 2, the lateral displacement of the embankment with FRLTP was calculated to be
seemly independent of an increase in the K0 value of FRLTP in a range of 0.5~1. Nev-
ertheless, it is important to note that when compared with the embankment without
FRLTP, the lateral displacement of the embankment with FRLTP was found to result
in a considerable reduction by approximately 45% as observed in Fig. 3 for the top of
the DCM column. Moreover, the introduction of FRLTP into the column-supported
embankment was predicted to have a strong effect on the reduced lateral displacement
of the embankment to a shallow depth. For example, the difference in the lateral defor-
mation from the embankment with FRLTP to the embankment without FRLTP were
observed to be significant for the column top, but it became smaller as the column
length increased to about 5 m depth below the ground surface. In view of the authors,
652 C. C. Dang and L. C. Dang

the significant improvement in the total and differential settlements, and the lateral dis-
placement of the column-supported embankment with FRLTP could be attributed to
the enhanced engineering properties (e.g., the high shear strength, tensile strength and
stiffness) of fibre reinforced materials, used as a load transfer platform.

5 Conclusions

This paper presents numerically investigating the de-

formation behaviour of an embankment built on
FRLTP and floating DCM columns improved soft
soils. The findings of the 2D finite element analysis
indicate that the at-rest lateral earth pressure coeffi-
cient (K0) of FRLTP changing in a range of 0.5~1 was
numerically predicted to have insignificant effects on
the embankment behaviour in terms of the total and
differential settlements during the construction in
stages, and the 2 years post-construction lateral dis-
placement with depth. However, it is of interest to note Fig. 3 Lateral settlement with vari-
that by comparing with the column-supported embank- ous K0 values of FRLTP under the
ment without FRLTP, the column-supported embank- embankment toe
ment reinforced with a 0.5 m thick FRLTP was found to significantly decrease the total
and differential settlement by 20% and 75%, and cut down 45% of the lateral displace-
ment, respectively. The findings of this research work show that the FRLTP and DCM
columns supported embankment placed on soft soils could be an effective and practical
solution to ensure the embankment stability during the staged construction and services.

References
1. Jaky, J. Pressure in silos. in Proceedings of the 2nd International Conference on SMGE. 1948.
2. Schmidt, B., Earth pressures at rest related to stress history. Can. Geot. J., 1966. 3(4): p. 239-242.
3. Chai, J.-C., Shrestha, S., Hino, T., Ding, W.-Q., Kamo, Y., and Carter, J., 2D and 3D analyses of an
embankment on clay improved by soil–cement columns. Com. Geot., 2015. 68: p. 28-37.
4. Venda Oliveira, P.J., Pinheiro, J.L.P., and Correia, A.A.S., Numerical analysis of an embankment built
on soft soil reinforced with deep mixing columns: Parametric study. Com. Geot., 2011. 38(4): p. 566-576.
5. Dang, L.C., Dang, C.C., and Khabbaz, H., Modelling of columns and fibre-reinforced load-transfer
platform-supported embankments. Ground Improvement, 2019. 0(0): p. 1-19.
6. Dang, L.C., Dang, C.C., and Khabbaz, H., Numerical analysis on the performance of fibre reinforced
load transfer platform and deep mixing columns supported embankments, in Ground Improvement and
Earth Structures, M. Bouassida and M.A. Meguid, Editors. 2018, Springer: Switzerland. p. 157-169.
7. Dang, L.C. and Khabbaz, H., Experimental investigation on the compaction and compressible properties
of expansive soil reinforced with bagasse fibre and lime, in Recent Advancements on Expansive Soils, J.S.
McCartney and L.R. Hoyos, Editors. 2019, Springer: Switzerland. p. 64-78.
8. Dang, L.C., Khabbaz, H., and Fatahi, B. Evaluation of swelling behaviour and soil water characteristic
curve of bagasse fibre and lime stabilised expansive soil. in PanAm-UNSAT 2017. Texas, USA.
9. Dang, L.C., Khabbaz, H., and Fatahi, B. An experimental study on engineering behaviour of lime and
bagasse fibre reinforced expansive soils. in 19th ICSMGE. 2017. Seoul, Korea.
10. Dang, L.C. and Khabbaz, H. Shear strength behaviour of bagasse fibre reinforced expansive soil. in
International Conference on Geotechnical and Earthquake Engineering 2018. Chongqing, China.
 Nonlinear Behavior Analysis of SFRC Foundation
Considering Homogeneous and Inhomogeneous
Soil Interactions

Tran-Trung Nguyen1,2,*, Radim Cajka2, Phu-Cuong Nguyen3,**, Thanh-Tuan Tran4

1
Faculty of Architechture, Van Lang University, Ho Chi Minh City, Vietnam
2
Technical University of Ostrava, Ostrava, Moravian-Silesian Region, Czech Republic
3
Faculty of Civil Engineering, Ho Chi Minh City Open University, HCMC, Vietnam
4
Department of Civil Engineering, Kunsan National University, Republic of Korea
*
nguyentrantrung@vanlanguni.edu.vn ; nguyetrantrungvlu@gmail.com
**
cuong.pn@ou.edu.vn ; henycuong@gmail.com

Abstract. Current structural design and analysis commonly use numerical sim-
ulation. Interactive problems often use this method to express the behavior of
different types of structures when working together subject to external loads.
This study shows the relationship between compression forces and displace-
ments of a slab foundation that are directly interactive with inhomogeneous soil
when using the stress-strain curve of steel fiber reinforced concrete material
proposed by Barros and Figueiras [1]. The obtained results are better than the
previous study RI 9DVNRYD DQG ýDMND [3] when both experimental test results
are compared. Otherwise, only using one type of element for both the founda-
tion and the soil does not affect this interaction problem. The material model is
an essential part of the simulation, and the problem should consider nonlinear
behavior analysis.

Keywords: ANSYS, Finite element analysis, Homogeneous soil, Inhomogene-

ous soil, Steel fiber reinforced concrete, Nonlinear analysis.

1 Introduction

Nowadays, new materials are being used widely for foundation structures. Steel fiber
reinforced concrete (SFRC) is a composite material which has fibers dispersed
randomly in concrete. There are three types of fibers: plain, straight, and round.
Round fibers are the most common type with their diameter ranging from 0.25 to 0.75
mm. The use of such material has been suggested for many structural applications,
with or without traditional internal reinforcement. The use of SFRC is, thus,
particularly suitable for slab foundation structures when they are subjected to
compression loads over the serviceability limit state regarding bending and shear, and
researched characteristics in the study [1].
The substructure and subsoil are interdependent. Therefore, the effect of the under-
lying soil loads on the upper structure cannot be neglected. In the analysis, it is neces-

© Springer Nature Singapore Pte Ltd. 2020 653

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_103
654 T.-T. Nguyen et al.

sary to take into account the interaction of the substructure and subsoil. Several theo-
ries of the origin of the soil models in the paper [2] are proposed. In which, half-
elastic space theory for inhomogeneous soil model is still valid. Inhomogeneous soil
model is a multi-layer model in which each layer has the same thickness, but the
Young modulus is different. This parameter increases with the depth of each layer.
The interaction between the substructure and subsoil was analyzed by authors [3-8]
using the finite element method (FEM). Previously mentioned studies [3-8] use dif-
ferent finite element software to solve the problem that is being discussed by compar-
ing results from the deformation of the slab, subsidence of the subsoil, bending mo-
ments and contact stress.
The feature in the simulation model of these studies was ANSYS parametric design
language (APDL) software for solving the problems. SOLID65 element with two
specific parameters for elastic analysis such as elastic modulus and Poisson¶V ratio
was used for the slab foundation, while nonlinear analysis used all of the nine con-
stants. However, how to determine those constants were not shown in those studies.
Besides, the subsoil model in the above studies used an inhomogeneous soil model
with two parameters, elastic modulus and Poisson's ratio in the elastic analysis, while
nonlinear analysis used Mohr-coulomb criterion. However, there was no explicit men-
tion of two essential values: cohesion c and internal friction angle I p . Besides, the
element used for the inhomogeneous soil was SOLID45. Perhaps some issues are not
considered, so when compared with the practical experimental test also in the study
[3], the results of the deformation of the slab were not as good as expected.
As mentioned above, the SFRC material was assigned through 11 parameters to the
SOLID65 element. This study introduces the stress-strain relationship curve of SFRC
with two parameters that are easily defined [1], namely mean compressive strength
f cf and parameters of geometry L f , D f of fibers in [3]. The authors report the in-
homogeneous soil model as additional parameters in the Mohr-coulomb criteria such
as cohesion c and friction internal angle I p . On the other hand, the model of slab
foundation and inhomogeneous soil uses only one element SOLID186, and the prob-
lem is analyzed nonlinearly with ANSYS Workbench software. The results of the
load-displacement relationship of the slab foundation and compression load of the
proposed model achieved better results than those of 9DVNRYD DQG ýDMND [3] com-
pared with actual experimental results.

2 Experimental test

This study used the results of an experimental test at the Technical University of
Ostrava. The results of compression loads and displacement of slab foundation in the
studies [2; 4] were compared to the experimental results. The displacement of the
interaction between the slab foundation and the inhomogeneous soil at the center of
the slab foundation at each level of compression loads of 75 kN, 150 kN, 300 kN, and
450 kN is shown in Fig. 1.
Nonlinear Behavior Analysis of SFRC Foundation … 655

Fig. 1. Displacement at the central position of the SFRC foundation

in the study [3] transferred.

3 Material models

3.1 Barros and Figueiras 's SFRC material model [1]

The stress-strain relation curve was proposed by Barros and Figueiras (1999) [1]. The
curve is established through Eq.(1) to Eq.(4) as follows:
H
H pf
V f cf (1 q )/ p
(1)
§ H · § H ·
(1  p  q)  q ¨  p¨
¨ H pf ¸¸ ¨ H pf ¸¸
© ¹ © ¹
with
E pf
q 1 p  (2)
Ec
1 q
p  q  [0,1] and !0 (3)
p
fcf fcf
Epf and Ec 21500 3 (4)
H pf 10
where f cf - the average compressive strength. Use the least squares method for 3D
DRAMIX 65/60B6-25 kg.m-3 steel hooked-end fibers with L f 60 mm, D f 0.9 mm,
Lf / Df 65 .
H pf H co  0.00022Wf (5)
0.311W f
p 1  0.853e (6)
where H co = 0.0022 is the value of the deformation at the stress peak of the concrete
without steel and W f is the percentage of steel fiber weight present in the concrete
mix. In this study, the proposed value W f 0.83% concrete C25/30 has compressive
656 T.-T. Nguyen et al.

strength f cf = 33 MPa and elastic modulus Ecm = 31476 MPa [3].

3.2 Soil Model

Elastic Analysis. The soil was analyzed using two models that are homogeneous and
inhomogeneous. These two models behaved according to the theory of elastic half-
space. The similar soil model used two basic physical properties: Elastic modulus
Edef and Poisson¶V ratio P obtained from data in the study [3]. The inhomogeneous
soil model, the Edef parameter was calculated according to studies [3; 8]:
m
§ z · 1
Edef E0 ¨ ¸ where m 2 (7)
© z0 ¹ P
where E 0 is elastic modulus at the surface, z are coordinated in depth, z 0 is a unit
length, sometimes replaced by the width B or the diameter D of the foundation, m is
the coefficient depending on Poisson¶V ratio P , Edef - elastic modulus of layers with
a thickness of 0.25 m calculated according to Eq.(7).

Inelastic Analysis. The mechanical properties of the soil used more strength parame-
ters, such as cohesion c and internal friction angle I p . The type of soil mentioned in
this study is low-plasticity clay type - the Mohr-Coulomb model with parameters
cohesion c 6 kPa and peak friction angle I p 24 degrees was used in this study.

4 Numerical analysis

Fig. 2. Modeling the interaction between the slab foundation and the soil.

The interaction between the foundation slab and the soil was simulated by using
ANSYS Workbench version 19.0. The slab foundation with dimensions
2.00u 2.00u 0.15 m [3], using SOLID186 element with SFRC material from Eq.(1)
to Eq.(6). The SOLID186 element models the soil. Geometric size of
6.00u 6.00u 6.00 m [3] with two types of soil, considers adding coefficients c và I p
for nonlinear problems. The quarter model was simulated for analysis to save the
computational time, as shown in Fig. 2. Constraint conditions include boundary con-
Nonlinear Behavior Analysis of SFRC Foundation … 657

ditions, symmetry conditions, and contact conditions between the slab foundation and
the soil or each layer of soil. All conditions are shown in Fig. 3.

Fig. 3. Constraint conditions for interaction problem.

As analytical results, the influence of a material model to the interaction problem was
studied. Fig. 4 shows the behavior of the slab foundation, considering the interaction
with the inhomogeneous soil. The behavior of the slab foundation and the inhomoge-
neous soil was expressed through the relationship curve between the displacement of
the slab foundation and the compressive load. The proposed result was nearly identi-
cal to the experimental test. It can be seen that the proposed model is better than the
model of study [3].

Fig. 4. The results of the displacement of the slab foundation.

658 T.-T. Nguyen et al.

5 Conclusion

This paper develops the FEM modeling successfully for capturing the behavior of the
slab foundation when considering the interaction of foundation and inhomogeneous
soil. The material model plays an essential role in predicting the behavior of the
whole structure. The model has a few parameters as possible but must fully
demonstrate the mechanical properties of the material. Specifically, the SFRC
material curve model is established from three parameters interacting with
inhomogeneous soil. Results obtained from the proposed model are better than the
previous study [3] when compared with actual experiments. This study also shows
that using one type of element for both types of structures not only shows good results
but also saves the computational time for solving nonlinear problems.

References
1. Barros, J. A., Figueiras, J. A.: Flexural behavior of SFRC: testing and modeling. Journal of
Materials in Civil Engineering, 11(4), 331-339 (1999).
2. Feda, J.: Stress in subsoil and methods of final settlement calculation. Elsevier (2013).
3. 9DVNRYD - ýDMND 5 ' 1XPHULFDO 0RGHO RI WKH 6XEVoil-Structure Interaction and
Comparison of the Results with Experimentally Measured Values. In: the Key
Engineering Materials, 298-309. Trans Tech Publ, Switzerland (2017).
4. ýDMND 5 .ĜLYê 9 6HNDQLQD ' 'HVLJQ DQG GHYHORSPHQW RI D WHVWLQJ GHYLFe for
H[SHULPHQWDOPHDVXUHPHQWVRIIRXQGDWLRQVODEVRQWKHVXEVRLO7UDQVDFWLRQVRIWKH9â%±
Technical University of Ostrava, Civil Engineering Series, 11(1), 1-5 (2011).
5. Cajka, R., Burkovic, K., Buchta, V.: Foundation slab in interaction with subsoil. In: the
Advanced Materials Research, 375-380. Trans Tech Publ, Switzerland (2014).
6. ýDMND5/DEXGNRYi-0\QDUþtN31XPHULFDOVROXWLRQRIVRLO-foundation interaction
and comparison of results with experimental measurements. International Journal of
GEOMATE, 11(23), 2116-2122 (2016).
7. /DEXGNRYi-ýDMND5'QXPHULFDOPRGHOLQ the nonlinear analysis of the interaction
between subsoil and SFCR slab. International Journal of GEOMATE, 13(35), 120-127
(2017).
8. 9DãNRYi - ýDMND 5 ,QWHUDFWion of nonlinear numerical model of SFRC slab and
nonlinear numerical subsoil model. International Journal of GEOMATE, 15(47), 103-110
(2018).
9. Bencardino, F., Rizzuti, L., Spadea, G., Swamy, R. N.: Stress-strain behavior of steel fiber-
reinforced concrete in compression. Journal of Materials in Civil Engineering, 20(3), 255-
263 (2008).
 Analyses on drainage capacity and sliding resistance of
large diameter vertical wells for deep-seated landslide
stabilization

Ba Thao Vu1,3, Van Minh Pham1(*), Quoc Dung Nguyen1,2,

Huy Vuong Nguyen1, Thi Thanh Huong Ngo4
1
Department of Geotechnical Engineering, Hydraulic Construction Institute, Vietnam Academy
for Water Resources, 3/95 Chua Boc Street, Ha Noi, Viet Nam.
2
Vietnam National Committee on Large Dams and Water Resources Development - VNCOLD,
10/95 Chua Boc Street, Ha Noi, Viet Nam.
3
Department of Geotechnical Engineering, College of Civil Engineering, Tongji University,
Shanghai 200092, China.
4
Faculty of Civil Engineering, University of Transport Technology, 54 Trieu Khuc, Thanh
Xuan, Ha Noi, Viet Nam.
(*) Corresponding author: vanminhvtc@gmail.com

Abstract. Deep-seated landslides occurring near roads and building areas cause
serious damages. High groundwater level due to long periods of rainfall or infil-
tration from reservoir saturates and softens soils landslides. This article intro-
duces an approach which combines the restraining effects of stabilizing piles and
the dewatering effect from large diameter vertical wells, to increase the stability
of landslides. Effectiveness in resisting sliding force and lowering the groundwa-
ter level was evaluated using the finite-elements method. The results show that
the saturation line drops significantly, and the safety factor of the slope increases
by approximately 20% compared to the case without the wells.

Keywords: Deep-seated landslide, Slope stabilization, Slope drainage, Large

diameter vertical well.

1 Introduction

Common stabilization methods, which are often used for landslides prevention, include,
retaining wall, slope alteration in combination with water drainage channel, integration
between soil anchors and surface cover of the slope and soil nails [1, 2, 3]. For the deep-
seated landslides with high level of groundwater in rainy seasons, these methods face
various limitations, such as, (1) the groundwater level increases due to rainfall infiltra-
tion, low capacity of drainage system, stuck or even lack of drainage system; (2) the
combination of drains and surface cover limits the volume of surface water that pene-
trates onto the slope, thus lowering the existing groundwater in the slope; (3) using only
horizontal drains is not effective because the length and the diameter of horizontal
drains are limited by drilling technique. In that case, reducing the sliding forces by

© Springer Nature Singapore Pte Ltd. 2020 659

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_104
660 B. T. Vu et al.

drainage is more important than increasing the resisting forces by using structural
countermeasures, such as retaining walls, soil anchors or soil nails. Therefore, it is es-
sential to find the method that integrates reducing the driving forces by subsurface
drainage system, as well as increasing the resisting forces by large diameter stabilizing
piles [4, 5, 6], for better results.
In this article, a deep-seated landslide stabilization method using large diameter vertical
(LDV) wells is introduced. Effectiveness of the method in both lowering groundwater
levels and increasing the sliding resistance is analyzed through a three dimensional fi-
nite element method (FEM).

2 Stabilization principles of LDV wells

Long-term heavy rainfall can cause saturation of the soil, rising groundwater level, re-
duced shear resistance and increased self-weight of soil on the slope, resulting in in-
creased sliding force that finally culminates in landslides. Therefore, lowering the
groundwater level and preventing water infiltration are very important for slope stabi-
lization. For deep-seated landslides (sliding plane depth ranges from 5.0 m to 40.0 m)
with high groundwater level, the conventional stabilization methods cannot decrease
the sliding forces. This study proposes a new contermeasure using large diameter well
drainage (1.5 m to 6.0 m) to stabilize deep-seated landslides (Fig. 1).
1 ± natural slope; 2 ±
normal water table; 3 ±
potential sliding sur-
face; 4 ± RC well; 5 ±
filter mesh hole; 6 ±
horizontal drain; 7 ±
backup pump; 8 ± sur-
face ditch; 9 ± ground-
water level when wells
are operating; 10 ± rain-
fall.

Fig. 1. Structure of LDV well

Water collected from the holes on the well shaft is evacuated through the horizontal
drains to the surface channels. The filter mesh is inserted into the holes to prevent soil
to enter. Horizontal drains may be positioned in many levels to augment the drainage
capacity. Due to the large diameter well, many holes can be created on the well shaft,
making it possible to collect a large volume of water. To prevent choking of horizontal
drains or the huge volume of water due to prolonged rain, backup pumps were needed
to be installed. The groundwater level could be determined based on the number of
wells per area unit of slope and the permeability of soils.
In addition to the function of lowering the groundwater, LDV wells act as stabilizing
piles. The large diameter wells could prevent large seated landslide phenomenon be-
cause of the large shear resistance of reinforced concrete shaft. Besides, using wells
minimizes deforestation, protects the environmental landscape and controls surface
Analyses on drainage capacity and sliding resistance … 661

run-off.

3 Application of LDV wells for a deep-seated landslide

3.1 Introduction of deep-seated landslide

The investigated slope is in Hoa Binh province. After prolonged heavy rains at the be-
ginning of October 2017, 18 cracks appeared on the surface of slope with the width
ranges from 2 cm to 15 cm and the length varies from 10 m to 90 m, forming a potential
sliding surface of 300 m long, 200 m wide and around 30 m deep. The sliding block of
approximately 1.8 million m3 soil is shifted down from 5 cm to 80 cm. The sliding
block has inclined the 110 kV electric tower and the retaining wall at the bottom of the
slope. It also causes the cracking on the nearby under constructed buildings (Fig. 2).

Fig. 2. Current state of the landslide

Fig. 3. Cross-section of slope

Since the slope has a sliding arch of great length, using 04 wells located respectively
at level +17.0 (02 wells of 20 m deep); level +31.0 (02 wells of 16 m deep), whereas
662 B. T. Vu et al.

the level of the foot of the slope is +0.00, the top level of the slope is +42.0.
The wells have a diameter D1 = 3.0 m; the well shaft is composed of two layers: the
first layer, which has a thickness of 15 cm, is for protection while digging the well; the
second layer, which has a thickness of 30 cm, is the main structure. The dimensions of
holes are 1m x 1.2 m. Horizontal draining steel pipes, with a diameter of 150 mm, and
thickness of 10 mm were arranged. The reverse filter system of 60 cm thickness was
installed at the bottom of the well (Fig. 3)
Mechanical properties of soil layers were obtained by laboratory tests and the me-
chanical properties of concrete referenced from the TCVN 5574:2012³Concrete and
reinforced concrete structures ± 'HVLJQVWDQGDUG´>@DUH shown in Table 1.
Table 1. Physical and mechanical properties of soils and concrete
Jnatural Jsaturated E K c M
Layer Q
(kN/m3) (kN/m3) (MPa) (m/s) (kPa) (°)
1st layer 18.5 19.5 0.3 60 1.50×10-6 20 15
2nd layer 18.9 19.9 0.3 50 1.85×10-7 15 18
3rd layer 20.2 21.2 0.28 110 5.04×10-7 35 35
4th layer 21.0 22.0 0.28 180 1.30×10-8 60 35
Concrete 25.0 25.0 0.20 25000 1.00×10-12 1000 40

3.2 Landslide stability analysis

In this study, we have used geo-technical engineering software Midas GTS for seepage
and stability analysis. Midas GTS is a modern FEM-based geotechnical and tunneling
software developed by Midas IT since 1989 [8]. We used 2D models for slope stability
analysis with the calculation procedure as following: (1) seepage analysis to determine
saturation line and (2) slope stability analysis using results received from previous prob-
lem. On the other hand, a 3D model was adopted for seepage analysis to evaluate the
effectiveness of lowering groundwater.
Seepage analysis: The seepage analysis model developed in Midas GTS to determine
the saturation line and the water pressure applying on the wells shaft is based on the
partial differential equations for steady-state flow as following:
w § wH · w § wH · w § wH · w4
¨ kx ¸  ¨ ky ¸  ¨ kz ¸Q (1)
wx © wx ¹ wy © wy ¹ wz © wz ¹ wt
in which: H: pressure head (m); kx, ky, kz: coefficients of permeability in x-, y- and
z-directions respectively (m/s); Q: inflow/outflow quantity in volume units per unit
time, (m3/s); 4: volumetric water content; t: time (s).
Stability analysis: The selected method for slope stability analysis is Strength Re-
duction Method, which gradually reduces the shear strength ( M, c) of slope ground ma-
terials until failure surface takes place. To determine the minimum safety factor for a
slope, the modulus of Elasticity aQG3RLVVRQ¶VUDWLRDUHDVVXPHGWREHFRQVWDQW7KH
cohesion (c) and friction angle (M) are progressively reduced, and the safety factor is
determined at the diverging point (Eq. (2)).
W
Fs (2)
Wf
in which: W: shear strength of slope material, W c  V n tan I (kPa); Wf: shear stress at
Analyses on drainage capacity and sliding resistance … 663

the sliding surface is expressed as W f c f  V n tan I f (kPa).

3.3 Results Analysis

Results of 2D and 3D calculation models are shown in figures 4 and 5, respectively.

The groundwater level is lowered significantly, at the section 2-2 where there is no
well, the saturation line is at 12.3 m away from the ground; and at section 3-3 where
well exists, the saturation line is at 12.8 m away from the ground. Overall, it can be
seen that the saturation line is dropped down around 5 m, thereby indicating that the
proposed solution is effective.

Fig. 4. Saturation line in case without Fig. 5. Saturation line in case using
wells wells

Fig. 6. Effectiveness in lowering the saturation line at the cross-sections.

Fig. 7. Safety factor in case without wells Fig. 8. Safety factor in case using wells
The slope stability analysis results are presented in Fig. 7 and Fig. 8. It can be seen that
664 B. T. Vu et al.

slope stabilization method using the LDV wells increases the safety factor from 1.01 to
1.23, and the sliding surface does not form continuously.

4 Conclusion

This paper presents a novel method to improve the stability of deep-seated landslide by
larger diameter vertical wells. The well is easily constructed, and has no effect on
nearby landscapes. This countermeasure is suitable to stabilize the large deep-seated
landslides. The larger diameter vertical well was effectives in lowering the groundwater
level and increase the resisting sliding forces. Consequently, it leads to a drop in the
phreatic line significantly, and the safety factor of the slope increases approximately by
20% compared to the case without the wells.

Acknowledgements

This study was supported by the Independent National Research Project "Research and
Propose Models of Integrated Solutions for Collection, Storage and Exploitation of Wa-
ter Sources for Drinking and Irrigation Water in Water Scarcity Regions in Dien Bien
Province", Ministry of Science and Technology of Vietnam.

References
1. R. Whitlow. Soil mechanics, Education publisher (1999).
2. Braja M.Das. Foundation Engineering, China machine press (2016).
3. B.K. Nguyen. Design and construction of deep excavation, Construction publisher (2010).
4. Richard L. Bateman, Francke C. Walberg. Pile Design Procedure for Stabilizing Channel
Slopes, Int. Con. on Case Histories in Geotechnical Engineering, 483-490 (1998).
5. D. Mujah, F. Ahmad, H. Hazarika, N. Watanabe. The Design Method of Slope Stabilizing
Piles: A Review, Int. Journal of Current Eng. and Tech., 224-229 (2013).
6. DZ/T 0219-2006. Specification of design and construction for landslide stabilization, Min-
LVWU\RI/DQGDQG5HVRXUFHVRIWKH3HRSOH¶VRepublic of China (2006).
7. 7&91³&RQFUHWHDQGUHLQIRUFHGFRQFUHWHVWUXFWXUHV± Design standard.
8. Midas Geotechnical and Tunnel Analysis System. MIDAS Info. Tech. Co., Ltd., (2014).
Shape effects in the dynamical response of flints/boulders
during pile driving ± a numerical study.

Emilio Nicolini 1 and Ivan Terribile2

1
Cathie Associates France, 92000 Nanterre, France
emilio.nicolini@cathie-associates.com
2
Cathie Associates France, 92000 Nanterre, France
ivan.terribile@cathie-associates.com

Abstract. Today, more than eighty percent of the total power of the active off-
shore wind turbines is installed in Europe, where the offshore wind industry has
gained full reliability and is now targeting to continue reducing the construction
cost. About 20 to 30% of the cost of an installed turbine is the foundation,
where installation plays a major role and thus cost reduction will be mostly im-
pacted by developments . This paper presents the following of a study of the
dynamic reaction offered by a flint to the pile tip, which was presented in the
2018 Hanoi Conference, taking into consideration a circular and elliptical shape
and the possibility of a non-centered impact of the pile wall on the boulder/flint.
The work is aimed at the exploration of the effect of the shape of the boul-
der/flint and concludes with a simple prediction method to reduce the risk of
pile tip damage.

Keywords: Pile driving, Boulder, Flint, Chalk, Dynamic numerical study.

1 Introduction

In the last decade, the development of the offshore wind-energy industry led to the
installation of large monopiles in the North Sea. This area is characterized by the
presence of chalk deposits. Being the foundation a significant part of the construction
and installation cost (about 30%), the cost reduction of wind turbine fields will in-
volve the optimization of the installation of the foundation, which today is largely
constituted by large diameter driven piles. The diameter is often larger than 6m, with
8 and 10m diameter piles not rare. In such a trend, with the need of optimizing the
steel volume (i.e. reducing the steel pile wall thickness), the diameter to thickness
ratio is increased considerably. This exposes the pile tip to be damaged by the impact
with boulders which can be found in glacial tills; such event is becoming more and
more a concern in the industry, due to the consequences of pile tip damage on the
monopile installation. The same concern is anticipated for the flint bands which are
frequently found in chalk, where driven monopiles are in some cases foreseen to be
installed with the so called drive-drill procedure, which requires the pile tip to be
undamaged during the driving phase. In this context, limited publication is available
and the Authors are carrying out a research program. Chalk is an ultra-fine-grained

© Springer Nature Singapore Pte Ltd. 2020 665

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_105
666 E. Nicolini and I. Terribile

limestone made of very pure, low magnesium carbonate, retaining high porosity and
moisture content [1]. In chalk, bands of nodular flints are found widely. Flints are
siliceous rocks, with a very high resistance. During pile driving in chalk, when a flint
is found, two phenomena can occur: the flint breakage or its plastic displacement. Due
to the low strength of the chalk and the high tensile strength of the flint, the latter is
mostly displaced into the chalk [2]. The force to displace the flint arises at the
flint-pile contact and its magnitude and direction are dependent on the flint size,
shape, on the chalk strength/stiffness and on the position of the impact point. In this
configuration, the formation of an initial imperfection at the pile tip may lead to a
progressive wall yielding during driving, leading to full pile failure. To evaluate the
magnitude of this force during pile driving, parametric dynamic FEA on the pile-flint-
soil interaction were done. The paper focuses on the flint shape effect, with the aim of
improving the results presented in Nicolini&Gargarella (2018) and generalizing the
safe driveability assessment criteria to different flint geometries.

2 Dynamic Finite Element model

Dynamic FE analysis have been performed with one circular and two different ellip-
soidal flint shapes. To study the effect of the flint shape on the flint/pile-wall interac-
tion, the flint width was kept constant and equal to the circular flint diameter d, while
3 different sizes of the vertical axis h have been considered, with a ratio h/d equal to
1, ½ and ¼, (Fig. 1). The size effect was evaluated by using two different values for
the width d, for a total number of six different geometries. The flint behavior was
modelled with a linear elastic model (E=80GPa, Ȟ=0.145), while for the chalk an elas-
tic-perfectly plastic behavior was chosen, by using the Undrained C feature in
PLAXIS. For comparison purposes, the numerical model contains characteristics
similar to the model previously adopted in Nicolini&Gargarella (2018) [2]. Three
different chalk properties have been used, whose values are reported in Table 1.

Table 1. Chalk parameters.

Ȗ Su Eu νu Vp
Material
[kN/m3] [kPa] [GPa] [-] [m/s]
Chalk 1 19.0 100 0.1 0.496 1475
Chalk 2 19.0 500 0.5 0.5 2102
Chalk 3 19.0 1000 1 0.1 2443
Fig. 1. Flint models

Since the compressibility of the water and the

porosity of the material are not taken into account into the chosen soil behavior mod-
HOWKH3RLVVRQ¶VUDWLRwas calibrated to obtain realistic P-wave velocity values. After
a sensitivity analysis, an interface with Rinter=1 was considered between the flint and
the embedding soil to allow relative displacement.
Shape effects in the dynamical response … 667

For the material and numerical damping, same approach presented in

Nicolini&Gargarella (2018) has been used. The material damping has been introduced
in the model through the frequency-dependent Rayleigh formulation [5], while for the
numerical damping the standard setting of Newmark implicit method has been used.
To limit boundary effects and reduce the computational time, the size of the models
has been kept 20 times the flint maximum axis. The average mesh element size has
been chosen based on the characteristic of both the soil and the input signal, as 1/10 of
the wavelength associated with the maximum frequency component of the input sig-
nal [3]. To consider radiation damping and avoid wave reflections, viscous boundaries
have been introduced in the model, by using the standard values of C1=1 and
C2=0.25 for the relaxation factors in PLAXIS [4]. Concerning the input signal, a
driveability analysis considering a 7m diameter pile, with a thickness of 77mm, has
been performed (see [2]) to obtain the time history of the compression force at the pile
toe transmitted to the flint through the contact point. The obtained input signal, nor-
malized with respect to its maximum has been used as a dynamic multiplier of 5 dif-
ferent peak force values (1000kN, 2000kN, 3000kN, 40000kN and 5000kN). A para-
metric static analysis was done for each geometry to evaluate the influence of the
impact point on the response of the flint and flint centre and half the horizontal flint
semi-axis were selected as more representative. In plane strain conditions, the circular
flint is equivalent to a cylinder displaced in the soil. This allowed to validate the mod-
el by comparing the obtained results with the T-bar and laterally loaded pile solutions,
in static loading. The failure mechanism obtained for the circular flint is similar to the
simple rotational mechanism [8] obtained for a laterally loaded pile. The difference is
mainly due to the presence of the pile wall, which slightly modifies the soil flow
around the flint (Fig. 2).

Fig. 2. Comparison between the failure mechanism obtained with (left) and without (right) the
pile wall in static conditions

Summarizing, 180 dynamic analyses were performed by combining 3 flint shapes, 2

sizes, 3 material properties for the chalk and 5 peak forces at 2 different impact
points.
668 E. Nicolini and I. Terribile

3 Results

For each analysis, the permanent displacement left after the end of the free oscilla-
tions was retained (Fig. 3). For what concerns the reaction force opposing to the dis-
placement, it was calculated as the vector sum of the applied peak force and of the
consequent maximum horizontal force acting at the pile-flint contact, which comes
from the presence of the pile. The latter has been modeled in a simplified way, i.e. as
a spring whose stiffness simulates the lateral stiffness of the pile section. Finite ele-
ment analysis performed by Aldridge et al. (2005) [6] on the end section of a large
diameter pile indicate that the relationship between the pile toe lateral deflection and
the lateral force acting on the pile can be expressed by the following equation:
ߩ௟௔௧ ൌ ሺͲǤʹͲͶ‫ܨ‬௟௔௧ ‫ ܦ‬ଶ ሻȀሺ‫ ݐ‬ଷ ‫ܧ‬௣௜௟௘ ሻ (1)

where Ulat is the lateral deflection, Flat is the lateral force, D is the pile diameter, t is
the pile wall thickness and E is the pile Young modulus. By inverting the previous
equation, for a pile with a diameter of 7m, a thickness of 77mm (D/t=90) and a steel
<RXQJ¶V PRGXOXV RI  *3D WKH IROORZLQJ HQG VHFWLRQ ODWHUDO VWLIIQHVV KDV EHHQ
obtained:
݇௖ ൌ ሺ‫ ݐ‬ଷ ‫ܧ‬௣௜௟௘ ሻȀሺͲǤʹͲͶ‫ܦ‬ଶ ሻ ൌ ͻͻͻ͸݇ܰȀ݉ (2)

This equation is in agreement with the force obtained by applying a unit horizontal
displacement to the tip of a pile with same characteristics in PLAXIS 3D [7].

Fig. 3. Ellipsoidal flints displacement versus time

To find a rule to protect the pile tip from buckling during pile driving, the relation-
ship between the final flint displacement and the maximum force at the flint-pile con-
tact point has been calculated. Results have been normalized as follows:
ி ௨೤
݂ൌ Ǣ ‫ ݔ‬ൌ (3)
ඥௌೠ ாೠௗ௅ ௅
Shape effects in the dynamical response … 669

where F is the modulus of the maximum force acting at the pile tip-flint contact, Su
and Eu DUHUHVSHFWLYHO\WKHXQGUDLQHGUHVLVWDQFHDQGWKHXQGUDLQHG<RXQJ¶VPRGXOXV
of the embedding chalk, d is the horizontal axis dimension, L is the flint length (1m
due to plane strain conditions) and uy is the permanent vertical displacement. Fig. 4
shows the distribution of the obtained dimensionless quantity in a semi-logarithmic
plot. Permanent vertical displacement less than 1mm have not been considered in line
common practice refusal criteria (rounded to 1000blows/m).

Fig. 4. Distribution of FEA analysis results

A change in the response is noticed in correspondence of x=0.04 and f=1.2. Beyond

these values an increase of the module of the force is observed. The relationship be-
tween f and x can be represented by means of the following two parameters function
on the logarithmic graph.

݂ ൌ ݇ଵ Ž‘‰ଵ଴ ‫ ݔ‬൅ ݇ଶ (4)

If x<0.04, the force can be expressed by using k1=0.56 and k2=2, while if the normal-
ized displacement is greater than x=0.04, the force acting on the pile can be expressed
through the parameters k1=8.6 and k2=13.2. For large diameter piles, it can be as-
sumed that the presence of the flint has no significant influence on the pile toe global
resistance. Then, at a global scale the advancement of the pile is not influenced as
well. and the flint/boulder will undergo essentially an imposed displacement, which
will require a certain reaction force (above figure). Based on this assumption, it is
possible to use the new relationship for a safe drivability assessment.
1. evaluation of the local force to damage the pile tip with the relationship pro-
posed by Aldridge et al (2005); as the actual direction of the force is not pre-
dictable, it is used the formula with the lower multiplicator, that is:
‫ܨ‬௣௘௔௞ ൌ ͳǤͶߪ௬ ‫ ݐ‬ଶ (5)
670 E. Nicolini and I. Terribile

2. given the chalk strength parameters (Su, Eu) and the flint presumable size en-
countered (d, L), it is possible evaluate the minimum displacement per blow
that could damage the pile tip by using equation 5.
3. once the maximum admissible displacement is known, the minimum blows
rate to not damage the pile is evaluated as:
ܾ݈‫݁ݐܽݎݏݓ݋‬ሺܾ݈‫ݏݓ݋‬Ȁ݉ሻ ൏ ͳȀ‫ݑ‬௬ (6)

4. finally, through a drivability analysis it is possible to evaluate the maximum

hammer energy to prevent the local buckling during pile driving, by impos-
ing the blow rate is higher than above values.

4 Conclusions and further developments

Several damages to the pile tip have been reported during pile driving, especially in
chalk or in glacial till, where flints or boulders could be encountered. The force that
arises at the flint-pile contact point could generate an initial imperfection followed by
pile damage propagation. With the aim of finding a relation between the pile ad-
vancement and the potential pile tip damage, parametric dynamic FEA analysis have
been performed. By assuming that for large diameter piles the presence of the flint has
no influence on the overall pile toe resistance during pile driving, a preliminary as-
sessment of a safe driveability procedure has been proposed. Focusing on the pile-
flint contact point, a relationship between the module of the force transferred by the
flint at the pile tip and the flint permanent vertical displacement has been evaluated.
By limiting this force to the value that could damage the pile, given the chalk strength
parameters and the foreseen flints size encountered, it is possible to evaluate the
maximum displacement per blow to prevent the pile tip buckling.
In conclusion, this paper provides a simple analytical method to define a driveability
criteria to avoid pile tip damage, taking into account both the expected flint size and
surrounding soil properties, for large diameter piles compared to flint/boulder size.

References
1. Lord J. A., Clayton C. R. I., Mortimore R. N.: Engineering in Chalk. CIRIA, London
(2002).
2. Nicolini E., Gargarella P., Numerical study of flint/boulder behavior during pile driving.
In: Randolph,M.F.,Dinh Hong Doan, Anh Minh Tang, Man Bui, Van Nguyen, (eds.) Pro-
ceedings of the 1th Vietnam Symposium on Advances in Offshore, pp. 183±189. Springer
Nature Singapore, (2018).
3. Lysmer, J., Kuhlemeyer, R.L.: Finite dynamic model for infinite media. J. Eng. Mech.
Div.
95(4), 859–878 (1969).
4. Kuhlemeyer, R.L., Lysmer, J.: Finite element method accuracy for wave propagation
problems. J. Soil Mech. Found. Div. 99, 421–427 (1973). Technical Report 1973.
5. Clough, R. & Penzien, J.: Dynamics of structures. McGraw Hill, New York (1993).
Shape effects in the dynamical response … 671

6. Aldridge, T.R., Carrington, T.M., Kee, N.R.: Propagation of pile tip damage during instal-
lation. In: Gourvenec, S., Cassidy, M. (eds.) Proceedings of the International Conference
on Frontiers in Offshore Geotechnics, vol. 823. Taylor and Francis (2005).
7. Gargarella, P.: Numerical study of flint/boulder behaviour during pile driving. 0DVWHU¶V
thesis, Politecnico di Torino, Torino (2018).
8. Martin, C., Randolph, M.: Upper-bound analysis of lateral pile capacity in cohesive soil.
Géotechnique, 56 (2), pp. 141-145 (2006).
 The failure envelope of a new shaped cross-plates
foundation for deep-water soft soils

Emilio Nicolini - Cathie Associates, Alessandro Alberio - Politecnico di Milano and

Matteo Castelletti - Cathie Associates

emilio.nicolini@cathie-associates.com
alessandro1.alberio@mail.polimi.it
matteo.castelletti@cathie-associates.com

Abstract. The offshore deep-water oil&gas fields are very often facing the need
of installing heavy structures on soft seabed soils, with complex loading acting
on the foundations. Service structures which are installed to control the oil&gas
production and flowlines like PLETS, ITA, seabed manifolds, are examples of
such structures. Different types of foundations are today available for the deep-
water seabed (mudmats, suction piles and anchors, etc.), which require a fit for
purpose construction and thus are not well suited for automated installation. As
it is believed that automated construction will be favourable to the necessary
cost reduction of deepwater fields, a research in this direction has been acti-
vated. This paper will introduce the results of part of this research program
aimed at the study of a new foundation concept, which is being developed
within a wider project of automated deep-water foundation construction. Re-
sults that will be presented regard a further step of the optimisation study to find
the best shape of a foundation formed by crossed plates inserted vertically in
soft soils. Geotechnical verification of the resistance of the foundation will be
presented in terms of shear and axial force failure envelopes obtained by nu-
merical analyses. The optimisation process will take into account, additionally
to the geotechnical performance, the structural compatibility.

Keywords: Foundation, Cross-shaped, Offshore.

1 Introduction - Foundation geometry and Soil profile

Following the work presented in a previous paper [1], some new features about the
cross-shaped foundations will be herein outlined.
The cross-shaped foundation consists of a central pipe, with 1m diameter, equipped
with vertical steel plates which are crossed at 90° in their middle in a symmetric struc-
ture. 7KHWZR VWHHOSODWHVFUHDWH IRXU ³ZLQJV´DURXQGWKH FHQWUDOWXEH. The width of
the wings from the foundation central point to the outer vertical edge, ܹ଴ , is 4.45m,
while vertically their embedment, ‫ Ͳܪ‬, is 4.50m. On the horizontal plane the structure
can be inscribed inside a square whose side is ‫ܤ‬଴ , equal to 6.25m. As reinforcements,

© Springer Nature Singapore Pte Ltd. 2020 673

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_106
674 E. Nicolini and M. Castelletti

four additional tubular steel beams are added on the top part of each wing. The top of
the central tube is closed with a circular plate which is assumed rigid for calculations
purposes. Loads were applied on this top plate at its centre point.
The cross-shaped foundation analysed in this paper is installed in a very soft clayey
normally consolidated (NC), plastic soil (PI=50). A total unit weight ߛ ൌ ͳͷ݇ܰȀ
݉ଷ was accounted for, while the undrained shear strength was assumed to vary line-
arly with depth: ‫ݏ‬௨ ሺ‫ݖ‬ሻ ൌ ʹ ൅ ͳǤͷ ൉ ‫ݖ‬ሾ݇ܲܽሿ with z starting from seabed level. A stiff-
ness ± strength ratio ‫ܧ‬௨௦ Ȁ‫ݏ‬௨ ൌ ͵ͲͲ, was assumed based on Duncan and Buchignani
work [2]. Finally, for computational purposes, D3RLVVRQ¶VUDWLRߥ ൌ ͲǤͶͻͷ was used
to achieve a constant volume deformation condition without causing numerical prob-
lems. To account for soil remoulding due to the installation procedure, DQ³LQWHUIDFH
FRHIILFLHQW´ ߙ ൌ ͲǤͷ was used. Installation was treated in a parallel research work
and will be referred once concluded; the installation system will be part of the founda-
tion solution.

2 Finite Element model and Preliminary analyses

To perform Finite Element Analyses the commercial software Plaxis (version
2016.02) was used. For further information about finite element coding, soil model
and generated mesh further references are provided in the previous paper [1]. All the
elements which form the foundation were modelled with a purely elastic material
(steel). <RXQJ¶VPRGXOXV‫ܧ‬௦ ൌ ʹͳͲ‫ ܽܲܩ‬DQG3RLVVRQ¶VUDWLRߥ௦ ൌ ͲǤ͵ were assumed.
Reinforcing beams at the top edges of the foundation wings are made by a pipe with
diameter of ߶௕௘௔௠ ൌ ͲǤͷ݉ and a wall thickness ܹܶ ൌ ʹͷ݉݉. Plates were consid-
ered weightless. Multiple thicknesses, PT, were modelled in the structural sensitivity
study.
As first step, starting from the case previously studied [1], some analyses were per-
formed to verify that the new finite element model was correctly built. To reduce the
computational time needed for these preliminary analyses just three load cases have
been analysed; with reference to Fig.1, a force with inclination ܾ ൌ Ͳιǡ ͶͷιandͻͲι
respect to the horizontal axis and ܽ ൌ Ͳι was applied to the foundation top plate. For
details about angle ܾ and ܽ refer to Fig.1.
In order to draw the comparison points (five yellow points in Fig.2) on the previous
failure envelope, failure was associated to a conventional displacement of the load
application point of 2m as was previously assumed in [1]. The blue dots and the red
curve represented in Fig. 2 are those of the previous paper [1] while the triangular
yellow points represent the results of preliminary analyses. The normalization of load
components is obtained as follows:
‫ܨ‬௓ ‫ܨ‬௫௬
‫ܨ‬௓ ǡ ݊‫ ݉ݎ݋‬ൌ Ǣ‫ܨ‬௫௬ ǡ ݊‫ ݉ݎ݋‬ൌ (1)
‫ܤ‬଴ ή ‫ܪ‬଴ ή ‫ݏ‬௨ǡ௕ ‫ܤ‬଴ ή ‫ܪ‬଴ ή ‫ݏ‬௨ǡ௕
where ‫ݏ‬௨ǡ௕ is the ‫ݏ‬௨ at the base of the foundation, computed for‫ ݖ‬ൌ ‫ܪ‬଴ .
The slight discrepancy between the results is due to model improvements, mesh
changes of the new FE model with respect to the one used in [1]. Seen the results is
concluded that the new modelling is perfectly in line with the previous one.
The failure envelope of a new shaped cross-plates … 675

Fig.1. Load angle inclination Fig.2. Comparison of results with [1]

3 Influence of plate thickness on foundation resistance

For all analyses in previous paragraphs, a plate thickness of 10 cm has been consid-
ered (Case I), to have nearly a perfectly rigid behaviour of the structure. To study the
behaviour of the foundation in case of different thicknesses, two other cases have
been investigated:

ƒ Case II: ܲܶ ൌ ʹǤͷܿ݉

ƒ Case III: ܲܶ ൌ ͳǤͷܿ݉, with steel vertical stiffeners(thk:ʹܿ݉, variable h)

In case III the foundation is formed by the cross-shaped elements (wings), reinforced
by five stiffeners, as shown in Fig. 3. A simplified approach was used to model this
kind of reinforcements: each wing has been divided in five parts and for each of these,
the average value of the moment of
inertia with respect to local ‫ ݏ‬axis has
been computed:
ͳ
j ‫ܫ‬௜ ൌ ή ሺʹ‫ܦ‬௜ ሻଷ ή ܹܶ௥௣ (2)
ͳʹ
where ‫ܦ‬௜ is the average height of the ݅-
th horizontal strip and ܹܶ௥௣ is the
thickness of the stiffeners. The total
i
moment of inertia ሺ‫ܫ‬௥௘௔௟ ሻ accounting
Fig. 3. Reinforced and equivalent wing, plan
for the 5 vertical reinforcement plates is view and details
then computed as 5 times ‫ܫ‬௜ . To analyse this
foundation geometry a simplification has been used: an equivalent plate with wall
thickness of ͳǤͷܿ݉ with five different values of E decreasing with depth, is used to
simulate the increasing inertia moment due to the stiffeners. The new wing is divided
676 E. Nicolini and M. Castelletti

in five horizontal bands ሺ݆ ൌ ͳǡ Ǥ Ǥ ǡͷሻ, see right sketch of Fig.3. In this case the I of the
equivalent plate is be computed as:
ͳ ଷ
‫ܫ‬௕௔௡ௗೕ ǡ௘௤ ൌ ή ൫ܹܶ௘௤ ൯ ή ‫ܮ‬௝ (3)
ͳʹ
Considering that, the relation:
‫ܫ‬௥௘௔௟ǡ௜ ή ‫ܧ‬௥௘௔௟ ൌ ‫ܫ‬௕௔௡ௗǡ௝ ή ‫ܧ‬௘௤ǡ௝ (with ݅ ൌ ݆) (4)
is always true; assuming ‫ܧ‬௥௘௔௟ ൌ ‫ܧ‬௦ is possible to compute the ‫ܧ‬௘௤ǡ௝ for each band.
Note that ‫ܧ‬௘௤ǡ௝ decreases with depth, due to the reducing contribution of the reinforc-
ing plates. Simplified plates for the FEA will be treated as orthotropic plate, with
GLIIHUHQWYDOXHVRI<RXQJ¶VPRGXOXV: ‫ܧ‬ଵ ൌ ‫ܧ‬௦ in the direction parallel to the central
tube, and ‫ܧ‬ଶ ൌ ‫ܧ‬௘௤ǡ௝ in the direction normal to the central tube. In Fig. 4 a comparison
between the results for three plates geometry is presented. The foundations were
loaded with the same external force and with the two inclinations mentioned before
(ܾ ൌ Ͳι and Ͷͷι). As can be noted, in the 3 models the mobilisation of the foundation
reaction is obtained at significantly different displacement levels. However, it is noted
that if the load is kept within SLS range, i.e. by applying a FoS=2 (average of what is
required by standards), the difference in displacements is negligible.

Fig.4. Results comparison (Case I, II, III)

4 Structural checks on Plates

For all cases, structural checks of steel plates have been performed as prescribed in
Eurocode 3 [4]. According to it, following inequality must be satisfied in the plates:
ଶ ଶ ଶ
݂௬௞ǡ௦௧௘௘௟
ߪ௘௤ǡ௦ௗ ൌ ට൫ߪ௫ǡௌௗ ൅ ߪ௬ǡௌௗ െ ߪ௫ǡௌௗ ߪ௬ǡௌௗ ൅ ͵߬௫௬ǡௌௗ ൯ ൑ ߪ௘௤ǡோௗ ൌ (5)
ߛெ଴
௡ೣǡೄ೏ ௠ೣǡೄ೏ ௡೤ǡೄ೏ ௠೤ǡೄ೏ ௡ೣ೤ǡೄ೏ ௠ೣ೤ǡೄ೏
with ߪ௫ǡௌௗ ൌ േ ೟మ
Ǣߪ௬ǡௌௗ ൌ േ ೟మ
Ǣ߬௫௬ǡௌௗ ൌ േ ೟మ (6)
௧ ௧ ௧
ర ర ర

where ‫ ݐ‬is the plate thickness, ݂௬௞ǡ௦௧௘௘௟ is the yield stress, assumed as ʹ͹ͷ‫ܽܲܯ‬, and
ߛெ଴ ൌ ͳǤͳ is a partial factor for resistance of cross section. Fig.6 presents a graphical
representation of the stress state in the four plates for Case III, for ܽ ൌ Ͷͷι. A nor-
malization with respect to ݂௬ௗ ൌ ݂௬௞ Ȁߛெ଴ was used. To consider the presence of stiff-
eners in the structural checks, in case III, ‫ܯ‬ଵଵ and ܳଵଷ components have been as-
sumed to be supported by stiffeners themselves and considered as zero in previous
equation (6). Vertical stiffeners were checked too: they are generally within the al-
The failure envelope of a new shaped cross-plates … 677

lowable stress limit, except for some limited areas close to foundation foot where
larger or thicker stiffeners should be used.

5 Sensitivity analysis on
foundation geometry

From the geometry of ‫ܣ݁ݏܽܥ‬, two different

cases have been studied by varying the wings
measures and foundation embedment:

ƒ‫ܤ݁ݏܽܥ‬ǣܹ஻ ൌ ܹ଴ Ȁʹܽ݊݀‫ܪ‬஻ ൌ ‫ܪ‬଴

ƒ‫ܥ݁ݏܽܥ‬ǣܹ஼ ൌ ܹ଴ ܽ݊݀‫ܪ‬஼ ൌ ‫ܪ‬଴ Ȁʹ

For each case, a load was applied, with varying

Figure 5. Internal action in plates in
Plaxis software inclination: b = ±90°, ±75°, ±60°, ±45°, ±30°,
±15, 0°, keeping the ܽ angle value constant
ሺͲιǡ Ͷͷιሻ. Failure load has been chosen as the one corresponding to a displacement of
ʹ݉ of the application point. Seen the limited effect of the foundation thickness on the
results (in terms of ultimate resistance), as outlined in Fig.4, a plate thickness of 10cm
was adopted, as it was done in ܿܽ‫ܫ݁ݏ‬.

Fig.6. Normalized plates stress state ሺܽ ൌ ͶͷιሻǢ 1.5cm without (left) and with (right)
comparison between plate with thickness of stiffeners (Case III).

Fig.7 shows a comparison between three analysed cases; results are normalized as
follows:
ƒ
ிೋ ிೣ೤
ƒ ƒ•‡ǣ‫ܨ‬௓ ǡ ݊‫ ݉ݎ݋‬ൌ Ǣ‫ܨ‬௫௬ ǡ ݊‫ ݉ݎ݋‬ൌ
ௐ಴ ήு಴ ή௦ೠǡ್ǡಲ ௐ಴ ήு಴ ή௦ೠǡ್ǡಲ
ிೋ ிೣ೤
ƒ ƒ•‡ǣ‫ܨ‬௓ ǡ ݊‫ ݉ݎ݋‬ൌ Ǣ‫ܨ‬௫௬ ǡ ݊‫ ݉ݎ݋‬ൌ
ௐಳ ήுಳ ή௦ೠǡ್ǡಳ ௐಳ ήுಳ ή௦ೠǡ್ǡಳ
ிೋ ிೣ೤
ƒ ƒ•‡ǣ‫ܨ‬௓ ǡ ݊‫ ݉ݎ݋‬ൌ Ǣ‫ܨ‬௫௬ ǡ ݊‫ ݉ݎ݋‬ൌ
ௐ಴ ήு಴ ή௦ೠǡ್ǡ಴ ௐ಴ ήு಴ ή௦ೠǡ್ǡ಴
678 E. Nicolini and M. Castelletti

Fig.7. Different geometries results comparison

where ‫ݏ‬௨ǡ௕ǡሺ஺ǡ஻ǡ஼ሻ is the ‫ݏ‬௨ at the base of the foundation for each case. Interpolation
curves shown in the plots were computed according to the same approach outlined in
the previous paper [1]. Plots well show the symmetrical behaviour of the foundation
in tension and compression.

6 Conclusions and way forward

The numerical analyses presented herein with reference to previous paper [1], show
that the structural compatibility of the cross-shaped foundation is fulfilled. The use of
vertical stiffeners of variable height allows having a fully verified structure which
remains within the elastic range at the ULS geotechnical failure load of the founda-
tion, and which in the SLS load level range (estimated by assuming a FoS=2 on the
failure load) provides marginal displacements. As well, this paper presents the failure
domains of 2 variations of the base crossed-plates foundation; by comparing the last
two new explored geometries, B and C, it can be noted how the different geometries
react to different load components. All structural elements were verified; performed
stress checks prove the structural compatibility. Effects of cyclic loads were not
treated in this paper, but this aspect will be dealt with in a future development of this
research line.
This paper refers to a fixed shape of the cross foundation; the impact and the modifi-
cation of the shape of this foundation (e.g. the adaptation of the length and width to
the applied loads) are currently being explored by the Authors.

7 References
1. Nicolini & Castelletti: Numerical analyses to determine the failure envelope of a cross-
shaped plate foundations, OSIG 2017, pp.940-946
2. Duncan JM and Buchignani AL. (1976). An Engineering Manual for Settlements Studies.
Dept. of Civil Engineering ± University of California ± Berkeley - 94pp.
3. Eurocode UNI EN 1993-1-3, 2018
 Enhanced Oil Production in the lower Miocene reser-
voir by Multistage Fracturing, Offshore Viet Nam
Huu Truong Nguyen1, W. Bae2, Van Hung Nguyen1)
1
Petrovietnam University, Ba Ria ± Vung Tau, Vietnam
2
Sejong University, Gwangjin-ku, Seoul, Korea
truongnh@pvu.edu.vn, wsbae@sejong.ac.kr, hungnv@pvu.edu.vn

Abstract. Oil production has been declined in the fractured basement reservoir
from these fields of the Cuu Long Basin as those reservoir pressures were high-
ly decreased. To produce oil, those reservoirs from these fields have been stim-
ulated by using various techniques for enhanced oil production. Water injection
at secondary recovery to maintain the reservoir pressure has been flooded be-
cause this technique has still brought high efficiently oil production, but the
high-water cut has been very challenged while producing oil from the reser-
voirs. The most efficient way to recover oil in the Cuu Long basin is to stimu-
late the lower Miocene reservoir by hydraulic fracturing. This study presents
the effective multi-stage hydraulic fracturing in the lower Miocene reservoir for
single X well of the Cuu Long basin on the continental shelf offshore Viet Nam.
The calculated results of multistage fracturing show that the average effective
wellbore radius, average pseudo-skin, average productivity ratio, fracture con-
ductivity are higher than those of the blanket and un-stimulated case. The inte-
grated model of multistage fracturing consists of fracture geometry, fracture pa-
rameters, conductivity and production model.
Keywords: Multi-stage fracturing, Lower Miocene reservoir, effective hydrau-
lic fracturing.

1. Introduction

In the few decades, hydraulic fracturing is a stimulation technique to carry out

enhanced oil and gas production from either low permeability reservoir or medium to
high permeability reservoir by increasing fracture conductivity. The lower Miocene
reservoir of the White Tiger oil field is a sandstone reservoir with high heterogeneous
level and complex fracture networks at which sedimentary of the lower Miocene
sequence has high heterogeneity and complicated geological structures. The reservoir
can be divided into two structures namely Northern dome and Southern dome. The
reservoir permeability varies from 0.1mD to 1000mD and the porosity ranges from
15% to 23% in the Northern dome. Due to the fault and rock compaction with high
sedimentary thickness, the porosity generally gradually decreases towards the
southern and eastern faults. The reservoir depths currently range from 2759m to
2998m with differential lithological compositions. In detail, the sandstone has high
percentage ranging from 40% to 65% of lithology, and feldspars ranging from 10% to
25%, micas from 2% to 5%, fragments from 2% to 13%, and clay or carbonate
cements from 12% to 15% of the lithology. The reservoir temperature distribution
varies from 800C to 1000C but some production and injection wells record
1.0670C/100ft thermal gradient [1]. In single well, the valuable fracture gradient is not

© Springer Nature Singapore Pte Ltd. 2020 679

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_107
680 H. T. Nguyen et al.

too high and is a criterion to design net pressure and to select pump rate as well as
pump horsepower. Thus, the hydraulic fracturing stimulation is a good method to
increase not only the reservoir permeability but also the fracture conductivity from
fractured well. In fact, single well has several layers and needs to be stimulated by
hydraulic fracturing to increase its fracture conductivity. Therefore, the multi-stage
fracturing for single well in the lower Miocene reservoir has been considered because
this technique has brought the number of benefits when compared with blanket
fracturing. This study presents these benefits of multiple stage fracturing including the
effective wellbore radius, average skin, net pressure, fracture geometry, net present
value and economic performance in transient regime.

2. Material and methods

2.1. Fracture geometry model

Most fracture geometries have been described in either two dimensional (2D) or three
dimensional (3D) models in which models show an approximate description of real
fracture geometry during proppant slurry pumping under high pressure [2][3]. Both
2D and 3D models without fluid leak-off are not accurate to describe fracture
geometry for permeable reservoir. In the lower Miocene reservoir, these reservoir
porosity and permeability are relatively low. Then, the 2D PKN-C model with fluid
leak-off is significant to illustrate fracture geometry for this study [4][5]. Without 2D
PKN-C model, radial, GDK, 3D, radial-C, GDK-C models are not satisfied to show
an exact fracture geometry for the lower Miocene reservoir.

2.2. Transient oil production

Oil production from the fractured well as the transient flow regime with a bottom hole
pressure can be calculated as Economides et al., [6]:
162.6q o Boȝ k
(pi -p wf )= (logt+log -3.23+s) (1)
kh )ȝFt Uw' 2
Pi = Reservoir pressure, psi
t = Oil production at transient time, months
k =Reservoir permeability, mD
Ct = Total compressibility, psi-1
s = Skin factor of fractured well
h = Reservoir thickness, ft
ߤ = Oil viscosity, cp
Bo = Formation volume factor, bbl/STB
r¶w = Effective wellbore radius, ft
Pwf = Flowing bottom hole pressure, psi

2.3. Net present value model

Enhanced Oil Production in the lower Miocene … 681

The net present value can be calculated as the following model below:
N
Rn
NPV=¦ n
-Ctr (2)
n=1 (1+i)
In which:
NPV = Net present value, $
Rn = Revenue of year N
i = Interest rate, %
Ctr = Total treatment cost, $

3. Application for a single well in the lower Miocene reservoir

Table 1. Reservoir data and fracturing parameters [7][8][9]

Parameters Stage 1 Stage 2 Stage 3 Blanket

Depth, TVD, ft 9439 9175 9086 9224
Drainage radius, ft 1640 1640 1640 1640
Tubing size, ft 0.328 0.328 0.328 0.328
Reservoir thickness, ft 88.6 86.4
Porosity, %
Permeability, mD
Reservoir pressure, Pi, psi 2970 2970 2970 2970
BHP, psi 2000 2000 2000 2000
Sandstone Poison¶s ratio 0.25 0.25 0.25 0.25
Shale Poisson¶s ratio 0.35 0.35 0.35 0.35
Modul young, psi
Reservoir temperature, 0C
Oil API
Formation volume factor,RB/STB 1.4 1.4 1.4 1.4
Total compressibility, psi-1
Oil viscosity, cp 1.074 1.074 1.074 1.074
Bubble point pressure, psi 2000 2000 2000 2000
Closure pressure, psi 4861 4844 4916 4704
Injection rate, bpm
Injection time, min
End of the job, EOJ, ppg
Flow behavior index, n 0.341 0.341 0.341 0.341
Consistency index, K, lbf.sn/ft2 0.12 0.12 0.12 0.12
Leak- off coefficient, C, ft/min0.5 0.00227 0.00227 0.00227 0.00227
Proppant type Sintered Ball Bauxite sized 16/30): Intermediate Strength Proppant (ISIP), proppant
density 227.76lb/ft3, average diameter 0.038in, proppant packed permeability 400,000mD,
proppant packed porosity 0.35. Proppant damaged factor 0.5
Fracturing fluid type YF540HT: Surfactant 2pptg, Stabilizer 15pptg, Crosslinker 0.35pptg, Buffer
7.5pptg, Breaker Na2S2O8 8pptg, Encapsulated, gelling agent 11.2pptg, activator 1pptg, and Bio-
cide 0.5pptg

Table 1 shows the well production data such as tubing size, drainage radius, reservoir
porosity, reservoir permeability, oil API, reservoir temperature, flowing bottom hole
pressure, oil viscosity, bubble point pressure, formation volume factor, total com-
pressibility, reservoir pressure, those use for production simulation at transient re-
682 H. T. Nguyen et al.

gime. The fracturing parameters are used to design fracture treatment including leak-
off coefficient, flow behavior index, consistency index, end of the job, closure pres-
sure, proppant type and fluid type. The three sandstone layers have underlying and
overlying shale layers and are designed three stages fracturing, which are seen in the
table 1. With the blanket fracturing, all the intervals are as one large layer and it is
fractured from the midpoint at 9224ft.

4. Results and discussions

Table two compares the net pressure and fracture geometries between multistage,
blanket fracturing. The table also shows the average maximum fracture width and
fracture half-length of the first stage from multistage being 0.372 inch and 696.5 ft
respectively that are higher compared to fracture geometry of blanket. This is because
both fracture half-length and fracture width are currently inversely proportional to the
fracture height while the fracture width is directly proportional to fracture half-length.
While the fracture height of the blanket is 265ft, which is the highest value compared
to fracture heights of multistage. As the result, the fracture geometry of blanket is
smaller than that of multistage. The net pressure of the blanket is 195psi, which is too
much lower compared to the average net pressures from multistage of 655.7psi. In the
petroleum industry, the net pressure should be designed below 1000psi in order to
avoid formation breakdown pressure. In addition, table 2 presents the average fracture
width varied from 0.22in to 0.236in of multistage fracturing. Those are four times
higher the proppant diameter, which is the criteria for proppant slurry entered to frac-
ture and kept fracture opened Schechter [10]. Oil production rate depends on the val-
uable fracture conductivity which allows reservoir fluid entering the wellbore. A frac-
ture conductivity depends on the number of factors including closure pressure, prop-
pant concentration among fractures, proppant types, proppant pack permeability,
proppant pack porosity. Table 3 presents the production analysis from multistage and
blanket stage fracturing. To investigate the oil production rate, the average negative
skin factor of multistage fracturing is -7.00, which is much lower than that of negative
skin from blanket stage of -5.90 and the average effective wellbore radius of multi-
stage is 357.9ft which is much higher than the wellbore radius of the blanket stage of
120.3ft. Thus, this is an explanation that the cumulative oil production and the
productivity index of multistage are 7,854,300bbls and 5.5, respectively, which are
much higher than that of blanket stage and un-stimulated case at transient regime as
seen in table 4. Fig.1 compares the production rate among three cases in the transient
regime and shows the production rate from multistage fracturing is higher in compari-
son with the blanket and un-stimulated case. The fig. 2 presents these NPV versus the
multistage fracturing, blanket and un-stimulated case, where the NPV of the multi-
stage fracturing is the largest. Thus, multistage fracturing provides more benefits than
blanket stage and un-stimulated case. Table 5 gives the economic data in the field
experience, which is used to calculate the NPV from multistage, blanket, and un-
stimulated case.
 Enhanced Oil Production in the lower Miocene … 683

Table 2. Results from fracture pressure and fracture geometry

Parameters Stage 1 Stage 2 Stage 3 Blanket

Net pressure, Pnet, psi 665.3
Treating pressure, Ptreat, psi 5526 5469 5593 5068
Pump horse power, HHP 2708 2680 2741 2483
Half-length, ft 729.1 634.1 726.3 234.2
Maximum fracture width, in 0.379 0.362 0.376 0.330
Average fracture width, in 0.24 0.22 0.236 0.207
Fracture height, ft 88.6 86.4

Table 3. Fracture conductivity and production profile analysis

Parameters Stage 1 Stage 2 Stage 3 Blanket

Fracture conductivity, md.ft 5232 5244 5194 5224
Closure pressure, psi 4861 4844 4916 4704
Dimensionless conductivity, FCD 0.3 0.34 0.3 0.93
Pseudo-skin factor, Sf -7.04 -6.90 -7.03 -5.90
Effective wellbore radius r¶w, ft 374.6 325.8 373.2 120.3
Productivity index, J/J0 5.8 5.3 5.8 3.3

700
400
Multitage fracturing
600
Blanket 350
Production rate, 1000bbl

500 Un-stimulated case

300
NPV,1,000,000, $

400 250

300 200
150
200
100
100
50
0 0
0 200 400 600 800 1000 1200 Multitage fracturing Blanket Un-Stimulated case
Time (days)

Fig. 1. Comparison oil production rate among mul- Fig. 2. Comparison NPV among multistage, blanket and
tistage, blanket and un-stimulated case in three un-stimulated case in three years
years

Table 4. Total cost and economic analysis in three years

Parameters Fluid used, Proppant used, HHP Proppant and fluid

gals lbs cost($)
Blanket 70,956.32431 141,396.2 5068 158,985.8
Multistage 215,642.9945 467,608.5 16,588 471,597.4
Parameters Cumulative oil Oil ($) Total cost ($) NPV ($)
(bbl)
Un-stimulated case 1,318,900 142,968,000 0 118,617,326.82
Blanket 5,789,800 347,388,000 158,986 289,544,184.75
Multistage 7,854,300 471,258,000 417,597 393,807,290.65
684 H. T. Nguyen et al.

Table 5. Economic data

Parameters Value
Proppant cost, ($/lbm) 0.4
Fluid cost, ($/gal) 1
Pump cost, $/hr/HHP 3.25
Fixed cost, ($) 15000
Interest rate, (%)
Oil price ($/bbl)
Years, NPV 3

5. Conclusions

Research on multi-stage fracturing for a single well in the lower Miocene reservoir
leads to following conclusions:
For fractured wells, oil productivity index of multistage fracturing is much higher
than that of base case and blanket stage.
The average negative skin factor from multistage fracturing is lower than that of
the negative skin factor of blanket stage and un-stimulated case.
The proppant concentration among fractures from multistage fracturing is much
higher than that of blanket stage fracturing, which results in higher fracture conductiv-
ity as well as higher productivity index.
The average net pressure from multistage fracturing is higher than that of blanket
fracturing, which results in longer fracture half-length and wider fracture width as
well as higher fracture volume.

References

1. Ngoc T.B. Nguyen, Cuong T.Q. Dang, Wisup Bae, Taemoon Chung.: Integrating
Geological Characterization and Historical Production Analysis to Optimize Field
Management of Lower Miocene Reservoir in the White Tiger Field, Vietnam.
137045-PA SPE Journal (2010)
2. T.K.Perkins, L.R.Kern.: Width of hydraulic fractures. Journal of Petroleum Technol-
ogy. 13(9): p. 937 ± 949 (1961)
3. K.G.Nolte.: Determination of proppant and fluid schedules from fracturing pressure
decline. SPE Production Engineering. 1(4): p. 255 ± 265 (1986)
4. R.P.Nordgren.: Propagation of a vertical hydraulic fracture. Society of Petroleum
Engineers Journal. 12(4): p. 306 ± 314 (1972)
5. Peter Valkó, Michael J.Economides.: Hydraulic fracture mechanics. John Wiley &
Sons, Chichester, England (1995)
6. Cinco-Ley, H. and Samaniego, V.J.: Transient Pressure analysis for fractured wells´,
Gas TIPS (Fall), pp.347 (1974)
7. Haiqing Yu and M.M.Rahman.: Pinpoint Multistage Fracturing of Tight Gas Sands-
An Integrated Model with Constraints. SPE 152439, SPE Middle East Unconvention-
al Gas Conference and Exhibition held in Abu Dhabi, UAE (2012)
8. BJ-PVDrilling JV Company Ltd.: Fracturing report summary (2011)
9. BJ.: Hydraulic fracturing post job report for wells (2010)
10. Schechter, R. S.: Oil Well Stimulation. Englewood Cliffs, NJ: Prentice Hall (1992)
 Properties of geopolymer modified sludge generated in
landslide area designed by Taguchi method

Minh Chien Vu1, Tomoaki Satomi2 and Hiroshi Takahashi3

1
Doctor student, Graduate School of Environmental Studies, Tohoku university
chien.minh.vu.p5@dc.tohoku.ac.jp
2
Assistant Professor, Graduate School of Environmental Studies, Tohoku university
satomi@mail.kankyo.tohoku.ac.jp
3
Professor, Graduate School of Environmental Studies, Tohoku university
htaka@mail.kankyo.tohoku.ac.jp

Abstract. It has been claimed that heat treatment is necessary for the reaction of
geopolymer to achieve the desired strength and form. +RZHYHU LW¶V DOVR WKH
drawback affecting the application of Geopolymer in the construction site. There-
fore, instead of heat treatment, the other important factors for synthesizing the
geopolymer matrix in the soft soil environment, is the main focusing of this pa-
per. The Taguchi method was used to determine the most important factor that
affects several properties of geopolymer modified sludge by minimizing the ef-
fects of variation, but without eliminating the cause. The results indicate that the
geopolymer content, clay content in sludge, initial water content and moisture
content play a unique role in different properties of modified sludge.

Keywords: Sludge, fly ash, geopolymer, Taguchi method

1 Introduction
Soft and highly compressible soils (sludge) constitute a major problem in geotechnical
engineering projects. Cement treatment of the sludge is commonly used as stabilization
technique that modifies the workability, compaction characteristics and increases the
compressive strength of soil [1]. However, the production of cement does need a huge
amount of energy and release the carbon dioxide to the atmosphere. Geopolymer is a
kind of inorganic polymer, alternatives to replace cement, produced by the reaction of
aluminosilicate materials with alkaline solutions [2]. One remarkable point about geo-
polymer is the elimination of cement usage, and 44-64% reduction of greenhouse gas
emission [3]. Compare to Portland cement, geopolymer could present the challenge on
requirement elevated temperature curing [4]. However, several works have been con-
ducted on fly ash based geopolymers modified sludge all indicate the effects of the
others parameters [5,6]. Therefore, there are no works incorporating all factors and take
into account the domination of each factor affecting on the performance of modified
sludge.
To overcome this discrepancy, by using a suitable design of experiment method
(DOE) one may consider some of the factors affecting properties. Therefore, the aim of
this study proposes the suitable conditions for using geopolymer modified sludge (high
amount of initial water content) in landslide area by concerning the most frequent

© Springer Nature Singapore Pte Ltd. 2020 685

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_108
686 M. C. Vu et al.

factors resulting in failure strength and failure strain. Taguchi method was used to in-
vestigate the effects of more parameters simultaneously since this systematic approach
can significantly reduce empirical H[SHULPHQWV¶ WLPH DQG FRVWV [7]. Four main con-
trolled factors, including geopolymer content, clay content of sludge, initial water con-
tent of sludge and moisture content of modified sludge were considered. A total number
of 25 experiments were conducted according to the L25 array proposed by acquired
DOE method. Then obtained results were evaluated by Signal to Noise Ratio (S/N)
derived from the Taguchi approach.

2 Design of experiment and methodology

2.1 Materials

The samples of sludge were collected from the landslide area in Tohoku, Japan. Be-
cause of the difficulty to maintain the quality of reality sludge, the imitation sludge
(clay: silt at ratio 3:2) which has the same physical and chemical properties was used.
The grain size distribution of 2 samples of soft soil and imitative soil were obtained as
shown in Fig.1 Fly ash class F (FA) used as the main source of aluminosilicate material
for making geopolymer (Geo). Figure 1 shows the grain size distribution curve of FA.
The average grain size of FA is 13.76 μm. The specific gravity of FA is 2.14. The liquid
alkaline activator (AA) is a mixture of sodium silicate (Na2SiO3) solution with the
weight ratio of SiO2/ Na2O = 2.8 and sodium hydroxide (NaOH) with a concentration
of 12 M.
100
90
Cumulative finer volume [%]

80
70
60
50
40
Imitation sludge
30
Fly ash
20
Reality sludge 1
10 Reality sludge 2
0
0.1 1 10 100 1000 10000
*UDLQVL]H>ȝP@

Fig. 1. Grain size distribution of imitation sludge, reality sludge and fly ash

2.2 Design of experiment

Four main considered factors for the Taguchi experimental design were comprised of:
³JHRSRO\PHUFRQWHQW´³clay content´³LQLWLDOZDWHUFRQWHQW´³moisture content´ZKLOH
each implemented at five different levels. The variation levels of each parameter have
Properties of geopolymer modified sludge … 687

been presented in Table 3. The L25 array is designed (Table 1) that is suggested by
Taguchi method for four factors of five levels.

Table 1. The introduced levels for each factor in experiment design

Factor Description Level 1 Level 2 Level 3 Level 4 Level 5

A Geopolymer 5 10 15 20 25
content [%]
B Clay content 0 30 60 70 100
[%]
C Initial water 40 50 60 70 80
content [%]
D Moisture con- 26 28 30 32 34
tent [%]

2.3 Experiment procedure

This procedure is imitating one of the methods that are often used in the treatment
sludge process on the construction site. The sludge was prepared by mixing clay, silt
and water (the initial water content following Table 2) at 5 - 8 minutes. Then, fly ash
and alkaline activator solution are mixing with the sludge about 8 - 10 minutes to
achieve a uniformity. After mixing the mixture will be poured into the container (plastic
box) and stored in the oven for initial curing. The initial curing temperature and curing
temperature is set at 20oC (±2oC) because it is average temperature in Japan and the
countries in temperate zone. The total mass of the container and sludge changing by the
time are carefully measured and recorded. When the moisture content of mixture
dropped to the design condition, the specimens were made following the previous re-
search [6]. These specimens were wrapped by the vinyl sheet and kept in the oven. For
quick recovery in landslide area, the curing time is set at 7 days. After that, the speci-
mens were unwrapped and investigated the characteristic by Unconfined compression
test (UC test) in term of failure strength and strain.

3 Result and discussion

3.1 Failure strength

Table 2 shows the experiment result of failure strength of modified sludge after 7 days
curing. Table 3 indicates the response table of input factors on the failure strength of
modified sludge after 7 days curing. According to this result, geopolymer content is the
most influent factor on the failure strength of modified sludge. On the contrast, the
changing of moisture content seem OHVVHIIHFWWKDQDQRWKHUIDFWRUV,W¶VFOHDUO\WRVHH
that the increasing the geopolymer content the failure strength of modified sludge in-
creasing, meanwhile, the increasing of initial water content shows the opposite. The
tendency. failure strength of modified sludge increasing when the clay content
688 M. C. Vu et al.

increasing from 0% to 60%. Then, as the clay content is over than 60% up to 100%, the
failure strength of modified sludge is decreased. The reason that causes this phenome-
non is the particle size of clay is smallest compared to particle size of fly ash and silt.
It could fulfil the porous inside the structure of modified sludge. But too much clay
content could reduce the larger size particle, that is mainly loaded the compressive
strength.

Table 2. Mixture design with results of failure strength, failure strain and moist density

No. Geopolymer Clay Initial Moisture Experiment results

content [%] content water content Failure Failure Moist
[%] content [%] strength strain density
[%] [kN/m2] [%] [kg/m3]
T1 5 0 40 26 217.5 0.72 1344.7
T2 5 30 50 28 210.5 1.16 1636.9
T3 5 60 60 30 230.9 4.53 1778.4
T4 5 70 70 32 188.1 4.57 1782.2
T5 5 100 80 34 157.6 3.19 1788.5
T6 10 0 50 30 188.1 1.10 1482.9
T7 10 30 60 32 212.7 1.34 1612.2
T8 10 60 70 34 270.1 3.62 1688.3
T9 10 70 80 26 189.6 1.09 1515.4
T10 10 100 40 28 243.0 1.18 1792.6
T11 15 0 60 34 222.1 1.39 1584.8
T12 15 30 70 26 177.1 0.28 1499.9
T13 15 60 80 28 243.0 0.55 1518.0
T14 15 70 40 30 334.9 3.27 1773.4
T15 15 100 50 32 212.7 2.78 1764.6
T16 20 0 70 28 212.7 0.94 1481.0
T17 20 30 80 30 231.4 0.80 1600.3
T18 20 60 40 32 416.0 4.60 1724.4
T19 20 70 50 34 294.2 6.12 1695.5
T20 20 100 60 26 209.9 1.37 1663.7
T21 25 0 80 32 204.4 0.75 1561.0
T22 25 30 40 34 381.3 6.67 1691.7
T23 25 60 50 26 547.4 1.25 1728.4
T24 25 70 60 28 431.6 2.50 1807.4
T25 25 100 70 30 319.9 2.11 1733.3
Properties of geopolymer modified sludge … 689

Table 3. Response Table for Signal to Noise Ratio of Failure strength

Level Geopolymer con- Clay content Initial water Moisture content

tent [%] [%] content [%] [%]
1 45.99 46.39 49.79 47.70
2 46.79 47.38 48.53 48.23
3 47.33 50.15 47.98 48.13
4 48.41 48.72 47.15 47.44
5 51.08 46.95 46.14 48.10
Delta 5.10 3.76 3.65 0.79
Rank 1 2 3 4

Opposite to the geopolymer content, the increasing amount of initial water content
reduces the failure strength of modified sludge. The high amount of initial water content
could flocculate soil particle, raise the plastic limit as well as reduce the maximum bulk
density of modified sludge. The influence of moisture content on the failure strength
performance is not explicitly because the gap of each level of moisture content is not
high enough (2%).

3.2 Failure strain

Table 4. Response Table for Signal to Noise Ratio of Failure strain

Level Geopolymer con- Clay content Initial water Moisture content

tent [%] [%] content [%] [%]
1 7.23 -0.07 7.72 -1.36
2 3.70 1.72 6.93 1.23
3 1.30 7.21 5.84 6.13
4 5.90 9.63 3.84 7.23
5 6.43 6.07 0.24 11.32
Delta 5.93 9.70 7.48 12.68
Rank 4 2 3 1

Table 2 and Table 4 shows the failure strain of modified sludge after 7 days and contri-
bution level of each factor evaluated by S/N ratio on this characteristic, respectively.
As the result, the most effect variable on the failure strain of modified sludge is the
moisture content, meanwhile, the geopolymer content presents the less impact among
four factors. The increasing of moisture content from 26% to 34% could increase the
failure strain of modified sludge cause the high water reflecting on the flexible of spec-
imen (Fig.5). However, increasing the initial water content accordant to increasing the
initial curing time leads to the decreasing of strain performance. As mentioned above,
higher of clay content not only could reduce the porosity but also improve the transfer
690 M. C. Vu et al.

loading stress capacity of modified sludge. But as the clay content is over than 70%,
this phenomenon is less effective. Furthermore, the failure strain of modified sludge
decreased as geopolymer content from 5% to 15% and increased as geopolymer content
from 15% to 25%.

4 Conclusion

This study is conducted to investigated the role of several factors on the characteristic
of geopolymer modified sludge by the application of Taguchi method to design the
experimental condition and analysis the results. Based on the results from this experi-
mental research, the following conclusions have been drawn:

ņ The decrease level of factor that contributes to the failure strength of modified
sludge is geopolymer content, clay content, initial water content and moisture con-
tent. Meanwhile the increasing of geopolymer content is increasing the failure
strength of modified sludge, the increasing of initial water content shows the oppo-
site tendency. The clay content up to 60% increases the failure strength, over than
that, the failure strength is decreased.
ņ The decrease level of factor that contributes to the failure strain of modified sludge
is moisture content, clay content, initial water content and geopolymer content. The
moisture content and initial water content shows the contradictory of influence. The
increasing of clay content also increases the failure strain as the values is less than
70%. The geopolymer content from 5% to 15% decreases the failure strain but from
15% to 25%, it increases the failure strain.

References
1. Takahashi, H., Kuribara, H., and Satomi, T.: Study on durability for drying and wetting of
cover soil for radiation-contaminated soil made of tsunami sludge, Journal of JSEM 14 (Spe-
cial Issue), 309-313 (2014).
2. Duxson, P., Fernández-Jiménez, A., Provis, J. L., Lukey, G. C., Palomo, A., and Van
Deventer, J. S. J.: Geopolymer technology: the current state of the art, Journal of Materials
Science 42(9), 2917-2933 (2006).
3. Mclellan, B. C., Williams, R. P., Lay, J., Van Riessen, A., and Corder, G. D.: Costs and
carbon emissions for geopolymer pastes in comparison to ordinary Portland cement, Journal
of Cleaner Production 19, 1080-1090 (2011).
4. Shaikh, F. U. A. and Vimonsatit, V.: Compressive strength of fly-ash-based geopolymer
concrete at elevated temperatures, Fire and Materials 39 (2), 174-188 (2015).
5. Sukmak, P., Horpibulsuk, S., Shen, S. L., Chindaprasirt, P., and Suksiripattanapong, C.:
Factors influencing strength development in clay±fly ash geopolymer, Construction and
Building Materials 47, 1125-1136 (2013).
6. Chien, V. M., Tuan, L. A., Satomi, T. and Takahashi, H.: Study on effect of chemical com-
position of Geopolymer to improve sludge generated at the disaster sites by Fiber-Cement-
Stabilizes soil method, Advanced Experimental Mechanic 2, 168-173 (2017).
7. Olivia, M. and Nikraz, H.: Properties of fly ash geopolymer concrete designed by Taguchi
method, Materials & Design 36, 191-198 (2012).
Granular flows through a model-scale forest: Influence of
tree density and implications for landslide mitigation

H.T.Luong1,2, J. Baker2 and I. Einav2

1
Hanoi Architectural University, Km10 NguyenTrai, Thanh Xuan, Hanoi, Vietnam
2
The University of Sydney, Darlington, NSW 2006, Australia
luonghanghd@gmail.com

Abstract. Landslides pose a severe threat to the man-made and natural envi-
ronment. As a result, many different artificial structures have been employed to
attempt to mitigate the damage from these hazards. However, natural systems,
such as forests of trees, can also play a defensive role by dissipating energy and
reducing flow speed. This study uses a small-scale model forest to investigate
the effectiveness of such sustainable measures. It examines how granular flows
interact with arrays of tree structures at different densities, comparing the re-
sults to flow down a bare plane. Using image analysis of high-speed camera re-
cordings, we track the lateral spread-out area and the speed of the flows. The
results show that the trees have a strong decelerating effect in the downstream
direction, and also slow the rate of lateral spread out. The knowledge gained
from these experiments can be applied to the field scale, and could provide use-
ful practical guidelines for alternative, sustainable hazard mitigation in moun-
tainous areas.

Keywords: Flow-obstacle Interaction, Hazard Mitigation, Mass Movement,

Granular Flow, Small-scale Experiments.

1 Introduction

Dense granular flows can be observed in a diverse range of industrial and geo-
physical contexts, such as mining and milling operations, slurry pipelines, debris
flows, landslides, sediment transport and snow avalanches. Every year, mass move-
ments in the form of landslides and debris flows cause direct significant damage to
life and property. They can also have serious financial implications for infrastructural
facilities located downslope of potential sources or triggers. These natural hazards
typically occur in mountainous regions with steep, unstable slopes, but predicting ex-
act locations, timings and extent remains a challenge for researchers and practitioners
alike.
There have been numerous studies aimed at understanding the dynamics of such
geophysical granular flows, ranging from field-scale observations [1] to large-scale
experiments [2] , and down to small-scale laboratory experiments [3]. However, these

© Springer Nature Singapore Pte Ltd. 2020 691

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_109
692 H. T. Luong et al.

investigations typically consider empty slopes, planes or chutes that are free from
obstacles or intrusions.
In the majority of real cases, the flow will encounter obstacles in its path, whether
natural topographical features or man-made structures. Of particular significance is
the presence of trees and forests, which may offer ready-made protection against these
hazards by dissipating some of the energy of the flow. However, the precise mecha-
nism of how granular materials flow through forested regions remains unclear and is
often overlooked.
Very severe flow-like events can generate extreme forces and completely cut
through forests, leaving broken and fractured tree stems, overturned roots and torn
branches. Such damage is usually difficult to predict or control, due to the complex
and variable nature of the flows. However, in more moderate shallow flows the trees
can withstand the impact pressure, and the flow tends to migrate through forests with-
out destroying tree trunks. The trees then act as a defensive structure by dissipating
energy from the flow and diverting material in different directions. In most practical
cases there will be some regions of the forest that are severely dam-aged, and other
areas where the trees remain standing. It is therefore conceivable that hazard engi-
neers can utilise natural forests to decelerate flows and minimise their run-out dis-
tance. This would require a detailed understanding of interactions between flows and
forests, and knowledge of how factors such as tree spacing or diameters influence the
behaviour.
This paper presents a comprehensive set of experiments to investigate the behav-
iour of unconfined flows through a small-scale model forest, focusing on how materi-
al spreads out and moves downslope over time for a range of inflow conditions and
tree spacing. It begins with a description of the experimental setup and measurement
methods in Section 2, followed by a presentation of the results in Section 3. In Section
4 we discuss the implications of this further, before closing with a summary and con-
clusions in Section 5.

2 Experimental Methods

The laboratory setup (Fig. 1) consists of a 600 mm wide and 900 mm long cork plane
WKDWFDQEHLQFOLQHGIURPDQDQJOHș 0 up to a maximum of 400 to the horizontal.
Attached to the plane is a regular hexagonal arrangement of cylindrical trees, each of
radius r =10 mm, with a height of 50 mm chosen so that the flow does not overtop the
trees [4, 5] . The trees are fixed normally to the plane, with the uniform spacing S
between their centres varied between different runs. Tests are also carried out in the
DEVHQFHRIWUHHVLH6 ’
Granular flows through a model-scale forest … 693

Fig. 1. Schematic of the laboratory setup, showing the inclined plane (main image) and ar-
rangement of trees (insert).

A reservoir is attached to the upper region of the plane, which has a double gate
mechanism to allow precise control over the release of granular material. This reser-
voir has a fixed width of W = 97 mm, which is much smaller than the width of the
plane to ensure that material is free to spread out in the horizontal directions. It also
has a variable height H ranging from 13 mm to 43 mm.
For all experiments the reservoir is filled with spherical glass beads of mean diam-
HWHUG PPDQGSDUWLFOHGHQVLW\ȡ NJP-3. Note that using these mono-
dispersed glass beads prevents the formation of fingering [6, 7], which can also have
important implications for hazard mitigation but will not be discussed here. To ensure
a steady release of material for the duration of the tests, the reservoir is filled with a
sufficient mass of grains, in this case approximately 5.0 kg of the glass beads.
A high-speed camera is used to record the progression of the flow from above after
the initial release of the gate, at a rate of 100 fps.

These images are then processed using ImageJ and custom Python code to:
i. Threshold and binarize into regions containing granular material and empty re-
gions.
ii. Calculate the total area A of granular material for each image.
iii. Subtract subsequent images, threshold and binarize into spreading and non-
spreading regions.
iv. Calculate the downslope position x of the centroid of the spreading region for each
image.
694 H. T. Luong et al.

3 Experimental Results

3.1 Spread-out area of flows

Fig. 2 shows plots of the total flow area A as a function of time t for different gate
heights. In all cases it can be seen that the area increases approximately linearly with
time, except towards the end of the experiments when the hopper starts to empty. The
constant area growth rates ‫ܣ‬ሶ are given by the gradients of the lines of best fit, which
are increasing functions of the release height, meaning that thicker flows have a ten-
dency to spread out faster.
There are significant differences between the results with trees spaced at 40 mm
(S= 40mm, Figure 2a) and those on an empty plane (S = ’, Figure 2b). For all the
different gate heights we find that the material spreads out considerably less in the
presence of trees, as evidenced by the steeper gradients on Fig. 2b compared to 2a.
Other tree spacings investigated show a consistent trend, with denser forests slowing
down the spread-out-rate more than less-dense setups. Slope angle is also found to
play a role, and we observe flows spread out faster on steeper terrain.

3.2 Centroid velocity of flowing region

Another measure of the potential impact of the flow is the velocity at which material
moves downslope. Here we calculate the region of material that moves between sub-
sequent images, and then track its downslope centroid position x, at each time t. This
provides a way of comparing different shaped flows, since it takes into account the
whole area, not just the most advanced point like in previous studies [3-5, 8].
Fig. 3 shows plots of centroid position x against time, for the same set of experi-
ments as in Figure 1. We observe similar behaviour, with approximately linear growth
rates showing that the material moves with constant speed v. As before, higher release
gates cause steeper gradients, which now translates to faster flows. The influence of
the trees is again very significant, with flows through the 40mm forest (Fig. 3a) being
strongly slowed down, sometimes travelling up to five times slower in the downslope
direction than those on an empty plane (Fig. 3b).
Granular flows through a model-scale forest … 695

Fig. 2. Plots of the flow area A against
time t for different gate heights H and
a fixed slope angle ș = 280: a) tree
spacing S = 40 mm; b) no-tree region
S=f.
Fig. 3. Plots of the downslope position
x (of moving material centroid) against
time t for different gate heights H. a)
Tree spacing S = 40mm; b) Empty
SODQH 6  ’ 6ORSH DQJOH IL[HG DW T =
28 degree.

4 Discussion

Our experimental results show the considerable effect that forests have on dense
granular flows, in terms of spread-out shape, spread-out velocity and downslope ve-
locity. These could be useful in a practical setting if one can precisely quantify the
LQIOXHQFHRIWKHFRQWUROSDUDPHWHUVLQWKLVFDVHJDWHKHLJKW+VORSHDQJOHșDQGWUHH
spacing S. On-going work is examining the effect of these factors to try and develop a
predictive model.
For geophysical hazards, other flow properties of interest include the flow thick-
ness and the force acting on each tree. It may be possible to measure these directly, or
to infer from existing measurements, for example by using total flux to infer thick-
ness. Our experimental setup was designed so that the forces on the trees were insuf-
ficient to break them or detach from the plane, but this could also be an interesting
avenue to explore.
696 H. T. Luong et al.

5 Concluding remarks

This paper has presented a comprehensive set of experimental results showing how a
mass of granular material flows through a regular array of cylindrical obstacles, or
‘tree’. By examining different density ‘forests’, as well as empty planes, we have
shown the effect that these obstacles have on the flow, finding that, even though they
increase the overall lateral spread out, they reduce the rate at which this spread out
occurs. The trees also have a significant decelerating effect on the flow velocity. The
findings of this paper could be used to analyse the potential impact of flows through
existing forests, or in predicting the effect of reforestation and replanting on reducing
the destructive force of natural flows.

Acknowledgements

This work has benefitted from fruitful discussions with Benjy Marks and François
Guillard regarding the experimental apparatus and image analysis. The authors are
very grateful to Todd Budrodeen and Garry Towell for the design of the reservoir and
the experimental space.

References
1. [1] B. W. McArdell, "Field Measurements of Forces in Debris Flows at the Illgraben: Im-
plications for Channel-Bed Erosion," International Journal of Erosion Control Engineer-
ing, vol. 9, no. 4, pp. 194-198, 2016.
2. [2] C. Johnson, %.RNHODDU50,YHUVRQ0/RJDQ5/D+XVHQDQG-*UD\*UDLQဨ
size segregation and levee formation in geophysical mass flows," Journal of Geophysical
Research: Earth Surface, vol. 117, no. F1, 2012.
3. [3] O. Pouliquen, "Scaling laws in granular flows down rough inclined planes," Physics
of fluids, vol. 11, no. 3, pp. 542-548, 1999.
4. [4] X. Cui and J. Gray, "Gravity-driven granular free-surface flow around a circular cyl-
inder," Journal of Fluid Mechanics, vol. 720, pp. 314-337, 2013.
5. [5] C. Ng et al., "Physical modeling of baffles influence on landslide debris mobility,"
Landslides, vol. 12, no. 1, pp. 1-18, 2015.
6. [6] O. Pouliquen, J. Delour, and S. B. Savage, "Fingering in granular flows," Nature, vol.
386, no. 6627, p. 816, 1997.
7. [7] J. Baker, Gray, N., Kokelaar, P., "Particle Size-Segregation and Spontaneous Levee
Formation in Geophysical Granular Flows " International Journal of Erosion Control En-
gineering vol. 9, no. 4, pp. 174-178, 2016.
8. [8] T. Faug, P. Childs, E. Wyburn, and I. Einav, "Standing jumps in shallow granular
flows down smooth inclines," Physics of Fluids, vol. 27, no. 7, p. 073304, 2015.
 Dynamic properties of loose sand using numerical
analysis-A case at Hong Thai Tay coal transportation
road project (Vietnam)

Nguyen Phi Hung1, Nguyen Van Dung1 and Tan Do1*

1
Hanoi University of Mining and Geology, Duc Thang, Bac Tu Liem, Ha Noi, Vietnam
domanhtan@khoaxaydung.edu.vn

Abstract. In this paper, a study was conducted to investigate the dynamic be-
haviors of loose sand collected from Hong Thai Tay coal transportation road
project (Vietnam). Cyclic direct simple shear tests (CDSS) were performed on
the soil using a finite element method program. The CDSS test was controlled
under drained strain condition with the vertical effective consolidation stress of
100 kPa. In addition, the other dynamic control parameter, cyclic stress ratio,
was subjected to vary in the range of 0.04-0.1. A total of 30 cyclic direct simple
shear tests were performed with respect to the number of uniform cyclic cycle
and cyclic stress ratio to determine the dynamic parameters of the sand plastici-
ty model. Eventually, the dynamic properties of the investigated sand were ob-
served through various dynamic aspects. As a result, the strain amplitude of the
loose sand in this study slightly increased within the first fifteen-loading cycles.
From the sixteenth cycle, huge increments were observed during the cyclic test.
This result agreed well with that of excess pore water pressure ratio, that is, the
development of excess pore water pressure reached about 1.0 after fifteen cy-
cles of loading. Moreover, the obtained results of the shear modulus reduction
ratio from this study were compared to those of sand from the literature re-
views.

Keywords: Dynamic properties, Loose sand, Numerical analysis, Hong Thai

Tay Project.

1 Introduction

Hong Thai Tay coal transportation road is one of the very important projects in the
development strategy of Dong Bac corp-port company. It is used as the main coal
transport route from mining areas to Hong Thai Port (e.g., Trang Bach coal mine,
Coal mine numbers 91, 618, and 397). From the site investigation, a thick loose sand
layer (i.e., up to 3 meters) was observed near the underpass through highway 18A of
Hong Thai Tay coal transportation road. As well-known, loose sand is susceptible to
liquefaction during the dynamic (e.g., earthquake) and cyclic (e.g., truck movement)
excitations. The liquefaction phenomenon is most often observed in saturated, loose

© Springer Nature Singapore Pte Ltd. 2020 697

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_110
698 N. P. Hung et al.

(low density or uncompacted), sandy soils because loose sand has a tendency to com-
press when a load is applied.
Many researchers have addressed the dynamic properties of sand using mathemati-
cal solutions [1, 2] and numerical analysis [3-5]. In this study, the Soil-Test option in
a Finite Element Method program, Plaxis 2D, was adopted to simulate the cyclic di-
rect shear test (CDSS). The CDSS test was controlled under drained strain condition
with the vertical effective consolidation stress of 100 kPa. In addition, the other dy-
namic control parameter, cyclic stress ratio, was subjected to vary in the range of
0.04-0.1. A total of 30 cyclic direct simple shear tests were performed with respect to
the number of uniform cyclic cycle and cyclic stress ratio to determine the dynamic
parameters of the sand plasticity model.

2 Methodology

2.1 Selection of material

The soil used for the simulation in this study was the loose sand from Hong Thai Tay
coal transportation road (Quangninh, Vietnam). The sand is classified as silty sand
(SM). Soil classification parameters based on the Unified Soil Classification System
(USCS). All physical properties of the sand, such as its unit weight, natural water
content, fine particle, and specific gravity, are tabulated in Table 1.

Table 1. Physical properties of the loose sand

Properties Loose sand

Unit weight (tons/m3) 1.90
Natural water content (%) 26.77
Specific gravity 2.69
D10 (μm) 5
3DUWLFOHVȝP 13.98
Color Yellow

2.2 Constitutive model used in the numerical analysis

PM4Sand model is the elasto-plastic, critical state compatible, bounding surface plas-
ticity model. It is a very attractive model for the industry due to a small number of
parameters to be calibrated. They are mostly related to the commonly available data
in the design practice [6]. It originally developed from the Dafalias-Manzari model
[7, 8] with substantial improvements made by Boulanger and Ziotopoulou [9]. There
are various inherent advantages of using PM4Sand model for the evaluation of dy-
namic properties of sand (e.g., good approximation of empirical correlations used in
practice including the post-liquefaction settlements, accurate stress-strain and pore
pressure build-up simulations, accurate simulation of the accumulation of shear strain
and strength degradation, easy prediction of number of loading cycles to liquefaction)
Dynamic properties of loose sand using numerical … 699

[10]. The input parameters used in the numerical analyses are detailed in Table 2. The
procedure for input parameters calibration is presented clearly in Plaxis Manual [6].

Table 2. Input parameters for numerical analysis

Soil types Loose sand
Site Quangninh, Vietnam
Unit weight (kN/m3) 26.6
Friction angle (degrees) 12
Cohesion (kPa) 13.1
Elastic modulus (Mpa) 6.0
Shear modulus coefficient (Mpa) 425.77
Relative density (%) 29
Contraction rate parameter hp0 0.05

3 Results and analysis

3.1 Dynamic properties from CDSS

The results from CDSS including shear stress-shear strain, effective normal stress-
shear stress, number of uniform cycles-shear strain, and the number of uniform cy-
cles-excess pore water pressure ratio are shown in Fig. 1. During cyclic loadings, the
shear strain continuously fluctuated with respect to uniform cycles and finally reached
the maximum value of 11.68 %. However, the fluctuation increased pronouncedly
after the fifteenth cycles (See Fig. 1a). This result agreed well with that of excess pore
water pressure ratio, that is, the development of excess pore water pressure reached
about 1.0 after fifteen cycles of loading (See Fig. 1b). It should be noted that the liq-
uefaction of sand arises after the pore water pressure ratio reaches 1.0. As well-
known, the nature of the soil behavior changes after the initial liquefaction reaches
(i.e., pore water pressure ratio reaches 1.0). These changes are exhibited in the shapes
of the stress-strain loops (Fig. 1c) the stress path (Fig. 1d). These changes are also
observed in Fig. 1c and Fig. 1d after 15 cycles. In particular, after the fifteenth cycle,
the stress path approached and then crossed the phase transformation line. This find-
ing can be explained by the concept of CDSS test. When an element of such soil is
loaded cyclically, it exhibits a tendency to contract or compress. Under saturated con-
ditions, this tendency for contraction results in an increase in pore water pressure. The
effective stress, therefore, decreases and the soil becomes softer. Note that the effec-
tive normal stress decreases with each cycle (see Fig. 1d), and the tangent shear mod-
ulus (i.e., the slope of the stress-strain curve in Fig. 1c), reduced as the effective nor-
mal stress diminished [11]. This is the reason for the dramatic increase in shear strain
after the initial liquefaction.
700 N. P. Hung et al.

(a) Shear strain under cyclic loading (b) Excess pore water pressure under cyclic
loading

(c) Shear stress vs shear strain under cyclic (d) Shear stress vs effective normal stress
loading under cyclic loading
Fig. 1. Dynamic properties of the loose sand of CDSS

Fig. 2. Cyclic stress ratio vs number of uniform cycles as compared to Boulanger and
Ziotopoulou [9]
Dynamic properties of loose sand using numerical … 701

The relationship between the cyclic stress ratio and the number of uniform cycles is
plotted in Fig. 2. As shown, the cyclic stress ratio decreases with an increase in the
number of uniform cycles. In this sense, larger loading amplitudes (i.e., involving the
cyclic stress ratio) produced the initial liquefaction in smaller numbers of cycles than
lower loading amplitudes. This finding agreed well with that reported by Boulanger
and Ziotopoulou [9].

3.2 Shear modulus reduction ratio curves

In this study, shear modulus reduction-shear strain curves with respect to the vertical
effective stresses (i.e., 100 kPa, 500 kPa, 1000 kPa, 1500 kPa, and 2000 kPa) were
built up based on the hysteresis loops obtained by CDSS tests (Fig. 3). As shown,
shear modulus almost keeps constant with small shear strain but it starts going down
with large strain (i.e., larger than 0.001%), regardless of the vertical effective stress
values. In addition, shear modulus reduction ratio was relatively sensitive to the verti-
cal effective stress. Expectedly, the slope of the tangent of shear modulus ratio curves
increased as an increase in vertical effective stress. Eventually, it is interesting to note
that all results from this study were in between the lower and upper bounds of Seed,
Wong Robert [12], regardless of vertical effective stress.

Fig. 3. Shear modulus reduction ratio curves of the studied sand as compared to
Seed, Wong Robert [12]

4 Conclusions

The result of the shear strain obtained from CDSS agreed well with that of the pore
water pressure ratio. The shear strain started increasing considerably and the devel-
opment of excess pore water pressure reached about 1.0 after the same uniform cycles
(i.e., fifteen cycles of loading). At this point, the shapes of the stress-strain loops also
changed, systematically.
The relationship between the cyclic stress ratio and the number of uniform cycles
was also clarified in this study. An increase in cyclic stress ratio resulted a decrease in
702 N. P. Hung et al.

the uniform cycles. In other words, the higher cyclic stress ratios would cause the
initial liquefaction in smaller numbers of cycles than lower cyclic stress ratios.
The shear modulus reduction curves were also built up in this study based on the
hysteresis loops. As an agreement with previous studies, a strong effect of the effec-
tive normal stress on the shear modulus reduction ratio was observed. The shear mod-
ulus of the soil reduced in a slower rate with the higher vertical effective stress.

Acknowledgement

This work was supported by Hanoi University of Mining and Geology and Dong Bac
corp-port company.

References

1. Chen, W. and Saleeb, Constitutive equations for engineering materials Vol. 1

plasticity and modeling. 1994.
2. Chen, W.-F.J.E., Constitutive equations for engineering materials Vol. 2 plasticity
and modeling. 1994.
3. Dinesh, S., S. Thallak, and J.S. Vinod, Dynamic properties and liquefaction behavior
of granular materials using discrete element method. Vol. 87. 2004.
4. Sitharam, T. and S.J.P.I.A.S. Dinesh, Numerical simulation of liquefaction behaviour
of granular materials using Discrete Element Method. 2003. 112(3): p. 479-484.
5. Shahnazari, H., Y. Dehnavi, and A.H. Alavi, Numerical modeling of stress±strain
behavior of sand under cyclic loading. Engineering Geology, 2010. 116(1): p. 53-72.
6. Plaxis Manual, Plaxis The PM4S model 2018. 2018.
7. F. Dafalias, Y. and M. T. Manzari, A Critical State Two-Surface Plasticity Model for
Sands. Vol. 47. 1997. 255-272.
8. Dafalias Yannis, F. and T. Manzari Majid, Simple Plasticity Sand Model Accounting
for Fabric Change Effects. Journal of Engineering Mechanics, 2004. 130(6): p. 622-
634.
9. Boulanger, R.W. and K. Ziotopoulou, A Sand Plasticity Model for Earthquake
Engineering Applications. 2015.
10. Vilhar, G., et al., Implementation, Validation, and Application of PM4Sand Model in
PLAXIS. Geotechnical Earthquake Engineering and Soil Dynamics 2018.
11. Arduino, P., et al., Dynamic stiffness of piles in liquefiable soils. 2002.
12. Seed, H.B., et al., Moduli and Damping Factors for Dynamic Analyses of
Cohesionless Soils. Journal of Geotechnical Engineering, 1986. 112(11): p. 1016-
1032.
 Application of Box-Behnken Design in Optimization of
Fracture Treatment Design for Lower Oligocene
Reservoir, Offshore Viet Nam
Huu Truong Nguyen1*, W. Bae2), Van Hung Nguyen1)
1
Petrovietnam University, Ba Ria ± Vung Tau, Vietnam
2)
Sejong University, Gwangjin-ku, Seoul, Korea
truongnh@pvu.edu.vn, wsbae@sejong.ac.kr, hungnv@pvu.edu.vn

Abstract. The paper presents an approach for optimizing the design of hydrau-
lic fracture treatment in the low permeability lower Oligocene sandstone reser-
voir in the Cuu Long Basin, offshore Viet Nam. Several design variables in-
cluding low behavior index, leak-off coefficient, pumping time, injection rate
and proppant, which are generally affecting the fracture mechanics, were con-
sidered in the optimization process in order to improve oil production. The in-
fluence of those variables and interaction variables on fracture conductivity and
net present value (NPV), hence, the production performance was analyzed us-
ing Box-Behnken design±based statistical modeling, which are experimental
designs for response surface methodology. The results show that the maximum
NPV was 50.5 million USD from optimal variables including injection rate of
36bpm, injection time of 78 minutes, slurry concentration end of the job of
9.1ppg and the leak-off coefficient of 0.0034ft/min0.5.
Key words: Optimum variable Conditions, Improved Oil Production, Lower
Oligocene reservoir, Net Present value, Box-Behnken design
1. Introduction

In the last decades, hydraulic fracturing has been widely used to enhance oil and gas
production. Haiqing Yu and M.M.Rahman [1] presented hydraulic fracturing for tight
gas sandstone reservoir, S. Jain et al., [2] reviewed case study of hydraulic fracturing
in Bach Ho field. However, the production performance from fractured wells was not
optimized due to using unchanged fracturing parameters. The authors think that an
efficient fracture treatment should be designed based on a surface parameter, which is
either maximum fracture conductivity or maximum net present value (NPV). Meng
and Brown, [3]; Aggour and Economides, [4]; Langedijk et al., [5] used a
maximization net present value as treatment design object. The NPV was obtained
from sensitivity analysis with the number of fracture treatments and fracture length.
Mohaghegh et al., [6] proposed a hybrid neuro-genetic for an optimal fracture
treatment design based on recorded data. Aggour [7] published a procedure for
fracture treatment design optimization in high permeability reservoir with NPV
objective by using gradient method to optimize of nonlinear issue. Economides et al.,
[8] presented a unified fracture design optimization, which is an excellent study, but
lacks formulation design constraints. Thus, in this paper, the authors have proposed
new study on predicted optimum variables condition using the box-Behnken design
and response surface methodology with maximum NPV objective based on PKN-C
model in the lower Oligocene sandstone reservoir.
2. Method used

© Springer Nature Singapore Pte Ltd. 2020 703

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_111
704 H. T. Nguyen et al.

2.1. Optimization tool

Response surface methodology (RSM) generally search for the correlation between
several variables and one or more response variables [9]. This statistical method was
originally developed by George E. P. Box and K. B. Wilson [10]. The main feature of
RSM is to consider a sequence of designed experiments to obtain an optimization of
response. Box-Behnken design is a class of rotatable or nearly rotatable second-order
designs based on three-level incomplete factorial designs [11]. In addition, the study
uses the Box-Behnken design for four operating parameters. As a result, the number
of experiments (N) required for the development of BBD could be found as equation
N= 2k(kí1) +C0, (where k is number of factors and C0 is the number of center points).

2.2. Fracture models

Most of existing fracture models used for designing hydraulic fracturing treatment
have been considered the fracture geometry prediction as function of different treat-
ment parameters. The fracture geometry is generally defined by half-length, height
and width of fracture, which could be in two-dimensional space or three-dimensional
space. However, for a hydraulic fracturing in the field, those models are sometimes
not be able to provide accurate prediction of the fracture geometry because of several
issues such as volume of fluid loss and spurt lost through the fracture area. For operat-
ing hydraulic fracturing in the tight oil as Oligocene reservoir, the study in this paper
used the maximum fracture width of the PKN-C model published in [12], which is
also given in Equation (1). The calculation of the maximum fracture width accounts
for leak-off coefficient, power law parameters and spurt lost, rock properties, injec-
tion rate and injection time. As for the fracture half-length/fracture height, the Carter
II solution for material balance was considered, which is expressed in terms of rate,
time and power law parameters:
n 1

 1+2.14n  2n+2 2n+2  qi n h f x f  2n+2

1 n 1 1-m

  (1)
 
2n+2 2n+2
w f =9.15 3.98 K ( )
 n   2 E' 
2.3. Integrated model

In-situ stress Free design Fracture model

-Minimum variables -Fracture length Material
Horizontal Stress -BBD and RSM -Maximum width Balance
-Fluid
volume
NPV Production Conductivity Model used
model Model -Dimensionless Fracture -Proppant
-Transient flow conductivity, Sf used

Fig. 1. Integration model study

The fig.1. presents the integration of the hydraulic fracturing and economic analysis
of post fractured well with the objective lower Oligocene reservoir.
Application of Box-Behnken Design in Optimization … 705

2.4. Net present value model

The net present value NPV can be calculated as follows:

N
Rn
NPV= ¦ n
-C tr ; Where: Rn is the revenue at year n, i is the interest rate (%),
n=1 (1+i)
Ctr is the total treatment cost, $.

3. Input data

3.1. Design variables

Free design variables consist of flow behavior index (n), leak-off coefficient, Cl
(ft/min0.5), pumping time, ti (minutes), injection rate, q (bpm), and proppant end of the
job, Pc (ppg). Ideally, the first five variables are designed on the surface because those
variables can be changed if the parameters can achieve good fracture treatment per-
formance of a fractured well based on either maximum net present value or high pro-
duction performance. Based on the field experience, the design parameters on the
surface are ranged within lower and upper bounds as follows:
”Ti ”ILHOGH[SHULHQFH
”W”ILHOGH[SHULHQFH
”3c ”LQGXVWU\SUDFWLFH0HQJDQG%URZQ) [3]
0.2 ”Q”; field experience
”Cl ”ILHOGH[SHULHQFH

3.2. Application of the lower Oliogcene reservoir

The table 1 shows the upper bound and lower bound to the center from the variables
with the Box- Behnken design for 25 runs as shown in the table 3.

Table 1. Range of four independent variables and their levels for BBD

Coded variables level

Low Center High
Variables Symbol -1 0 1
Pump rate, bpm X1 30 35 40
Injection time, min X2 60 75 90
EOJ, ppg X3 8 9 10
Leak-off coefficient, ft/min0.5 X4 0.003 0.005 0.007
706 H. T. Nguyen et al.

Table 2. Reservoir and well data

Parameters Value
Target fracturing depth, ft 11,482
Wellbore radius, ft. 0.25
Reservoir height, ft. 77
Reservoir porosity, % 12.8
Reservoir permeability, md 2
Reservoir fluid viscosity, cp 0.5
Oil formation volume factor, RB/STB 1.4
Oil saturation, % 48
Total compressibility, psi-1 1.00 ×10-5
Young modulus, psi 5×106
Sandstone Poisson Ratio 0.25
Initial reservoir pressure, psi 5610
Reservoir temperature, 0F 260
Oil API 40
Gas specific gravity 0.707
Bubble point pressure, psi 1,310
Flowing bottom hole pressure, psi 3,500
Closure pressure, psi 9,137
Consistency index, , K, lbf.sn/ft2 0.04
Proppant cost ($/lbm) 0.4
Fluid cost ($/gallon) 1
Interest rate, % 10
Pump cost ($/hr/HHP) 3.25
Fixed cost, $ 15000
Oil price, $/bbl 60
Fracturing fluid type: Fracturing Lighting 3500 (LF-3500) consists of clay treatment- 3C of
1ppg, clay master 5C of 1pptg, inflo-250 of 2pptg, BF-7L (cross linking Buffer) of 2pptg,
XLW-30 (Cross linker) of 3pptg, GBW-41L(Breaker) of 2pptg, Hiperm CRB (Encapsulated
Breaker) of 1ppg, XLFC-5C (Gel Liquid Concentrate) of 8.75pptg for fracturing fluid system
3URSSDQWW\SHVL]HG&$5%2-/LWHZLWKGHQVLW\RIOEIW,63&RQGXFWLYLW\GDPDJHG
IDFWRURI

The well data used to simulate oil production from fractured well and predicts the
NPV based on the table 2.

4. Results and discussion

Fig. 2. The levels of variables effects on Fig. 3. Response surface plots to present of
NPV NPV
Application of Box-Behnken Design in Optimization … 707

Table 3. Independent variables and results from conductivity and NPV observed BBD

Coded level of those Actual of those level variables

variables
Run Rate,
X1 X2 X3 X4 bpm
1 -1 -1 0 0 30 60 9 0.005 1002.1 39.23
2 1 -1 0 0 40 60 9 0.005 1242 42.94
3 -1 1 0 0 30 90 9 0.005 1160 41.74
4 1 1 0 0 40 90 9 0.005 1444.8 45.72
5 0 0 -1 -1 35 75 8 0.003 1551.7 47.04
6 0 0 1 -1 35 75 10 0.003 1755.5 49.37
7 0 0 -1 1 35 75 8 0.007 931.12 37.98
8 0 0 1 1 35 75 10 0.007 1044.8 39.93
9 -1 0 0 -1 30 75 9 0.003 1471.9 46
10 1 0 0 -1 40 75 9 0.003 1836.7 50.26
11 -1 0 0 1 30 75 9 0.007 881.7 37.12
12 1 0 0 1 40 75 9 0.007 1093.9 40.7
13 0 -1 -1 0 35 60 8 0.005 1056.9 40.09
14 0 1 -1 0 35 90 8 0.005 1225.2 42.64
15 0 -1 1 0 35 60 10 0.005 1187.4 42.11
16 0 1 1 0 35 90 10 0.005 1379.2 44.79
17 -1 0 -1 0 30 75 8 0.005 1021.7 39.49
18 1 0 -1 0 40 75 8 0.005 1267.6 43.24
19 -1 0 1 0 30 75 10 0.005 1147.4 41.44
20 1 0 1 0 40 75 10 0.005 1427.7 45.45
21 0 -1 0 -1 35 60 9 0.003 1514 46.55
22 0 1 0 -1 35 90 9 0.003 1782.4 49.64
23 0 -1 0 1 35 60 9 0.007 914.51 37.67
24 0 1 0 1 35 90 9 0.007 1055.2 40
25 0 0 0 0 35 75 9 0.005 1219.4 42.52

The correlation between variables and interaction variables on NPV as follows:

NPV=48.32+1.69333X1 +1.32583X 2 +0.800832X3 -4.28833X 4 -2.43542X12 -3.83667X 22
(2)
-2.15167X32 -2.69042X 42 +0.075X1X 2 +0.065X1X3 +0.08X1X 4 +0.0325X 2 X3 -0.19X 2 X 4
+0.155X3X 4
Figure 2 can be divided in two regions, including the region below to zero and the
region above to zero. In detail, the first region shows the factors of these variables and
interaction variable are below zero including those for X1.X1, X2.X2, X3.X3, X4.X4,
and X2.X4 for predicting the decrease level of the NPV when these variables change.
The second region is above zero, where the factors of the variables and interaction
variables present the positive coefficients namely X1, X2, X3, X1.X2, X1.X3, X1.X4
and X2.X3, which define the NPV level to analyze the Pareto chart in the fig 2. To
present the effect plots, the modelling design of experiment software 9.1 is used to
analyze these coefficients levels of the variables and interaction variable effects on
the NPV as the fig.2. Figure 3 presents the contour plots and build the relationship
between main variables and interaction variables on the NPV as seen in the equation 1
with R2 of 0.913 and R2adjust of 0.8. Figure 3 found the best prediction of the varia-
bles when maximizing the NPV was on the red smallest region of which the maxi-
mum NPV reached 50.5 $mm. The prediction of optimal variables was the injection
708 H. T. Nguyen et al.

rate of 36bpm, injection time of 78 minutes, slurry concentration end of the job of
9.1ppg and the leak-off coefficient of 0.0034ft/min0.5.

5. Conclusions
An optimization approach has been proposed in this paper for improving the design of
hydraulic fracture treatment in the lower Oligocene sandstone reservoir in the Cuu
Long Basin, offshore Viet Nam. The BBD and RSM with maximum NPV objective
based on PKN-C model were considered. The following remarks could be drawn from
the calculation:
- When the leak-off coefficient variable is increased, the fracture geometry is pro-
duced lower as well as NPV decrease.
- Those variations of injection rate, injection time and proppant concentration end
of the job are increased, the fracture geometry is higher as NPV increase.
- Those factors of the interaction variables and main variables are positive which
lead to increase NPV.

References
1. Haiqing Yu and M.M.Rahman.: Pinpoint Multistage Fracturing of Tight Gas Sands- An
Integrated Model with Constraints. SPE 152439, SPE Middle East Unconventional Gas
Conference and Exhibition held in Abu Dhabi, UAE (2012)
2. S. Jain, A. Prestridge, P. Dellorusso, and N. C. Nghi, Schlumberger and D.D. Lam and
V.Q. Hung.: Case Study from 12 Successful Years of High Temperature Fracturing in
Bach Ho Field Offshore Vietnam. SPE 106712, SPE Production and Operations Symposi-
um held in Oklahoma City, Oklahoma, U.S.A (2007)
3. Meng, H.-Z., and Brown, K. E.: Coupling of production forecasting, fracture geometry
requirements and treatment scheduling in the optimum hydraulic fracture design. SPE
16435, Low Permeability Reservoirs Symposium, Denver, CO (1987)
4. Aggour, T. M., and Economides, M. J.: Optimization of the performance of high-
permeability fractured wells. SPE 39474, International Symposium on Formation Damage
Control, Lafayette, LA (1998)
5. Langedijk, R. A., Al-Naabi, S., Al-Lawati, H., Pongratz, R., Elia, M. P., and Abdulrab, T.:
Optimization of hydraulic fracturing in a deep,multilayered, gas-condensate reservoir. SPE
63109, Annual Technical Conference and Exhibition, Dallas, TX (2000)
6. Mohaghegh, S., Balan, B., and Ameri, S.: A hybrid, neuro-genetic approach to hydraulic
fracture treatment design and optimization. SPE 36602, Annual Technical Conference and
Exchange, Denver, CO (1996)
7. Aggour, T.: Optimization strategies for hydraulic fractures in high permeability reservoirs.
SPE 68131, Middle East Oil Show, Bahrain, March 17±20 (2001)
8. Economides, M. J., Oligney, R., and Valko, P.: Unified Fracture Design. Alvin, TX: Orsa
Press (2002)
9. Cornell, J.A.: How to Apply Response Surface Methodology. 2nd Edn., American Society
for Quality Control, Milwaukee, WI (1990)
10. Box, G. E. P., & Wilson, K. B.: On the experimental attainment of optimum conditions.
Journal of the Royal Statistical Society. Series A (General), 13, 1-45 (1951)
11. Box, G.E.P. and J.S. Hunter.:. Multi-factor experimental designs for exploring response
surfaces. Ann. Math. Stat., 28(1): 195-241 (1957)
12. Valkó, P.P., and Economides, M.J.: Hydraulic Fracture Mechanics. John Wiley & Sons,
Chichester, England (1995)
 Experimental and numerical assessments of seepage effect
on embankment behaviours by the time

Thanh Quang Nguyen1, Truong Linh Chau1 and The Hung Nguyen1
1
University of Science and Technology, Danang University, VietNam
nguyenthanhquang1987@gmail.com

Abstract. The incidents of dam failure occur due to some different reasons, in
which the reason for seepage accounts for 35%. The seepage leading to the dam
incidents can be divided into piping, internal erosion, solution of soluble rock,
excessive internal pressures and/or saturation and excessive uplift, heave, or
blowout. Currently, empirical method is commonly used to study the influence
of seepage on the dam incidents. This paper presents a research work on the de-
velopment of the experimental equipment and procedure for evaluating seepage
effects on several physical properties of embankment, which include cohesion,
friction angle, density and permeability coefficient. The testing results provide
important material inputs to the finite element model used for the stability anal-
ysis of the main dam of Ta Trach reservoir (Thua Thien Hue Province, Vi-
etnam) caused by the seepage phenomenon.

Keywords: Dam failure, Earth dam, Seepage, Physical properties of embank-

ment.

1 Introduction

Upon the development of irrigation works, the incidents related to earth dams also
increased significantly, causing large material losses. Therefore, it is more important
to take measures to prevent and mitigate these unsafe incidents.
Seepage always occurs in the dams and if its magnitude is within design limits, it
may not harm the stability of the dam. Seepage depends on several factors, including
permeability of the soil and the pressure gradient, essentially the combination of forc-
es acting on water through gravity and other factors. Seepage causing problems that
can lead to dam failure are classified as follows [1]: piping, internal erosion,
solutioning of soluble rock, excessive internal pressures and/or saturation and exces-
sive uplift, heave, or blowout.
Currently, empirical method is commonly used to study the influence of seepage
on the dam incidents. In order to better understand the effect of seepage, numerous
physical models have been developed and tested by different research groups. For
example: such as Awa et al. [2] used model test of a dam to study piping failure of
natural dam; Bhattarai et al. [3] conducted some model test to investigate the dynamic
lateral widening process and breach characteristics of a river dyke; and Ke et al. [4]

© Springer Nature Singapore Pte Ltd. 2020 709

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_112
710 T. Q. Nguyen et al.

carried out a series of seepage tests to evaluate the characteristics of suffusion and its
mechanical consequences on saturated cohesionless soil. Fig. 1 briefly shows these
physical models used for studying the influence of seepage. The observation, meas-
urement, collection and analysis of data will certainly provide important input to the
assessment of the influence of seepage on the dam stability.

(a) (b)

Fig. 1. Some physical models studying the

influence of seepage
(a) Study on piping failure of natural dam [2]
(b) Experimental study on river dyke breach
characteristics due to overtopping [3]
(c) Experimental investigations on suffusion
characteristics and its mechanical consequences
on saturated cohesionless soil [4]
(c)
This paper mainly focuses on development of laboratory testing equipment for inves-
tigating potential effects of seepage on some important physical properties of embank-
ment such as cohesion, friction angle, density and permeability coefficient. The experi-
mental results were then considered in the stability analysis of the main dam of Ta
Trach reservoir (Thua Thien Hue Province, Vietnam) using the commercial finite ele-
ment program Plaxis 2D.

2 Testing equipment of seepage phenomenon

Several experimental principles and construction of triaxial compressor have been

evaluated and applied to develop the testing equipment, which is able to approximate-
ly simulate the influence of seepage on the embankment materials during reservoir
operation.

Fig. 2. Testing equipment of seepage effect on physi- Fig. 3. The structure of each cell
cal properties of embankment
Experimental and numerical assessments of seepage … 711

Fig. 2 illustrates some main features of the testing device including flow pump, au-
tomatic pump control system, pressure tank, cell pressure, seepage pressure, etc.
Some other items such as vacuum split mold, latex membranes, etc, are aslo presented
in the figure. Fig. 3 presents a sketch of the structure diagram of each single cell of the
testing equipment of seepage effect. The cell accommodates a specimen of 0.035 m-in-
diameter and 0.070 m-in-length. The increments of inflow rate system consist of a 15-
35atm flow pump and a pressure tank applied to the specimen through an automatic
pump control system. The seepage flow moves through the specimen by two pores in
two metal sample caps (see Fig. 3 7KH ÀRZ WXEHV DUH GHVLJQDWHG DV  m-in-
diameter. During the experiment, the seepage pressure is always in accordance with the
calculated value in the experimental investigations.
To determine the changes of some physical properties of cohesive soils by time
cause by the seepage, the specimen is installed into Geocomp triaxial testing equip-
ment (see Fig. 5). During operation, all the water supply to the pump and the water output
(through the water filter system) go into the water cell, ensuring that the system is closed.

Fig. 4. Installing soil specimens into Fig. 5. Geocomp triaxial testing equipment in the lab
the equipment of Da Nang University of Science & Technology
By using this testing equipment in combination with other experiments, we can de-
termine the changes of physical properties of embankment due to seepage effect by
the time including cohesion (C); friction angle ijGHQVLW\ȖDQGpermeability coef-
ficient (Kt).

3 Experimental investigations

The soil is exploited in the 5th expansion zone of the VD2 'UDZLQJ5HI1ƒĈ-
Ĉ&-VL-·1ƒĈ-Ĉ&-VL-10) [5] in the downstream of the VD2 ± Ta Trach res-
ervoir in Thua Thien Hue Province (see Fig. 6).

Fig. 6. The location of taking

specimens (layer 2B is ex-
ploited in the 5th expansion
zone of the VD2) ± Ta Trach
reservoir, Thua Thien Hue
Province, Vietnam

The degree of compaction of the test specimens is 0.97. All specimens used in this
experiment were first fully saturated, creating the consistency of the initial humidity
712 T. Q. Nguyen et al.

or saturation. Soil specimens are considered to be fully saturated when pore water
pressure coefficient B reaches a value greater than or equal to 0.95 (Head, 1986 [6]).
After finishing the saturation, the specimens are consolidated under cell pressure and
seepage pressure, in other words, the soil specimens are isotropically consolidated by
effective pressure. The value of effective pressure is calculated according to the effec-
tive stress principle applied to the main dam of Ta Trach reservoir, Thua Thien Hue
Province.
The cell pressures determined for each cell: 0.5 ız; 1.0 ız; 2.0 ız with the actual
conditions are 1.6 atm - 3.2 atm - 6.4 atm respectively. And the seepage pressures
determined for each cell: 0.5 ıt; 1.0 ıt; 2.0 ıt with the actual conditions are 0.85 atm -
1.7 atm - 3.4 atm respectively. The experiment is conducted under two conditions
including actual conditions (case 1) and conditions of increasing the seepage pressure
- doubling the seepage pressure (case 2).
The changes of physical properties of embankment by time due to seepage effect
are determined at at T = 5, 15, 30, 45, 60, 90 days.

4 Test results

The measurement results obtained from the testing equipment allow to determine
several material parameters including cohesion (C) ± IULFWLRQDQJOHij± GHQVLW\Ȗ±
permeability coefficient (Kt), which are changing by the time.
- Density changes by the time
Table 1. The result to determine the density Ȗ changing by the time (Unit: kG/m3)
Time (day) 0 5 15 30 45 60 90
Case 1 18,90 18,80 18,80 - - -
18,90
Case 2 18,70 18,20 17,50 17,50 17,40 17,40

- Shear strengh changes by the time

Table 2. The result to determine the cohesion (C) changing by the time (Unit: kG/cm2)
Time (day) 0 5 15 30 45 60 90
Case 1 21,75 21,30 20,69 - - -
22,16
Case 2 21,50 20,00 19,50 19,40 19,15 19,10
Table 3. The result to determine the friction angle ijchanging by the time (Unit: °)
Time (day) 0 5 15 30 45 60 90
Case 1 14,975 14,580 14,177 - - -
15,210
Case 2 14,830 14,320 13,900 13,780 13,710 13,650

- Permeability coefficient changes by the time

Table 4. The results to determine the permeability coefficient (Kt) changing by the time (Unit: cm/s)
Time (day) 0 5 15 30 45 60 90
Case 1 1.10-6 3,5.10-6 5.10-6 - - -
1.10-6 -6 -6 -6 -6 -6
Case 2 1,5.10 4.10 4,5.10 5.10 5,5.10 6.10-6
Fig. 7 summarises the main testing results by plotting the changes of physical
properties of embankment due to seepage effect in function of time.
Experimental and numerical assessments of seepage … 713

(a) (b)
Fig. 7. The graph shows a change physical
properties of embankment due to seepage
effect by the time
(a) The graph shows the results of the density
change by the time
(b) The graph shows the results of the cohe-
sion change by the time
(c) The graph shows the results of the friction
(c) angle change by the time

5 Calculation application

During the design of the earth dam, the determination of the safety factor (FOS) is
greatly significant to the reservoir safety. However, people have always calculated the
dam stability based on the conditions and the input materials that are deterministic. It
means that average values of some key input parameters such as physical properties
RIHPEDQNPHQWȖij&.t) are generally assumed for all positions on the devastating
slope failure. Moreover, the potential effect of seepage is usually ignored. Based on
the experimental results mentioned above, the authors will determine FOS consider-
ing the effect of changing physical properties of embankment due to seepage effect by
the time by dividing the embankment by saturation margin.
Ta Trach reservoir is not a homogeneous earth dam. It has 60m height, crest length
of 1187m and impoundment weight of 10m. Dam foundation is mainly mild weath-
ered rock layer treated by spraying cement mortar waterproofing membrane to a depth
of 26m. The construction of a dam is described in Figure 8a. Stability analysis of the
main dam of Ta Trach reservoir in Thua Thien Hue Province, Vietnam was performed
using the commercial finite element program Plaxis 2D. The classical Mohr-Coulomb
constitutive model was considered. Dam stability results are calculated according to
the change of some physical properties of embankment due to seepage effect by the
time at time T = 100 years.
Fig. 8. FOS results of the main dam of Ta
Trach reservoir, Thua Thien Hue Province,
Vietnam
(a) Simulation of cross section on Plaxis
(b) The position of the dangerous sliding sur-
face after the seepage is stable (K=1.328)
(c) The position of the dangerous sliding sur-
(a) face after the water level is lower (K=1.124)
714 T. Q. Nguyen et al.

(b) (c)
Based on the calculation results on the numerical model by Plaxis software, it
shows that during 100 years, considering the changes of physical properties of em-
bankment due to seepage effect by the time, FOS of the downstream slope is K =
1.328 and upstream slope is K = 1.124. In the design level, the service life is 100
years, according to the allowable FOS under current Vietnamese standard, the down-
stream and upstream slope are unstable at this time.

6 Conclusion

An efficient testing equipment of seepage effect on some key material properties of

embankment, which include cohesion, friction angle, density and permeability coeffi-
cient, have been developed and presented in this paper. Different regression lines
were established from the experimental tests, which could be used to estimate the
changes of these material properties by time caused by the seepage phenomenon.
The testing results were then used in the stability analysis of the main dam of Ta Trach
reservoir using the finite element method. It can be concluded from the calculated results
that the combination of the testing equipment developed in this paper and the finite ele-
ment method can provide an efficient tool for the assessment of embankment stability
due to seepage as well as make early warning of the potential instability of dam.

References
1. O. E. Omofunmi, J. G. Kolo, A. S. Oladipo, P. D. Diabana, and A. S. Ojo, A Review on
Effects and Control of Seepage through Earth-fill Dam, Current Journal of Applied Sci-
ence and Technology 22(5); Article no.CJAST.28538 (2017).
2. Ripendra Awal, Hajime Nakagawa, Masaharu Fujita, Kenji Kawaike, Yasuyuki Baba and
Hao Zhang), Study on piping failure of natural dam (2011).
3. Pawan Kumar Bhattarai, Hajime Nakagawa, Kenji Kawaike, Hao Zhang, "Experimental
study on river dyke breach characteristics due to overtopping," Journal of JSNDS , vol. 33,
pp. 65-74 (2014).
4. Lin Ke, Akihiro Takahashi, Experimental investigations on suffusion characteristics and
its mechanical consequences on saturated cohesionless soil, Soils and Foundations, Vol-
ume 54, Issue 4, August 2014, Pages 713-730 (2014).
5. Geological report of Ta Trach reservoir ± Thua Thien Hue Province, Vietnam, Stage con-
struction drawings (2010).
6. Head, K.H., Manual of Soil Laboratory Testing, John Wiley and Sons, 1986.
 A numerical homogenized law using discrete element
method for continuum modelling of boundary value
problems

Trung Kien NGUYEN1,2, Jacques DESRUES2, Gaël COMBE2, Duc Hanh NGUYEN1
1
National University of Civil Engineering, Hanoi, Vietnam
2
Laboratory 3SR, University of Grenoble, CNRS, France
trungkiennguyen@hotmail.com

Abstract. This paper presents a new way to define constitutive laws based on
particles interactions within the Discrete Element Method (DEM). The concept
is based on Volume Element (VE) for which the constitutive law is constructed
through a numerical homogenization process. The new law fully accounts for
the discrete nature of granular materials. By using the response envelope dia-
grams proposed by Gudehus [1], a graphical representation of the constitutive
law is obtained. The results suggest that the current law can fully capture the
main features of granular materials such as anisotropy, path dependence and
non-linearity.

Keywords: FEMxDEM Modelling, Granular materials, Homogenization.

1 Introduction

Granular materials are a special class of materials: some authors consider their behav-
ior as intermediate between solid and fluid (see for instance [2]). Their behavior is
heterogeneous and anisotropic by nature. In the context of continuum modeling of
boundary value problems (BVP) by Finite Element Method (FEM), seeking after a
constitutive model that can reproduce all the relevant properties of these materials, or
a significant part of them, has a long story. Many works have been dedicated to this
subject, attempting to model as much as possible the complex behavior of granular
materials. They are often based on the theory of plasticity. Only a few of them allow
representing an anisotropic behavior, e.g. [3,4]. Taking into account anisotropy in the
plastic part in elastoplasticity response requires complex mathematical descriptions
which can be difficult to implement, and to calibrate as well. In view of these persis-
tent difficulties, a new trend was introduced in the geomechanical constitutive model-
ling landscape since the early 2000’ years; it consists in building constitutive models
directly based on discrete modeling, using a step by step computational homogeniza-
tion [5,6,7].
Discrete Element Method (DEM) has been well known as an effective method for
modeling granular material at the grain scale. Nevertheless, despite the advantages of
the DEM approach, it seems to be impossible in most cases to model boundary value

© Springer Nature Singapore Pte Ltd. 2020 715

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_113
716 T. K. Nguyen et al.

problems involving both real-size grains and real-size structures. Thus, building a
constitutive law based on this method (i.e. DEM) and then implementing this homog-
enized law (NHL) into a FEM framework is of real interest. Such a model is expected
to naturally describe specific features of granular materials. By using this new class of
constitutive laws, real-size grain micro-structure on real-size macroscopic problems
can be performed, without facing the intractable problem of dealing with trillions of
grains in a full DEM model.
In the present paper, we focus on the response characteristics of this new constitu-
tive law. The paper is thus structured as follows: Section 2 presents the DEM based
constitutive law. A graphical representation of this constitutive law by response enve-
lopes is studied and discussed in section 3. Some conclusions are given in section 4.

2 Numerical law based on DEM

RVE initial RVE deformed

Microstructural evolution
DEM computation

Displacement gradient  kl Total stress state  ij






ij




kl

Constitutive law based on DEM

Fig. 1. Numerical constitutive law based on DEM

The constitutive law is a key ingredient of a Boundary Value Problem (BVP). This
law expresses the local stress as a function of the history of the local deformation
gradient. Constitutive models can be categorized into phenomenological models and
micromechanical models (or macro- versus micro-scale models). Micromechanical
constitutive models require a thorough understanding and models for each constituent
which is not given for all materials. However, we believe that this is definitely the
case for granular materials which can reliably be modelled using DEM (and a Volume
Element). The latter can be considered to be similar to a constitutive law with a large
(but delimited) number of internal variables such as: grain geometries, contact net-
work and contact forces. The procedure to build this law was detailed in [5,7].
The overall picture of this law is illustrated in Fig. 1. A Volume Element (RVE) is
used to define the constitutive law. At any state, this RVE is characterized by the size
and interaction of the grains of which it is made. By applying a deformation gradient
𝛿ℎ#$ , the system is simulated by DEM computation. The Cauchy stress 𝜎&' results
from microscopic forces between grains [8]:
) -/ -
𝜎&' = (,-∈ 𝑓& . 𝑟' (1)
*
A numerical homogenized law using discrete element … 717

where 𝑆 in two dimensional problem is the area of the microstructure; 𝑓&-/ , 𝑟'-
are respectively the contact forces acting in contact 𝑐 = (𝑝, 𝑞, and the branch vector
joining the mass center of two grains (𝑝, 𝑞 in contact; 𝐶 is the set of all contacts in the
granular assembly. The VE is a granular assembly of polydisperse circular 2D parti-
cles. All grains interact via linear elastic laws and Coulomb friction when they are in
contact. The normal repulsive contact force 𝑓<$ is related to the normal apparent inter-
penetration 𝛿 by normal stiffness coefficient 𝑘> (𝛿  0 if a contact is present, 𝛿 = 0 if
there is no contact). In order to model cohesive-frictional materials, a local cohesion
is introduced for each contact by adding an attractive force 𝑓A to 𝑓<$ . Thus, the overall
normal contact force is written as:

𝑓> = 𝑓<$ + 𝑓A = −𝑘> 𝛿 + 𝑓A (2)

A degradation of the cohesion is taken into account by considering that 𝑓A can van-
ish when a contact separation occurs. The tangential contact force 𝑓D results from an
accumulation of increments Δ𝑓D = 𝑘D Δ𝑢D computed at each time. Δ𝑢D is the tangential
relative displacement in the contact and 𝑘D is the corresponding tangential stiffness.
The Coulomb friction is considered as follow [9].

𝑓D ≤ 𝜇𝑓<$ (3)

3 Characterizing the response of the NHL: Gudehus’ response

envelopes

Gudehus 1979 proposed the concept of response envelopes as a powerful tool for the
graphical representation of an incremental constitutive law [1]. By applying a defor-
mation increment Δ𝜀 in different directions in the deformation increment space, dif-
ferent stress increments Δ𝜎 are obtained, that can be plotted in the stress increment
space. The response envelopes are polar diagrams in the stress increment space of the
stress increments resulting from different unit deformation increments in the defor-
mation increment space. Their shapes and sizes depend upon the behavior of the con-
stitutive law, and the way it takes into account the history of previous deformation,
which can be very complex (e.g. cycles).
In the present paper, we use the response envelopes concept to analyze the incre-
mental response of our DEM-based Numerical Homogenized law (NHL), which is
expected to naturally describe the properties of granular materials.
Incremental strain Δ𝜀 defined as Δ𝜀) = Δ𝜀 . 𝑠𝑖𝑛𝛼 and Δ𝜀N = Δ𝜀 . 𝑐𝑜𝑠𝛼 (0 ≤ 𝛼 ≤
360°) with Δ𝜀 = Δ𝜀) N + Δ𝜀N N = 1 is applied to the VE to compute numerically
the incremental stress responses Δ𝜎. Five initial stress states issued from a biaxial
loading are used, labelled A, B, C, I, L. They correspond to the cases of the VE in the
following states (Fig. 2): isotropic, intermediate anisotropic, failure (peak strength)
and so called “critical state” – i.e. post failure stationary behavior.
The VE used for the study presented here is a 2D assembly made of 400 circular
grains, with distribution of size between 𝑟T&> and 𝑟TUV such that 𝑟TUV 𝑟T&> = 5/3. The
initial coordination number and void ratio are respectively 4.2 and 0.18. The material
718 T. K. Nguyen et al.

is cohesive-frictional. The normal stiffness of contact 𝑘> is such that 𝜅 = 𝑘> 𝜎Y =

1000, where 𝜎Y is the 2D confining pressure. The stiffness ratio 𝑘> 𝑘D = 1. The cohe-
sion force is defined regarding to the confining pressure by cohesion level 𝑝∗ =
𝑓A 𝑎 𝜎Y = 1 with 𝑎 the typical diameter of grain in VE. The intergranular angle of
friction is set 𝜇 = 0.5. When sliding or disjunction occurs at one contact, the cohesion
is considered as destroyed, so 𝑝∗ vanishes for this contact.
5

3
4
C

3
2 I

 (%)
q/
2

vv
L 2

1 1

0
A
0
0 1 2 3 4 5 6 7 8
 (%)
11

Fig. 2. DEM responses subjected to biaxial loading. Studied points (A, B, C, I, L) are marked
in red (left); and direction of strain increment applied (right).

It is worth recalling that for each of these stress states –naturally associated with a
VE state – a full characterization of the stress increment response has been performed
over a 0-360° range of the deformation increment orientation, by 5° steps. Moreover,
for each orientation, various magnitude of the incremental strain Δ𝜀 is applied,
from 0.05% to 0.5% in five steps.
Response envelopes of the different initial stress state (A, B, C, I, L) are plotted in
Fig. 3. It is found that, at initial stress state (A – isotropic) (Fig. 3.a), with Δ𝜀 =
0.05% and Δ𝜀 = 0.1%, response envelopes are close to an ellipse centered at the
origin of stress plan. This shape corresponds to a linear behavior [1,10]. Any signifi-
cant deviation of this kind of response indicates a non-linear behavior (e.g. changes of
envelope center from origin point, or shape deformed); hence, if Δ𝜀  0.1% the
behavior becomes non-linear.
For B, C, I and L configurations, stress states are anisotropic. The response enve-
lopes show non-linear behavior and irreversible even with a very small incremental
strain applied. As seen in Fig. 3.(b,c), response envelopes become flat on one side,
and decentered, even for the smallest incremental strain magnitudes (0.05%, 0.1%) .
These results suggest that at an anisotropic state, the behavior of the NHL law is al-
ways strongly non-linear. By considering the evolutions of responses envelopes at
different states, it is observed that the rigidity of VE is decreased due to changes in
microscopic properties. It is interesting to note that we obtain these effects naturally,
while in classic continuum modeling, it is necessary to define and manage a set of
variables to describe history loading dependency.
From the isotropic elasticity equation, we can easily obtain that an ellipse centered
at the initial stress characterizes the envelope response and its major axis direction is
45°. It is because the horizontal and vertical stiffness are equal. In our results, as
A numerical homogenized law using discrete element … 719

shown in Fig. 3.a, at small strain increment ( Δ𝜀  0.1% the stress response is elas-
tic. The highest absolute values of the stress responses also occur in the direction 45°.
This means that, the preparation procedure of the specimen gives a quasi-isotropic
response at the initial state (isotropic stress state). However, the limit of elastic behav-
ior is very slight and it means that an anisotropic initial state can be obtained by
changing some parameters in the preparation procedure of the sample. Obviously, it
can be seen on Fig. 3.a that the extension part of response envelopes tends towards a
limit (clearly from Δ𝜀  0.2%). This corresponds to extension limit of VE character-
ized by the cohesion of intergranular contact. Meanwhile, the size of response enve-
lopes in compression part increase with strain increment Δ𝜀 , especially in the 45°
direction, corresponding to an isotropic compression loading 𝑑𝜀) = 𝑑𝜀N  0.Induced
anisotropy is an important property of granular materials. However, describing in-
duced anisotropy in usual constitutive law is a difficult task. This property can natu-
rally be described in our DEM model. By considering 3 VE at different states, corre-
sponding to a VE subjected to biaxial loading: A – isotropic, B – intermediate and C –
peak state, we observe that the anisotropy is induced throughout the loading. Rota-
tions and size of response envelopes at small and large increment strain as well indi-
cate clearly the changes in anisotropic properties.

Fig. 3. Incremental response envelopes

Obviously, at the same strain increment, response envelopes at different states

show history loading dependency of material behavior. As shown in Fig. 3, the size of
response envelopes decreases with increasing of mean stress from initial state. This
means that the stiffness decrease. This latter is due to the fact that increasing loading
cause failure in contact cohesion, and thus reduces the stiffness of material.
720 T. K. Nguyen et al.

At critical state (L), we can consider the response envelope is composed of two
sections: elastic unloading and elasto-plastic loading. The section representing elasto-
plastic loading is reduced, tending to a straight line. This means that the model pre-
dicts a very small stress rate for elasto-plastic loading at critical state. Two reasons
that explain why we cannot obtain a perfect straight line: firstly, state L is not abso-
lutely critical state because of the scattering of DEM responses; secondly, usual elas-
to-plastic model predicts zero stress state due to consistency condition.

4 Conclusion

A numerical law based on Discrete Element Modeling has been presented. This is
accomplished in the framework of numerical homogenization method applied to
granular media. A conventional way to study constitutive law by using response enve-
lopes is then applied. By performing a series of numerical test on VE, the overall
picture of the constitutive relation is captured. The results show the ability to take into
account many natural features of granular materials such as induced anisotropy, non-
linear, stress path dependent or reduction in stiffness according to mechanical loading.
This highlights a great advantage of this new model compared to conventional model
employed in geomechanical modeling.

References
1. Gudehus G.: A comparison of some constitutive law for soils under radially symmetric
loading and unloading. Proceedings of 3rd Int. Conf. on Numerical Methods in Geome-
chanics, 1309-1323, (1979).
2. Lanier J., Mécanique des Milieux Granulaires, Hermes Science Publications, 2001.
3. Dafalias, YF.: Bounding surface plasticity. I: Mathematical foundation and hypoplasticity.
Journal of Engineering Mechanics 112(9), 966-87 (1986).
4. Wheeler, SJ. et al.: An anisotropic elastoplastic model for soft clays. Revue Canadienne de
Géotechnique, 40(2), 403-418 (2003).
5. Desrues, J. et al.: From discrete to continuum modelling of boundary value problems in
geomechanics: An integrated FEM-DEM approach. Int. J. Numer. Anal. Methods in Geo-
mech, 42(5), 919-955 (2019).
6. Miehe, C. and Dettmar, J.: A framework for micro–macro transitions in periodic particle
aggregates of granular materials. Computer Methods. Comput Method Appl Mech
Eng, 193(3), 225-256 (2004).
7. Nguyen TK. et al.: FEM x DEM Modelling of Cohesive Granular Materials: Numerical
Homogenisation and Multi-scale Simulations. Acta Geophysica, 62(5), 1109-1126 (2014).
8. Weber, J.: Recherches concernant les contraintes intergranulaires dans les milieux pulvéru-
lents. Bul. Liaison Pont et Chaussées, No. Juillet-aout (1966).
9. Cundall, P.A. and Strack, O.D.: A discrete numerical model for granular assemblies. Ge-
otechnique, 29(1), 47-65 (1979).
10. Bardet J.P., Numerical simulations of the incremental responses of idealized granular ma-
terials. International journal of Plasticity, 10(2): 897-909 (1994).
 Influence of heterogeneous fractured fault damage zones
on shear failure onset during fluid injection

Trung Kien NGUYEN1, Jeremy ROHMER2, Ba Thao VU3

1
National University of Civil Engineering, Hanoi, Vietnam
2
Bureau de Recherches Géologiques et Minières, Orléans, France
3
Hydraulic Construction Institute, Vietnam Academy for Water Resources, Hanoi, Vietnam
trungkiennguyen@hotmail.com

Abstract. Fault stability analysis is traditionally performed by assimilating fault

systems to surfaces. Yet, faults are complex and heterogeneous geological sys-
tems, whose compartmentalized architecture generally corresponds to an inner
core (FC) of small thickness (i.e. principal fault plane) surrounded by outer, of-
ten fractured damage zones (DZ). Depending on the fractures’ network charac-
teristics, the latter compartment can be related to complex spatial distribution of
hydro-poro-elastic properties, which can strongly influence the shear failure
tendency of the fault zone during massive injection of fluid into reservoirs. Us-
ing the upscaled DZ properties derived from outcrop surveys at Cirques de
Navacelles (South of France), we investigate this issue using coupled hydro-
mechanical simulations in the framework of fully saturated orthotropic elastic
porous media. By comparing the shear failure tendency for the heterogeneous
DZ cases to the ones with homogeneous DZ, we highlight that: 1. Whatever the
stress regime (extensional or compressional), the maximum injection pressure is
greater in the heterogeneous cases; 2. Under extensional regime, the presence of
the DZ limits the development of shear failure tendency in the center of the first
DZ compartments directly adjacent to FC, whereas shear appears to rapidly de-
velop along the whole reservoir thickness for the homogeneous case; 3. Under
compressional regime, the presence of the DZ enhances the localization of
shear failure along FC-DZ interface, whereas shear failure preferably develops
in the injection zone in the homogeneous case.

Keywords: Fault stability, Fluid injection, Heterogeneous DZ.

1 Introduction

When fluid is injected into rock reservoir, changes in fluid pressure will affect stress
state of rock mass; reduce fault strength leading to possible failure (e.g. shear reacti-
vation of fracture). A central difficulty and the main challenge in evaluating the rela-
tionship between fluid pressure and fault slip lies in the variation in material proper-
ties across the fault zone. In the recent years, geomechanical analysis of fault zone
behavior has been conducted for fault stability during fluid injection operations (see
for instance for CO2 storage applications [1,2]).

© Springer Nature Singapore Pte Ltd. 2020 721

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_114
722 T. K. Nguyen et al.

Faults are complex and heterogeneous geological systems, which do not corre-
spond to discrete surfaces as already postulated by many authors (e.g., [3,4]). A fault
zone is generally composed of an inner core made of fine material, often impermea-
ble, and where slip is concentrated called fault core (FC). It is surrounded by an outer
damage zone (DZ) that often acts as a hydraulic pathway, because of the presence of a
fracture network, whose characteristics (fractures’ orientation, connectivity, lengths,
density, etc.) depend on the distance to the core. The behavior of rock mass is there-
fore dependent on fracture networks’ characteristics (spatial density, length, orienta-
tion, etc.), hence leading to heterogeneous hydro-mechanical DZ properties.
Some researches made the efforts to measure permeability and Young’s modulus
of fault zone [5]. Its results show that these properties are variable in DZ and strongly
depend on their architecture. More recently, Nguyen et al. [6] has been estimated
hydro-poro-mechanical properties by a numerical/analytical homogenization method.
They brought out that the dependence of these properties in function of fracture densi-
ty in DZ. A simple model by considering only reservoir part in fluid injection was
used in [6]. The results revealed that the hydro-poro-mechanical influent to stability
of FC. As a continuity of this recent work, a more realistic hydromechanical model is
proposed in this paper to assess the shear tendency of reservoir during fluid injection.
This is carried out by estimating the maximum sustainable overpressures that will not
lead to failure. As shown by [7,8], the maximum overpressure strongly depends on
the in-situ stress state, it means that compressional or extensional stress regime.
Therefore, we propose to see how heterogeneous hydro-poro-mechanical properties
influent shear tendency in reservoir during fluid injection.
In this context, the paper is organized as follows. Section 2 summarizes the ho-
mogenization process applied to fractured damage zone collected at Navacelles. Sec-
tion 3 presents the numerical geological setting to model hydromechanical behavior
of reservoir under influence of fluid injection. This model takes into account the usual
homogeneous hydro-poro-mechanical properties of rock mass and also heterogeneous
ones in different cases. Section 4 describes the numerical results with emphasis in the
identification of critical zone. Some conclusions and remarks are given in section 5.

2 Homogenization method of fractured damage zone

Geological surveys were conducted at the Cirque de Navacelles, in the late Jurrassic
platform carbonates of Languedoc, south of France. This site is located outside major
fault systems and can be considered as a good analogue of the low-fractured rock
formation targeted for CO2 storage reservoir. The fault systems investigated can be
assimilated to a layered system crossed by strike-slip fault zones characterized by
fractured damage zones surrounding a fault core of small thickness. Based on the data
collated on the fractures’ characteristics of the DZ, a geostatistical model was defined
in [6] to stochastically generate fracture networks representative of the onsite observa-
tions. A numerical homogenization strategy was then developed for determining the
effective hydro-poro-mechanical properties of DZ. The results showed that the hydro-
poro-mechanical homogenized quantities depend on the distance to fault core. Fol-
Influence of heterogeneous fractured fault damage … 723

lowing these observations, empirical relationships were given linking the Young’s
modulus, permeability and Biot’s coefficient with this distance (Fig. 1). The interest-
ed readers are invited to the work of [6] for more details about the procedure.

3 Model geometry and parameters

Fig. 1. a. Model geometry and boundary conditions; vertical (along x axis) and horizontal
(along y axis) DZ properties as a function of the distance d to the fault core (located at d=0): b.
Permeability; c. Young’s modulus; d. Biot’s cofficients.

A two-dimensional plain strain model (60m x 100m) was considered. Fig. 1 gives the
geometry and initial conditions of boundary value problems. The reservoir 20m in
thickness is assumed to be located at 1000m depth. It is bounded at top and bottom
by low permeability 40m caprock (of 10-19 m2). The domain is intersected by a fault
core (FC) of 0.5m with a dip angle of 85° and very low permeability (of 10-17 m2),
surrounded by two DZ.
Fluid injection is modeled at left boundary of reservoir’s level. An overpressure is
increased from 0 to 25MPa (quasi-static loading). The maximum sustainable over-
pressure (ΔPmax) corresponds to the onset of rupture which occurs in the reservoir
(shear slip in FC or in DZ). The problem was solved in the framework of fully satu-
rated isothermal elastic porous media. Hydromechanical behavior of reservoir is as-
sumed to be elastic orthotropic while the caprock behave as isotropic elastic materials.
The whole domain is discretized by 5194 linear triangular-elements and numerical
simulations are performed by using Code_Aster®.
In the present papers, two different tectonic stress regimes: compression (𝜎" =
1.5𝜎' ) and extension (𝜎" = 0.7𝜎' ) are studied. For each regime, we consider two dif-
ferent cases: homogeneous case (HO) and heterogeneous case (HE). The HO means
that DZ’s hydro-poro-mechanical properties are spatially constant while HE means
that these parameters depend on the distance to FC. Related to the latter case (HE),
724 T. K. Nguyen et al.

hydro-poro-mechanical properties used are issued from data of in-situ campaign and
then treated as described in Section 2. These properties in function of distance 𝑑 to
FC are described as follows, with 𝐾,, , 𝐸 , 𝐵 , 𝐵,, are vertical permeability, horizon-
tal Young’s modulus, horizontal and vertical Biot’s coefficient of DZ, respectively:
𝐾,, = 8. 10 𝑒 .  𝑚 (1)
20 GPa 𝑖𝑓 𝑑 > 20m
𝐸 = (2)
0.033 𝑑  + 5 GPa 𝑖𝑓 𝑑 ≤ 20m

0.15 𝑖𝑓 𝑑 > 20m

𝐵 = 0.033 𝑑 + 0.88 𝑖𝑓 5m ≤ 𝑑 ≤ 20m (3)
0.68 𝑖𝑓 𝑑 < 5m

0.4 𝑖𝑓 𝑑 > 20m

𝐵,, = 0.007 𝑑 + 0.37 𝑖𝑓 5m ≤ 𝑑 ≤ 20m (4)
0.5 𝑖𝑓 𝑑 < 5m

Other parameters 𝐾 (horizontal permeability), 𝐸,, (vertical Young’s modulus),

𝐵,, (vertical Biot’s coefficient), 𝜈 (Poisson’s ratio), 𝜙 (porosity) are kept constant in
these simulations. A bulk density of 2260kgm-3 is considered. The numerical parame-
ters of rock formations are summarized in Table 1. The following notations are adopt-
ed: CHE for Heterogeneous case in Compressional stress regime; CHO for Homoge-
neous case in Compressional stress regime; EHE for Heterogeneous case in Exten-
sional stress regime; EHO for Homogeneous case in Extensional stress regime.

Table 1. Properties of rock formations.

Case Exx Eyy Kxx Kyy Bxx Byy φ ν

(GPa) (GPa) (m2) (m2) - - (%) -
Fault core 1 1 10-17 10-17 1.0 1.0 1 0.25
-13
HE DZ Exx 10 10 Kyy Bxx Byy 15 0.25
HO DZ 20 10 10-13 Kyy 1.0 1.0 15 0.25
Caprock 10 10 10-19 10-19 1.0 1.0 1 0.25

4 Numerical results

In this section, we focus on the hydromechanical behavior of reservoir. By increasing

of overpressure ΔP at the left side of reservoir, pressure increases rapidly in left DZ
and almost stops at FC. Pressure increase induces general reduction in effective stress
that causes failure. The Mohr-Coulomb’s criterion 𝜏 𝜎′K ≤ 𝜇 is used for evaluating the
onset of rupture. By calculating 𝑚𝑎𝑥 𝜏 𝜎′K , these results present the most unfavora-
ble case at which the maximum sustainable overpressure can be obtained. For sim-
plicity and conservative reason, the coefficient of friction 𝜇 = 0.6 [2].
In these simulations, FC with low permeability plays a seal role. It limits the in-
creasing of overpressure in the left part of reservoir. Biot’s coefficients of FC are
𝐵 = 𝐵,, = 1.0. For DZ, these coefficients are spatial dependent in heterogeneous
Influence of heterogeneous fractured fault damage … 725

case while are constant and equal 1.0 in homogeneous case. If Biot’s coefficients are
equal in both FC and DZ, the whole domain (FC and DZ) has the same initial effec-
tive stress. It is the case of CHO and EHO. In heterogeneous cases, DZ properties are
spatial dependent. The nearby zone to FC has a smaller Biot’s coefficient than in the
far zone. Thus, the interface of DZ and FC is influenced by this property. Toward the
DZ, Biot’s coefficients are smaller, that’s why initial effective stress in FC is more
critical (dangerous) than in DZ.
For CHE case, first rupture is detected at two corners of left domain, in the inter-
face between FC and DZ, corresponding to ΔP = 15MPa (Fig. 2.a). After this first
observation, rupture continues to develop from two corners to DZ. From ΔP = 20MPa
a failure’s zone is also observed in the middle of FC, developed from interface DZ/FC
to FC (Fig. 2.c). Meanwhile in CHO, first failure is detected from upper corner at ΔP =
8.5MPa but a strong development of failure is then observed at left boundary (injec-
tion zone) (Fig. 2.b,d). It developed from this injection zone to FC and all the left do-
main is activated after 13MPa of overpressure.

Fig. 2 𝑚𝑎𝑥 𝜏 𝜎′K for compressional cases. Arrows indicate shear initiation.

Fig. 3. 𝑚𝑎𝑥 𝜏 𝜎′K for extensional cases. Arrows indicate shear initiation.

Failure appears very soon in extensional case, after about 3.5MPa in EHE and
2.6MPa in EHO (Fig. 3.a,b). In EHE case, it begins from the middle of FC and then in
726 T. K. Nguyen et al.

DZ after activates almost the whole FC. In EHO case, failure develops both in FC and
DZ. As influenced by heterogeneous hydro-poro-mechanical properties, failure de-
velopment is quite different from EHE to EHO. The first shear activation for EHE and
EHO is detected at the same overpressure but after that, at each level of Δ𝑃, the acti-
vated zone in EHO is wider than in EHE. Furthermore, while failure zone in EHE
appear from the middle to upper and lower boundary, but in EHO it does appear at the
same time from the middle and two small zones in top and bottom of reservoir. This is
directly related to the influence of caprock.

5 Conclusion

Numerical investigation has been presented in this paper to investigate the influence
of heterogeneous fractured fault damage zones on shear failure onset during fluid
injection into porous reservoir. Rock reservoir’s behavior has been treated as an or-
thotropic material with heterogeneous hydro-poro-mechanical properties issued from
the recent works of [6]. Different stress regimes such as compressional and extension-
al case have been simulated. The results confirm that increasing of overpressure re-
duce effective stresses and thus promote failure. Shear activation tendency has been
analyzed. The results reveal the most dangerous situation by evaluating the maximum
of fraction 𝑚𝑎𝑥 𝜏 𝜎′K which induce rupture according to Mohr Coulomb criterion.
Along the papers, heterogeneous cases are well distinguished with homogeneous
ones. The results show clearly the consequence of using hydro-poro-mechanical prop-
erties which describes heterogeneous fracture network. This demonstrates the fact that
due to heterogeneous properties, the maximum sustainable overpressure causes failure
could be increased. The results help to improve the understanding of different man-
ners and locations of failure while using “real” properties in numerical modeling.

References
1. Cappa, F. and Rutqvist, J.: Impact of CO2 geological sequestration on the nucleation of
earthquakes. Geophysical Research Letter 38(17) (2011).
2. Streit, JE., Hillis, RR.: Estimating fault stability and sustainable fluid pressures for under-
ground storage of CO2 in porous rock. Energy 29, 1445-56 (2004).
3. Rohmer, J. et al.: Off-fault shear failure potential enhanced by high-stiff/low-permeable damage
zone during fluid injection in porous reservoirs. Geophys. J. Int. 202(3), 1566-1580 (2015).
4. Wibberley CAJ., Yielding G., Di Toro G.: Recent advances in the understanding of fault
zone internal structure: A Review. J. Geol. Soc. (London) 299, 5–33 (2009).
5. Faulkner, DR. et al.: Slip on weak faults by the rotation of regional stress in the fracture
damage zone. Nature 444(9), 22-25 (2006).
6. Nguyen, TK. et al.: Integrating damage zone heterogeneities based on stochastic realizations of
fracture networks for fault stability analysis. Int. J. Rock Mech. Min. Sci 80, 325-336 (2015).
7. Sibson, RH: Brittle-failure controls on maximum sustainable overpressure in different tec-
tonic regimes. AAPG bulletin 87(6), 901-908 (2003).
8. Rutqvist, J., and Tsang, CF.: A study of caprock hydromechanical changes associated with
CO2-injection into a brine formation. Environmental Geology 42(2-3), 296-305 (2002).
 Influence of Compaction Factor on the mechanical
behavior of the soils used for the earth dams in the North
Central region, Vietnam

Van Toan TRAN1 and Van Hien TRAN2

1
Thuy Loi University, 175 Tay Son, Dong Da, Ha Noi, Vietnam
2 Hydraulic Engineering Consultants Corporation No.II, 169 Tran Quoc Thao, District 3, Ho
Chi Minh, Vietnam
tranvantoan@tlu.edu.vn and hienhecii@gmail.com

Abstract. Almost all the earth dams constructed in the North Central region
of Vietnam usually encounter problems related to the use of cohesive soils dur-
ing dam construction under humid climate conditions with long and heavy rain
periods. Therefore, it is very difficult to compact the soil to achieve the com-
paction factor of .• required by the designer. In this paper, the authors pre-
sent experimental results to determine the correlation between the compaction
factor (K) with the mechanical behavior of the soils used for the earth dams in
the North Central region of Vietnam. The research results are applied to select
the reasonable value of compaction factor for the soils used in dam construc-
tion, which is suitable with the requirements of the design standard.

Keywords: Earth dam, compaction factor, mechanical behavior, permeability

coefficient, soil moisture content.

1 Introduction

In recent years, most of the earth dams constructed in the North Central region of
Vietnam usually encounter problems related to the use of cohesive soils in dam con-
struction under humid climate conditions with long and heavy rain period. There are
usually two main types of soil used for waterproofing blocks and bearing blocks of
earth dams which are sediment and ruin soil. The key constructions in provinces from
Thanh Hoa to Thua Thien Hue such as: Truoi, Ta Trach, Ngan Truoi, Cam Thuy,
Thuy Yen – Thuy Cam, Da Han, Ban Mong, ... are in regions that always experience
high moisture content in the air and long rainy time [1]. The air moisture content in
many months is greater than 80%. Therefore, it is very difficult to compact the soil to
achieve the compaction factor of .• corresponding to the soil moisture W =
Wopt±3% (Wopt - Optimal moisture content) required by the design standard [2], [3],
[4]. As a result, the construction time lasts longer and it does not meet the construc-
tion progress or economic and social development requirements. This reduces the
effectiveness of the investment and leads to a long time of capital recovery.

© Springer Nature Singapore Pte Ltd. 2020 727

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_115
728 V. T. Tran and V. H. Tran

In this paper, the authors present the experimental results to determine the correla-
tion between the compaction factor (K) with mechanical properties of soils in the
North Central region of Vietnam: friction angle (ij), cohesion (c) and permeability
coefficient (k): (ij F N) ~ K [5]. The research results will help designers select the
reasonable value of soils compaction factor during earth dam construction [4].

2 Methodology

This work is a combination of experimental tests in the laboratory and in the execute
construction work for two types of soil: sediment and ruin rock. Laboratory experi-
ments determine the soil moisture content at the material storage at several periods of
the year, 17 mechanical properties, swelling, shrinkage and wet disintegration. Field
experiments determine the particle components, permeability coefficient k, cohesion
c, and friction angle ij after the soil covered in the dam to reach the requirements of
the design. Experimental results in the field will be compared and be verified with
laboratory experimental results.

Table 1. Statistical table of laboratory test number [5].

No Earth dam Experimental content No. of tests

17 mechanical properties 17
1 Ta Trach
.NijF: curve 22
17 mechanical properties 17
2 Ngan Truoi
.NijF:curve 11
17 mechanical properties 17
3 Thuy Yen
.NijF:curve 22

The experiments were carried out from August 2010 to February 2016 for soil sedi-
ment and ruins soil types in the works Ta Trach dam, Thuy Yen dam and Ngan Truoi
dam. The number of experimental sample group is summarized in Table 1.

3 Experimental results in the laboratory

3.1 Sediment soil

Experimental results in Table 2 show characteristics of sediment soil in the study
region. The results of the proctor experiment showed that the natural moisture content
of the sediment soil is usually 3% higher than the optimal moisture content.
Through laboratory studies, the author built relations between [compaction factor
(K) and moisture content of soil (W)] with [friction angle (ij), cohesion (c), permeabil-
ity coefficient (k) and dry density] as shown in Fig. 1, Fig. 2 and Fig. 3.
Influence of Compaction Factor on the mechanical … 729

Table 2. Characteristics of sediment soil

Earth dam
No Characteristics
Ta Trach Ta Trach 1 Ngan Truoi Thuy Yen
1 Maximum dry density Ȗcmax(T/m3) 1.55 1.62 1.47 1.78
2 Natural density Ȗw(T/m3) 1.94 1.99 1.86 2.08
3 Optimum moisture content Wopt(%) 22.1 21.9 23.5 16.7
4 Natural moisture content W(%) 25.2 23.1 26.2 19.7
5 Density ¨ 2.66 2.72 2.71 2.70
6 Porosity n(%) 48.2 48.4 40.9 38.6
7 Saturation G(%) 92.0 94.5 91.8 89.5
8 Permeability coefficient k(cm/s) 1.10-4 7.10-5 1.10-4 9.10-5
9 Cohesion c(kG/cm2) 0.19 0.24 0.19 0.21
10 )ULFWLRQDQJOHijo) 17o46’ 15o26’ 17o59’ 22o31’
Fig. 1 shows that the permeability coefficient decreases when compaction factor in-
creases. When .•, the permeability coefficient is N§u10-5(cm/s) and stability
when K increase.
Fig. 2 showed that the friction angle ij covariates with the compaction factor K and
rise slightly to reach the largest value when .•. This is suitable with the general
rule. If we increase K, the effect of ij will not increase much.
Fig. 3 shows that the cohesion c covariates with the compaction factor K. When
.•0.98, c reaches the maximum value, in agreement with the general rule TCVN
8216 [1] and TCVN 8297 [4]. When the compaction factor K=0.98, c increases 3.4%
compares to the case of K=0.97. After that, it did not increase significantly.

Fig. 1. Influence of compaction factor on the permeability coefficient (k ~ K) of sediment soil

Fig. 2. Influence of compaction factor on the friction angle (ija.) of sediment soil
730 V. T. Tran and V. H. Tran

Fig. 3. Influence of compaction factor on the cohesion (c ~ K) of sediment soil

3.2 Ruin soil

Experimental results in Table 3 also show natural characteristics of ruin soil in the
study region. The results of the proctor experiment also showed that the natural mois-
ture content of the ruin soil is usually 3% higher than the optimal moisture content.
Through laboratory studies, the authors also built relations between [compaction
factor (K) and moisture content of ruins soil (W)] with [friction angle (ij), cohesion
(c), permeability coefficient (k) and dry density] as shown in Fig. 4, Fig. 5 and Fig. 6.

Table 3. Characteristics of ruin soil

Earth dam
No Characteristics
Ta Trach Ta Trach 1 Thuy Yen Thuy Yen 1
1 0D[LPXPGU\GHQVLW\Ȗcmax(T/m3) 1.80 1.75 1.74 1.68
2 1DWXUDOGHQVLW\Ȗw(T/m3) 2.15 2.08 2.09 2.06
3 Optimum moisture content Wopt(%) 16.5 17.1 16.8 20.1
4 Natural moisture content W(%) 19.3 19.1 20.1 22.8
5 'HQVLW\¨ 2.72 2.67 2.68 2.69
6 Porosity n(%) 42.8 43.7 42.4 40.6
7 Saturation G(%) 68.4 65.7 89.5 90.8
8 Permeability coefficient k(cm/s) 2.10-4 1.10-4 6.10-5 3.10-4
9 Cohesion c(kG/cm2) 0.20 0.18 0.16 0.17
10 )ULFWLRQDQJOHijo) 16o19’ 17o40’ 22o06’ 20o00’

Fig. 4 shows that the compaction factor increases from K=0.90 to K=0.96, the perme-
ability coefficient obtains from k=10-4 (cm/s) to k=5u10-5 (cm/s). Then, when K in-
creases, k is almost decreases.
Fig. 5 shows that the friction angle ij correlates to the compaction factor K and rise
slightly to reach the largest value when .•.
Fig. 6 shows that the cohesion c covariates with the compaction factor K and when
.•8, c reaches the maximum value.
Influence of Compaction Factor on the mechanical … 731

Fig. 4. Influence of compaction factor on the permeability coefficient (k ~ K) of ruin soil

Fig. 5. Influence of compaction factor on the friction angle (ija.) of ruin soil

Fig. 6. Influence of compaction factor on the cohesion (c ~ K) of ruin soil

Comparing the experimental results in the field and the experimental results in the
laboratory (in Table 4), the authors noticed that:
x For the permeability coefficient k obtained from the experimental results in the
field there is no difference in the sediment soil in the earth dam when compac-
tion factor rises up to K=0.97 and K=0.95. This result is consistent with the
experimental results in the laboratory;
x For the friction angle ij and the cohesion c of samples from drill holes in the
earth dam is higher than samples from the quarry soil, but the difference is in-
significant.
732 V. T. Tran and V. H. Tran

Table 4. Experimental results of sediment soil characteristics on the actual site

No Year W (%) Ȗw (T/m3) Ȗk (T/m3) k (cm/s) ijo) c (kG/cm2) K

1 2010 17.42 2.02 1.72 6.91u10-6 21.12 0.22 •
2 2011 20.00 2.02 1.68 4.89u10-6 22.73 0.24 •
3 2012 16.98 1.99 1.70 2.87u10-6 21.51 0.22 •
4 2013 19.53 1.98 1.66 4.50u10-6 19.22 0.22 •

4 Conclusion

The permeability coefficient k decreases when the compaction factor K increases.

When the compaction factor K equals to K•0.94 corresponding to the soil moisture
content W=Wopt+3%, then k reaches the minimum value of k”2.10-5 cm/s. When the
compaction factor K is equal to K=0.95, the permeability coefficient in the field re-
spectively is k=3.63u10-6 cm/s.
The friction angle ij and the cohesion c covariates with the compaction factor K
and these factors reach maximum values when K•0.94 for ij and K•0.98 for c. The
results show that the friction angle values ij and the cohesion values c from holes
drills samples of the earth dam are higher than from the samples of the quarry soil, but
the difference is insignificant.
Thus, through the results of this experimental study, some knowledge is shown as
follows:
x For the sediment soil used for the waterproofing blocks in the earth dam, the
compaction factor should be chosen the value of .•.94 corresponding the
soil moisture content W=Wopt+3% instead of K•0.97 in TCVN 8216, which is
optimal in terms of waterproofing.
x For the ruin soil used for the bearing blocks in the earth dam, the compaction
factor should be chosen the value K=0.97 same as in TCVN 8216, which will
provide the highest stability for the earth dam.
x To ensure soil embankment on the earth dam is stabilized, it is necessary to
control reasonable dam speed during construction process.

References
1. Ministry of Agriculture & Rural Development, Profile of survey and design of the earth
GDPV%DQ0RQJ1JDQ7UXRLĈD+DQ7D7UDFKDQG7KX\<HQ+DQRL-2010).
2. Ministry of Construction, TCVN 8216:2009 “Design standard for compacted earth dam”,
Hanoi (2009).
3. Ministry of Agriculture & Rural Development, QCVN04-05:2012 “National technical reg-
ulation on hydraulic structures - The basic stipulation for design”, Hanoi (2012).
4. Ministry of Construction, TCVN 8297:2009 “Hydraulics structurers – Earth dam: Tech-
nical requirements for construction by compaction method”, Hanoi (2009).
5. Donald P.C., Geotechnical Engineering – Principles and Practices, Prentice Hall, Upper
Saddle, NJ07458 (1999).
 Enhanced oil recovery: A selection technique for the
energy and recovery of Bach Ho field in Vietnam

Nguyen Van Hung*1, Nguyen Minh Quy2, Hoang Long2, Le Quoc Trung2, Le Vu
Quan2
1Petrovietnam University, Faculty of petroleum, Department petroleum engineering, Vietnam
2Vietnam petroleum institute, Trung Kinh, Yen Hoa, Cau Giay, Ha Noi, Vietnam
*Corresponding author: hungnv@pvu.edu.vn

Abstract. With the increase in energy demand in years to come and the de-
crease in oil discoveries during the last decades, it is believed that the enhanced
oil recovery (EOR) technologies will play an important role in energy balance.
This paper presents a full review of EOR status and opportunities to increase
the recovery factors in some typical reservoirs known as in-situ combustion,
CO2 flooding, hydrocarbon flooding, water flooding... These methods were ap-
plied to the lower Miocene reservoir at the Bach Ho (White Tiger) oil field.
Several parameters including produced oil characteristics, oil production rate,
reservoir depth, etc. were considered. It was observed from some preliminary
results that the immiscible gas injection is the most feasible and practical EOR
method for the Bach Ho oil field.

Keywords: Bach Ho oil field, EOR, Energy demand

1 Introduction
During the expectancy of oil wells, the recovery of hydrocarbons passes commonly
three stages: primary recovery, secondary recovery, tertiary recovery [1]. In primary
recovery, the natural energy of the reservoir is used to transport hydrocarbons towards
and out of the production wells. Pumps and gas lifting are involved in the primary
recovery. In this stage, there are usually five important driving mechanisms (solution
gas drive, gas cap drive, water drive, gravity drainage, combination or mixed drive).
The primary purpose of secondary production is to maintain the reservoir pressure by
either a natural gas flooding or water flooding [2]. However, the rise in world oil pric-
es has encouraged the oil companies and authorities to apply the new technical devel-
opments. In addition, the rate of replacement of the produced reserves by new discov-
eries has been declining gradually in the last decades. The way to further increase oil
production is through the tertiary recovery method. Enhanced Oil recovery (EOR)
method is the implementation of various techniques to increase recovery from the
reservoirs depleted by secondary recovery [3]. EOR method is a collection of sophis-
ticated methods to increase the recovery factors from oil field. IOR, or Improved oil
recovery, is a general term which encompass EOR methods as well as new drilling
and well technologies, intelligent reservoir management and control, advanced reser-

© Springer Nature Singapore Pte Ltd. 2020 733

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_116
734 N. V. Hung et al.

voir monitoring techniques and the application of different enhancements of primary

and secondary recovery processes [4].
There were many EOR methods that have been used in the last decades, for the
recovery of light and heavy oils, as well as tar sand. But, EOR methods can be divid-
ed into two categories including thermal and non-thermal recovery methods [5].
Thermal solutions are primarily intended for heavy oils and tar sands, although they
are applicable to light oils in special cases. For these methods, the increase in temper-
ature will reduce the surface tension. But thermal requires substantial investment in
special equipment, such as steam injection, in-situ combustion, hot water flooding.
Non-thermal methods are normally used for light oils, for instances chemical flooding
(polymer, micellar polymer, alkaline surfactant polymer), gas flooding (miscible gas
injection, immiscible gas injection, LPG injection, enriched gas miscible process,
carbon dioxide injection, nitrogen immiscible flooding, nitrogen miscibility displace-
ment mechanism, multi-contact miscibility, gravity drainage) [2].
In 1997, Taber et al [6] listed more than 20 EOR methods that experienced inten-
sive laboratory, and in most cases, significant field testing. However, the authors pro-
vided screening criteria for the eight methods that are either the most important or still
have some promise. An appropriate EOR process selection of targeted is crucial and
an inappropriate EOR method needs to be quickly identified and dismissed at early
stages before detailed feasibility studies of possible applicable EOR processes take
place. Most EOR methods have been applied to onshore fields due to relatively lower
cost and risk than offshore fields. Thus, most EOR screening criteria have focused on
onshore oil fields. Nowadays, EOR application in offshore oil fields is receiving sig-
nificant attention and EOR solution is considered as a highly acceptable option. The
motivation of this study was specially designing to make quick evaluation of EOR
potential, screening studies and prediction of EOR methods. The screening techniques
are based on multi-criteria models, probed analytical solutions, industry experiences,
field practices and experts’ knowledge in EOR applications.

2 Methodology
In the previous section, several EOR methods, that can be applied to enhance the oil
recovery factor, have been briefly presented. This section focuses on discussing the
EOR screening criteria of 16 methods, which are given in Table 1. These criteria were
originally developed by Taber et al. [6] in 1996 and then updated by Aladasani and
Bai [7] for taking into account some other parameters such as porosity and permeabil-
ity ranges; microbial EOR, Water Alternating Gas (WAG) miscible, and hot-water
flooding as EOR methods, along with subcategories of immiscible gas flooding. This
approach enables to develop some calculation tools [8], but it is not intended to repre-
sent threshold limits.
Binary technical screening is based on the comparison of certain properties of fluids
and reservoir of a field under study with the criteria proposed by Aladasani and
Baojun Bai [7]. Properties such as porosity, permeability, viscosity, API, saturation
degree, thickness, depth, reservoir temperature, and lithology are analyzed. The meth-
ods of investigation are: Nitrogen, hydrocarbon, CO2, immiscible gas injection,
ASP/SP (Alkaline surfactant polymer/surfactant polymer), polymer, steam (continu-
ous and cyclic), combustion.
Enhanced oil recovery: A selection technique for the energy … 735

Table 1. Summary of screening criteria for EOR methods [7].

Oil properties Reservoir characteristics

# Gravity Viscosity Porosity Oil saturation Formation Permeability Net Depth Temperature
SN EOR method
Projec (API) (cp) (%) (%) Type (mD) Thickness (ft) (ft) (0F)
Miscible gas injection
Sandstone
28-45 35-0 3-37 15-89 1.5-4500 1500-13365 82-250
1 CO2 139 or Wide range
Avg. 37 Avg. 2.1 Avg. 14.8 Avg. 46 Avg. 201.1 Avg. 6171.2 Avg. 136.3
Carbonate
Sandstone
23-57 18000-0.04 4.25-45 30-98 0.1-5000 Thin unless 4040-15900 85-329
2 Hydrocarbon 70 or
Avg. 38.3 Avg. 286.1 Avg. 14.5 Avg. 71 Avg. 726.2 dipping Avg. 8343.6 Avg. 202.2
Carbonate
33-39 0.3-0 11-24 130-1000 7545-8887 194-253
3 WAG 3 Sandstone NC
Avg. 35.6 Avg. 0.6 Avg. 18.3 Avg. 1043.3 Avg. 8216.8 Avg. 229.4

10000-
Sandstone
38-54 0.2-0 7.5-14 0.76-0.8 0.2-35 Thin unless 18500 190-325
4 Nitrogen 3 or
Avg. 47.6 Avg. 0.07 Avg. 11.2 Avg. 0.78 Avg. 15.0 dipping Avg. Avg. 266.6
Carbonate
14633.3
Immiscible gas injection
16-54 18000-0 11-28 47-98.5 3-2800 1700-18500 82-325
5 Nitrogen 8 Sandstone
Avg. 34.6 Avg. 2256.8 Avg. 19.46 Avg. 71 Avg. 1041.7 Avg. 7914.2 Avg. 173.1

Sandstone
11-35 592-0.6 17-32 42-78 30-1000 1150-8500 82-198
6 CO2 16 or
Avg. 22.6 Avg. 2.1 Avg. 26.3 Avg. 56 Avg. 217 Avg. 3385 Avg. 124
Carbonate
22-48 4-0.25 5-22 75-83 40-1000 6000-7000 170-180
7 Hydrocarbon 2 Sandstone
Avg. 35 Avg. 2.1 Avg. 13.5 Avg. 79 Avg. 520 Avg. 6500 Avg. 175
Sandstone 2650-9199
Hydrocarbon 9.3-41 16000-0.17 18-31.9 100-6600 131-267
8 14 Avg. 88 or Avg.
+ WAG Avg. 31 Avg. 3948.2 Avg. 25.09 Avg. 2392 Avg. 198.7
Carbonate 7218.71
(Enhanced) Waterflooding

13-42.5 4000-0.4 10.4-33 34-82 1.8-5500 700-9460 74-237.2

9 Polymer 53 Sandstone NC
Avg. 26.5 Avg. 123.2 Avg. 22.5 Avg. 64 Avg. 834.1 Avg. 4221.9 Avg. 167

Alkaline
Surfactant 23-34 6500-11 26-32 68-74.8 2723-3900 118-158
10 13 Sandstone 596-1520 NC
Polymer Avg. 32.6 Avg. 875.8 Avg. 26.6 Avg. 73.7 Avg. 2984.5 Avg. 121.6
(ASP)

Surfactant + 22-39 15.6-3 16-16.8 43.5-53 60-60 625-5300 122-155

11 3 Sandstone NC
P/A Avg. 31 Avg. 9.3 Avg. 16.4 Avg. 48 Avg. 55 Avg. 2941.6 Avg. 138.5
Thermal/Mechanical
Sandstone
10-38 2770-1.44 14-35 50-94 10-15000 40-11300 64.4-230
12 Combustion 27 or >10
Avg. 23.6 Avg. 504.8 Avg. 23.3 Avg. 67 Avg. 1981.5 Avg. 5569.6 Avg. 175.5
Carbonate
5E6-3
8-30 12-65 35-90 1-15000 200-9000 10-350
13 Steam 271 Avg. Sandstone >20
Avg. 14.5 Avg. 32.2 Avg. 66 Avg. 2605.7 Avg. 1643.6 Avg. 105.8
32971.3
12-25 8000-170 25-37 15-85 900-6000 500-2950 75-135
14 Hot water 10 Sandstone -
Avg. 18.6 Avg. 2002 Avg. 31.2 Avg. 58.5 Avg. 3346 Avg. 1942 Avg. 98.5
Zero cold Mineable tar
15 Surface mining - 7-11. NC >8 wt% sand NC >10 >3:1 NC
flow sand
Microbial

12-33 8900-1.7 12-26 55-65 180-200 1572-3464 86-90

16 Microbial 4 Sandstone -
Avg. 26.6 Avg. 2977.5 Avg. 19 Avg. 60 Avg. 190 Avg. 2445.3 Avg. 88

3 Case study
The Bach Ho oil field (White Tiger oilfield) is one of the giant oil fields in the Cuu
Long basin of the East Sea located offshore due east of the Mekong Delta of Vietnam.
This study tends to study the Bach Ho lower Miocene reservoir that is a layering
sandstone reservoir. The average depth of this layer is from 2759 – 2998m. The per-
736 N. V. Hung et al.

meability changes drastically from several millidarcies to thousands millidarcies, 2.5-

2500mD (average value 239mD) and the average porosity is 0.177. Initial static res-
ervoir pressure is 28.8MPa, average value of bubble point is 27.80C. Reservoir tem-
perature varies from 800C to 1100C. The saturation is about 0.51 [8] (Table 2).

Table 2. Typical parameters of Bach Ho oil field.

Ini. Pres- Res. Tem-
Depth Permeability Bubble
Porosity sure perature Saturation
(m) (mD) point (0C)
(MPa) (0C)
2759- Avg.
2.5-2500 28.8 27.8 80-110 ~0.51
2998 0.177

In order to select the most appropriate EOR method for applying in our case study
reservoir, a quick screening was used. Table 3 gives the values of most critical pa-
rameters defining the condition of rock and fluid of reservoirs, which involve API
gravity, viscosity, composition, oil saturation, formation type, permeability, depth,
thickness and temperature.

Table 3. Rock and fluid of reservoirs in a case study of Bach Ho oil field.
Oil properties Reservoir characteristics
Tem-
Gravi- Oil For- Perme- Thick
Viscos- Composi- Depth pera-
ty satura- mation ability ness
ity (cp) tion (ft) ture
(0API) tion type (mD) (ft)
(0F)
Sand-
32 1.69 C1-C7 0.51 235 9,656 150 212
stone

4 Result and discussion

Each of the 9 properties presented in the previous section is compared with the crite-
ria of screening in table 1. A score between 0 and 100% is assigned. A score of 1 is
assigned when the property is within the range established by Taber et al. [6], and
zero when it falls in the opposite case. When the property meets only part of the
range, a score is the ratio of compliance. The screening criteria of the other authors
also are considered but are not accounted into the assignment of the score [9].
According the binary technique that technically are feasible to apply in the Bach Ho
lower Miocene layer are: Immiscible gas injection and ASP/SP. The potential EOR
methods that can be considered are: hydrocarbon, combustion, and steam. However,
the score for polymer EOR method is low. The score obtained for this study: N2
(50%), hydrocarbon (60), CO2 (44), immiscible gas injection (83%), ASP/SP (63%),
polymer (30%), steam (60%), combustion (59%). Additionally, steam flooding meth-
od is usually applied in heavy and viscous oils. Thus, steam flooding EOR method is
not suitable for our case study.
Enhanced oil recovery: A selection technique for the energy … 737

Fig. 1. Results of binary technical screening for Bach Ho lower Miocene.

The Figure 1 summarizes the results of calculation. It is found from the results of
the binary screening that the application of the immiscible gas injection could result in
better expectation of implementation than the polymer injection proposed by Lang
and Toan [8] for Bach Ho lower Miocene reservoir. The possible reason is the differ-
ent between result simulated in laboratory and on the screening criteria. However, it is
of the necessity for further analysis to determine which methods have a greater poten-
tial for application or running predictive model such as: recovery factor, economic
effectiveness, analyzing the pattern production and injection schedule, facilities of
equipment.

5 Conclusion
After depletion of reservoir natural pressure drives, a secondary recovery stage is
usually implemented in order to boost and maintain the reservoir. When the recovery
decrease, EOR techniques are to be performed. In this research, at first, different types
of EOR methods were summarized and screening criteria and brief descriptions are
presented for the major EOR methods. Finally, a screening approach has been applied
for a case study in Vietnam.
This paper illustrated eight EOR methods that are popular nowadays. The selection
of EOR method depends on 9 parameters, which are: (1) API gravity; (2) viscosity;
(3) composition; (4) Oil saturation; (5) Formation type; (6) permeability; (7) Depth;
(8) Thickness; (9) Temperature. Technical screening applied in this study for match-
ing enhanced recovery methods to a particular case in Vietnam. If only oil gravity is
available, there will be a wide choice of effective EOR methods. However, there is
often a wide overlap in choices.
The following are main conclusions drawn from the analyses of the considered low-
er Miocene reservoir at the Bach ho oil field.
Immiscible gas injection and ASP/SP are two best EOR methods that ob-
tained from our results
738 N. V. Hung et al.

Hydrocarbon, combustion, and steam are potential EOR solutions of this

reservoir
Our result is the first step for EOR method selection that shown the differ-
ent with the result provided in [8].

References

1. Nguyen Van Hung, Vu Manh Hao, Le Vu Quan, Luong Hai Linh: Các phương pháp tăng
cường thu hồi dầu và kết quả áp dụng tại một số mỏ trên thế giới, Petrovietnam Journal 8
(2018).
2. Abubaker H. Alagorni, Zulkefli Bin Yaacob, and Abdurahman H. Nour: An overview of
oil production stages – enhanced oil recovery techniques and nitrogen injection, Interna-
tional journal of environmental science and development, Vol. 6(9) (2015).
3. Mustafa Versan Kok, Mikhail A. Varfolomeev, and Danis K. Nurgaliev: Application of
different EOR techniques for the energy and recovery of Ashal’cha oil field, Energy
sources, part A – recovery, utilization, and environmental effects, Vol.40 (645-653) (2018.
4. Vladimir Alvarado, Eduardo Manrique: Enhanced oil recovery – An update review, Ener-
gies, Vol.3 (1529-1575) (2010).
5. Al-Anazi B. D: Enhanced oil recovery techniques and nitrogen injection, CSEG recorder,
pp. 29-33 (2007).
6. Taber J.J, Martin F.D, and Seright R.S: EOR screening criteria revisited (part 1) – intro-
duction to screening criteria and enhanced recovery field projects, SPE 35385 (1997).
7. Ahmad Aladasani and Baojun Bai: Recent developments and updated screening criteria of
enhanced oil recovery techniques, SPE 130726 (2010).
8. Lang L.D, Toan N.M: Current studies on enhanced oil recovery for Bach ho and Rong
fields, proceeding Vol. 3 (2012).
9. Trujillo M, Mercado D, Maya G, Castro R, Perez H, Gomez V, and Sandoval J: Selection
methodology for screening evaluation of enhanced oil recovery methods (2010).
 Overview of Geomechanics and its applications to
petroleum industry – a case study for minimum
overbalance pressure calculation

Nguyen Van Hung

Petrovietnam University, Vietnam

Faculty of petroleum, Department petroleum engineering, Vietnam
hungnv@pvu.edu.vn

Abstract. Geomechanics plays an essential role in drilling and production pe-

troleum industry from beginning with pre-drill well planning and continuing
with wellbore stability support while drilling and sand production management.
This paper will cover an overview about geomechanics from its fundamental
knowledge and geomechanical models to its applications in practical problems
of oilfield operations such as wellbore instability, sand production. Additional-
ly, a full review about input data and how to get them by measuring in laborato-
ry and calibration with field measurement will be presented. The result shows
that the Mohr – Coulomb criterion can be used for the wells with the borehole
inclination from 0 to 600 in this study area.

Keywords: Geomechanics, Petroleum, Drilling, Reservoir, Mohr-Coulomb,

Hoek-Brown, Modified Lade, Rock Mechanics

1 Introduction

Geomechanics was considered one of specific topic in petroleum industry because of

its related roles to geoscientists and engineers. Application of geomechanics is of
necessity in greater depth for drilling and production petroleum by using several
themes like earth models, geocell, Mohr-Coulomb failure criterion, and rock micro-
structure.
Geomechanics software applications utilize earth models as an ultimate repository
of gemechanical information and source of inputs and geocell, which is the basic 3D
building. Furthermore, many authors have successful summarized and evaluated the
geomechanics information in each geocell with law of behavior or diagram such as
Mohr’s stability diagram, while others tried to link concepts of petrophysics and geol-
ogy to different rock mechanical parameters that enable to build geomechanical mod-
els. Several features of geomechanics applications could be considered in prediction
and management of subsurface rock deformations by combining information from the
fields of geology, rock mechanics, geophysics, and petrophysics to characterize sub-
surface rock deformations. It is also used to identify the risks of oilfield operations
and to anticipate changes over oil and gas field life caused by changes of bottomhole

© Springer Nature Singapore Pte Ltd. 2020 739

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_117
740 N. V. Hung

pressure, pore pressure, and formation temperature. Additionally, understanding geo-

mechanics encountered in reservoir development helps eliminating variety of prob-
lems in field operations such as unexpected solids production, well blowout, and
steam flooding operation.
The discipline of geomechanics has involved rapidly over the last three decades in
response to industry challenges, and aided by the development of new downhole
measurements, greater computing power, and numerical modeling software. Histori-
cal highlights in downhole measurements included the commercialization of sonic
logging tools that measured both compressional and shear wave velocity and logging
tools that recorded digital images of the borehole surface. These new measurements
enabled researchers to apply and test rock mechanics principles, originally developed
in company laboratories for the mining industry basins. The new sonic logs provide
elastic parameters that improved prediction of rock strength and in-situ stress magni-
tudes. Borehole imaging technology provided new information about how the earth
responds to the stress concentrations while drilling operations. As a result, the predic-
tion of rock strength and in-situ stress using borehole imaging are possible feasibility.
It is clear that Mohr-Coulomb failure criterion required constraining the range of
rock strength and the in-situ stress magnitudes in formations penetrated by wells in
the field. If there are any reasons of without in-situ information about a field or reser-
voir, it is impossible to perform any geomechanics analyses whether for hydraulic
fracturing, wellbore stability, or sand production. Therefore, plot may represent an
entire field, a single reservoir stratum, or small element of a wellbore’s surface. In
fact, the diagram could be used to link between stability plot and an element of sub-
surface rock a geocell, and a 3D geomechanical model. In the other word, geome-
chanics modeling has involved from 1D near-wellbore models to 3D and 4D geome-
chanics models of entire fields. The technology development in logging while drilling
(LWD) tools, real-time monitoring of LWD data, 3D and 4D seismic data and 3D
earth modeling software are responsible for the increased application geomechanics in
the petroleum industry [1].
The motivation of this paper is to present the result of the minimum overbalance
pressure calculation using Mohr-Coulomb criterion and an overview about geome-
chanical model that is different to [2]. For more detailed description of theory and the
methodology can be found in [3-5].

2 Literature review

Theoretical foundations of geomechanics have been developed in the field of rock

mechanics, structural geology, civil engineering, and geophysics. The important ele-
ments to all geomechanics models and applications are: stress, hydrostatic stress,
normal stress, effective stress, normal stresses, strain, linear elasticity, modulus of
elasticity, Poisson’s ratio, Young’s modulus, elastic modulus, elastic wave velocity,
anisotropy properties, rock strength, compressive strength.
In general, geomechanics has variety of applications would be considered in petro-
leum industry such as drilling engineering (well surface location, blowout preventer
 Overview of Geomechanics and its applications to petroleum … 741

rating, wellbore stability, drilling fluid selection), completion engineering (sand con-
trol, hydraulic fracturing), and reservoir management (compaction and subsidence,
fractured reservoirs, diagnosis of failures).
The new full workflow of 3D geomechanical model for shale gas reservoir can be
seen in [6, which involves several complex phases and models. In other word, geome-
chanics would be applied for life of field from exploration and production. In this
section, there is one practical feature selected will be introduced as the typical exam-
ple. The processes of calculation are similar for other applications.
Table 1: Rock failure criteria [7].
Failure Principal stresses at borehole Governing equation
criteria Vertical well Horizontal well
Morh - σ r = Pw σ r = Pw τ = τ 0 + σ tan φ
σ θ = σ H + σ h − 2 (σ H − σ h ) cos 2θ − Pw σ θ = (σ v + σ h ) − 2 (σ v − σ h ) cos 2θ − Pw
Coulomb
σ z = σ v − 2ν (σ H − σ h ) cos 2θ σ z = σ H − 2ν (σ v − σ h ) cos 2θ
τ θ z = τ rθ = τ rz = 0
τ θ z = τ rθ = τ rz = 0
Hoek - σ r = Pw σ r = Pw σ1 = σ 3 + mC0σ 3 + sC02
σ θ = σ H + σ h − 2 (σ H − σ h ) cos 2θ − Pw σ θ = (σ v + σ h ) − 2 (σ v − σ h ) cos 2θ − Pw
Brown
σ z = σ v − 2ν (σ H − σ h ) cos 2θ σ z = σ H − 2ν (σ v − σ h ) cos 2θ
τ θ z = τ rθ = τ rz = 0
τ θ z = τ rθ = τ rz = 0
Modi- σ r = Pw σ r = Pw (
Pw = B − C
1/2

fied Lade σ θ = σ H + σ h − 2 (σ H − σ h ) cos 2θ − Pw
σ z = σ v − 2ν (σ H − σ h ) cos 2θ
τ θ z = τ rθ = τ rz = 0
) ( 2 A)
σ θ = (σ v + σ h ) − 2 (σ v − σ h ) cos 2θ − Pw A = σ z + S + P0
σ z = σ H − 2ν (σ v − σ h ) cos 2θ B = Aσ θ d − σ θ2z
D = (σ θ d + σ z + 3S − 3P0 ) / ( 27 + η )
3
τ θ z = τ rθ = τ rz = 0
{ }
C = B 2 − 4 A D − ( S − P0 ) ⎣⎡ A (σ θ d + S − P0 ) − σ θ2z ⎦⎤

Where: σ r , σ θ , σ z , τ θ z , τ rθ , τ rz , : stress components in cylindrical coordinate

σ v , σ H , σ h : in-situ principal stresses, vertical stress, major horizontal stress,
minor horizontal stress respectively
Pw: internal wellbore pressure
θ : rotation angle
τ 0 , ϕ : cohesion of the material, and angle of internal friction
S , η , ν : material constants, Poison’s ratio

One of the well-known applications in geomechanics is wellbore stability that re-

quires the understanding of three famous failure criteria in rock mechanics: Mohr-
Coulomb, Hoek-Brown, and Modified Lade (table 1). The motivation of this study is
to calculate the minimum overbalance pressure using Mohr-Coulomb criterion, two
others will be presented in the future paper. While some other criteria including Mo-
gi-Coulomb, 3D Hoek-Brown have been found to be the best criterion for the well-
bore stability analysis in some cases [7] compared to Mohr-Coulomb criterion, these
models require more the input data of mechanical parameters. As the result, the Mohr-
Coulomb criterion is considered as prior use in some oil and gas companies.
742 N. V. Hung

3 Result and discussion

Table 2 presents the parameters of overburden stress gradient, pore pressure, cohe-
sion, friction angle and Poisson’s ratio at the depth 10.000ft of the Bach Ho field that
is located in the same reservoir of case study in [2].
Table 2: Rock mechanical properties and stresses data.
Depth D 10.000 ft
σv 1 psi/ft
Stress gradient σH 0.9 psi/ft
σh 0.5 psi/ft
Pore pressure
Pp 0.4 psi/ft
Gradient
Cohesion c 1000 psi
Friction angle φ 0.5236 radian
Poisson's ratio ν 0.25

From figure 2 and table 3, overall, while the minimum overbalance increased as
borehole inclination increased from 0 to 90 degrees for both cases Azimuth equal 0
and 30 degrees, the figures for Azimuth of 60 degrees remained stable. Additionally,
the figure for Azimuth of 90 degrees decreased gradually from 0 to 60 degrees of
Azimuth, which was followed by a strong fluctuation. Finally, it can be also seen that
the value of minimum overbalance pressure was the lowest between 0 to 60 degrees
of Azimuth. Similarly, Adel Al-Ajmi 2006 [6] has obtained the same tendency of this
relationship. In addition, the lowest value of minimum overbalance pressure of our
study was at Azimuth of 60 degrees, except at 70 degrees when applying Modified
Lade criterion. It was the same figures of Adel Al-Ajmi 2006 [6] with the lowest val-
ue at approximately 50 and 70 degrees of Azimuth.
Table 3: Minimum overbalance pressure using Mohr-Coulomb criterion.

Borehole inclination (degree)

0 10 20 30 40 50 60 70 80 90
0 2650 2650 2750 2850 3000 3100 3250 3300 3350 3400
Azimuth (de-

30 2650 2650 2700 2800 2900 3050 3150 3200 3250 3250
gree)

60 2650 2600 2550 2500 2450 2400 2450 2550 2600 2650
90 2650 2600 2500 2300 2100 1900 1700 3700 3700 3700
Overview of Geomechanics and its applications to petroleum … 743

    




 !"#$
 !"#$

!"#$

 !"#$*
    
  

Fig 1. Relationship between minimums overbalance pressure and well trajectory applying
Mohr-Coulomb criterion.
The figure 1 shows similarly the result with [2] for the borehole inclination from 0
to 600. However, it is interesting to observe that somehow the overbalance pressure
with the well of azimuth of 900 increased significantly in this study compared to oth-
ers which was not seen in [2] for all three considered failure criteria. In this case,
some tests and the verification of other criteria are required to verify the applicability
of the Mohr-Coulomb criterion and its potential overestimation. In addition, it is im-
portant to note that the Mohr-Coulomb criterion can underestimate the rock strength
as well as neglect the potential effect of overburden pressure because it is ignoring
effect of the intermediate principal stress. The Mogi-Coulomb and the 3D Hoek-
Brown criteria can be the good candidates in this case study.

4 Conclusion

This paper provided a conceptual framework concerning geomechanics and its value
in petroleum operations. Following some principle points of this study:
The well inclination from 00 to 600 with Azimuth of 900 was observed as
the safest drilling direction.
The Mohr – Coulomb criterion can be used for the wells with the borehole
inclination from 00 to 600, more tests and other criteria are needed to study
with the range of borehole inclination between 600 to 900.

References
1. IHRDC. Petroleum geomechanics training course (2017).
2. Van Hung Nguyen, Luong Hai Linh, Minh Hoang Truong, Huu Truong Nguyen, Vu The
Quang, Nguyen Viet Khoi Nguyen, and Bui Tu An: Application of a geomechanical model
744 N. V. Hung

to wellbore stability analysis: A case study X-well, Bach Ho field in Vietnam, lecture
notes in civil engineering, Vol. 18 (2019).
3. R. Reza, R. Nygaard. Comparison of rock failure criteria in predicting borehole shear fail-
ure. International journal of rock mechanics & Mining sciences 79: 29-40 (2015).
4. Rahimi R., R. Nygaard. What difference does selection of failure criteria make in wellbore
stability analysis. ARMA 14-7146 (2014).
5. Mark D. Zoback. Reservoir geomechanics. Cambridge University Press, New York
(2007).
6. Jun Xie, Kaibin Qiu, Bing Zhong, Yuanwei Pan, Xuewen Shi, Lizhi Wang. Construction
of a 3D geomechanical model for development of a shale gas reservoir in the Sichuan ba-
sin. SPE 187828 (2018).
7. Adel Al-Ajmi. Wellbore stability analysis based on a new true-triaxial failure criterion.
Thesis, KTH Land and Water Resources Engineering (2006).
 Numerical model of hydro±mechanical coupling DEM±
PFV and application for simulation of settlement of soil
saturated in embankment due to static loading

Anh Tuan TONG1and Nguyen Hoang Phuong LUONG2

1
University of Transport and Communications, 3 CauGiay Street, Hanoi, Vietnam
2
Dong A University, 33 Xovietnghetinh Road, Danang, Vietnam,
tatuan@utc.edu.vn, hoangphuongtb@gmail.com

Abstract. This paper presents the numerical model of hydro±mechanical cou-

pling DEM±PFV using a combination of the discrete element method (DEM)
for the solid phase and a pore±scale finite volume (PFV) of the flow problem.
Numerical results at the microscopic scale on settlement of an embankment
with soil saturated are presented denoting that the capable apply of the model
DEM ± PFV for predicting the settlement of granular material used in road em-
bankments subjected to static loading. Comparison between the numerical re-
VXOW DQG WKH 7HU]DJKL¶V DQalytical solution of a soil consolidation problem
shows the validation of the model DEM ± PFV.

Keywords: Numerical model DEM±PFV, settlement, soil saturated, consolida-

tion.

1 Introduction

Compared with the Finite Element Methods, modeling the flow of pore fluid by Dis-
crete Element Method would allow a better understanding of the complex aspects of
flow in porous media, and also the interaction of flow in porous materials [1, 2] with
the mechanical behavior of granular materials at the microscale [3, 4]. DEM method
has been developing in the last few decades [5] and is most commonly applied in the
study of mechanical behavior of granular materials [6, 7, 8, 9, 10], however, this
method does not get used very often in Vietnam.
Numerical modeling of hydromechanical coupling of DEM method and flow in
porous media have attracted attention of the scientific community in the field of
geomaterials, soil mechanics, hydraulics, permeability [8], internal erosion [9, 10],
settlement consolidation [6, 7], granular seabed, bedload transport [7]. Coupled hy-
dromechanical models developed essentially differ in the modeling techniques
adopted for the fluid phase, and they can be sorted in three families: (i) Microscale
flow modeling based on a poral space discretization by FEM method [1] or Lat-
ticeBoltzmann scheme (LB) [11]; (ii) Continuumbased models in which flow and
solidfluid interactions are defined with simplified semiempirical models based on

© Springer Nature Singapore Pte Ltd. 2020 745

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_118
746 A. T. Tong and N.H.P . Luong

Darcy law [7, 9]; and (iii) Porenetwork models developed to predict the permeability
of materials and multiphase flow [1, 3].
In the first section of this study, we give the principles of the DEM and the govern-
ing equations of the DEMPFV coupling [2, 7-10] ± an open source software is de-
veloped in Lab 3SR1, written by C++ and Python, and based on Linux system. Appli-
cation of the hydro±mechanical coupling DEM±PFV for simulation of settlement of
soil saturated in embankment due to static loading, and numerical results are then
compared with solution of Terzaghi oedometric consolidation in the second section.

2 Coupled DEM  PFV model

2.1 Solid phase (Discrete Element Model)

The DEM is essentially a free mesh Lagrangian technique where each particle of the
material is a sphere identified by its own mass, radius and moment of inertia. For each
time step of the computation, interaction forces between particles and consequently
WKH UHVXOWLQJ IRUFHV DFWLQJ RQ HDFK RI WKHP DUH GHGXFHG IURP VSKHUHV¶SRVLWLRQV
WKURXJK WKH LQWHUDFWLRQ ODZ 1HZWRQ¶V VHFRQG ODZ LV WKHQ LQWHJUDWed through an ex-
plicit secondorder finite difference scheme to calculate the new position of spheres.

Fig. 1. Elastic-plastic contact model

Linear elastic law provides the contact force as a function of this relative displace-
ment between two interacting grains. The normal forceࢌ௡ and the shear force ࢌ௧ are
proportional to the normal displacement ࢁ௡ and tangential displacement ࢁሶ ௧ between
two grains, depend on the normal stiffness݇௡ and the tangential stiffness ݇௧ through
the particle radius and the characteristic elastic modulus ‫ܧ‬of the material (see Fig. 1).
Acceleration of movementࢄሷ௞ ൌ ‫࢞ۃ‬ሷ ௞ ǡ Zሷ ௞ ‫ۄ‬of a particle ݇obtained by applying the
1HZWRQ¶VVHFRQGODZRIPRWLRQ
௡ ௖ǡ௞
݉௞ ࢞ሷ ௞ ൌ ‫׬‬డ‫׊‬ೖ ߪௌ ࢔݀ܵ ൅ ‫׬‬డ‫׊‬ೖ ߩௌ ࢍ݀‫׊‬ൌ σ௝ୀ଴
೎
ࢌ ൅ ݉௞ ࢍ ൌ ࡲ௖ǡ௞ ൅ ݉௞ ࢍ (1)
௝

Where: ݉௞ mass of particle݇, ‫׊‬௞ volume of particle݇, ߪௌ ࢔ is pressure effect on the

surface particle in the direction of unit vector ࢔, ߩௌ density of particle, ࢍacceleration
of gravity; ࢌ௖ǡ௞ ௖ǡ௞
௝ interactions force between particles݆and݇, ࡲ total force applied to
particle ݇and ݊௖ number of interactions of particle ݇.

1
Laboratory Sols, Solides, Structures and Risques, France
Numerical model of hydro–mechanical coupling DEM … 747

With ͸‫ ܰݔ‬degrees of freedom of translational and rotational motion, position vec-

tor was written by matrix:

ࢄሷ ൌ ࡶିଵ ሺࢀ௖ ൅ डሻ (2)

Where ࡶ moment of inertial, ࢀ௖ force vector, डreceived by adding to mass matrix

ࡹcomponents of rotational motion, and ܰnumber of particles for the simulation.
Particles positionࢄ௧ାο௧ are updated at each time step ο‫ݐ‬by a discretization of ac-
celeration with a centered second order centered finite difference scheme:
ࢄ೟శο೟ ିଶࢄ࢚ ାࢄ೟షο೟
ൌ ࡶିଵ ሺࢀ௖ ൅ डሻ (3)
ο௧ మ

Once the new position for each particle is computed, the list of interactions is updated
and new interaction forces are computed at each time stepο‫ݐ‬:
ࡲ௖௧ାο௧ ൌ ࡲࢉ࢚ ൅ ࣜሺࢄ௧ାο௧ ǡ ࢄ௧ାο௧Ȁଶ ሻο‫ݐ‬
൝ ࢄ೟శο೟ ିࢄ೟

(4)
ࢄ௧ାο௧Ȁଶ ൌ
ο௧

DEM used the computation cycle in two steps, computation of particles position and
computation of interactions forces.

2.2 Flow phase (Pore-scale Finite Volume)

Pore-scale space discretization. Pore bodies and their connections are defined local-
ly through a regular Delaunay triangulation. (see Fig. 2).

Fig. 2. Adjacent tetrahedra in the regular triangulation and dual Voronoi network, in two
dimensions [9]

Continuity equation. The rate of volume change of the pore equals to the sum of
fluxes ‫ݍ‬௜௝ exchanged between adjacent tetrahedral through the facet ܵ௜௝ :
ሶ
௙ ௝ర ௝ర
ܸప ൌ σ௝ୀ௝ ‫ ׬‬೑ ሺ࢛௙ െ ࢛௖ ሻ ݀ܵ ൌ σ௝ୀ௝
భ ௌ
‫ݍ‬
భ ௜௝
(5)
೔ೕ
Where: ሺ࢛௙ െ ࢛௖ ሻis the velocity of the fluid relative to that of the solid.
)URP6WRNHVHTXDWLRQ'DUF\¶VODZDQG+DJHQPoiseuille relation, the relation be-
tween fluxes ‫ݍ‬௜௝ with the rate of volume change of a tetrahedral and the pressure field
ሼ‫݌‬௞ ሽcan be written as [10]:
ሶ
௙ ௝ర ௣೔ ି௣ೕ
ܸప ൌ σ௝ୀ௝ ‫ ݍ‬ൌ ݇௜௝
భ ௜௝
ൌ ‫ܭ‬௜௝ ሺ‫݌‬௜ െ ‫݌‬௝ ሻ (6)
௟೔ೕ
748 A. T. Tong and N.H.P . Luong

Forces on solid particles. The total force exerted by the fluid on a particle results
from the pressure field and can be evaluated by summing three terms: (i) the buoyan-
cy force ࡲ௕ǡ௞ , (ii) the piezometric pressureࡲ௣ǡ௞ , and (iii) the viscous shear stressࡲఔǡ௞ :
ࡲ௞ ൌ ‫׬‬డ‫׊‬ೖ ߩ݃‫ ܵ݀࢔ݖ‬൅ ‫׬‬డ‫׊‬ೖ ‫ ܵ݀࢔ כ݌‬൅ ‫ ׬‬ೖ ߬࢔݀ܵ ൌ ࡲ௕ǡ௞ ൅ ࡲ௣ǡ௞ +ࡲఔǡ௞ (7)
డ‫׊‬
With: ࢔the unit normal of the contour.
In each step time, the rate of volume change of the pores is computed, the pressure
field obtained by eq. 6, and forces are computed for the next step from eq. 7.

2.3 Coupled DEM-PFV model

Solution of the coupled DEM-PFV model is based on two matrix relations linking the
generalized displacements to the pressure field:
ሾࡳሿሼࡼሽ ൌ ሾࡱሿ൛ࢄሶൟ ൅ ൛ࡽ௤ ൟ +൛ࡽ௣ ൟ (8)
ሷ ௖
ሾࡹሿ൛ࢄൟ ൌ ሼࡲ ሽ ൅ ሼडሽ ൅ ሼࡲ ሽ ௙
(9)
Where: eq. 8 corresponds to the mass conservation equation, and eq. 9 describes the
movement of solid phase.
Pressure field will be obtained by a first explicit order scheme from eq. 10:
ࢄሺ௧ିο௧ሻିࢄሺ௧ሻ
ሾࡳሿሼࡼሺ‫ݐ‬ሻሽ ൌ ሾࡱሿ ቄ ቅ ൅ ൛ࡽ௤ ሺ‫ݐ‬ሻൟ +൛ࡽ௣ ሺ‫ݐ‬ሻൟ (10)
ο௧
The solid  fluid interactions are computed in equation (11):
ሼࡲ௙ ሽ ൌ ሾࡿሿ ሼࡼሽ (11)
Finally, an explicit ordinary differential equation of second order scheme is solved by
eq. 9.

3 Simulation of settlement of soil saturated in embankment due

to static loading by DEM ± PFV model

Numerical setup: (i) By generating a dense sphere packings in a cube; (ii) imposing
pressure at the top and bottom boundaries; (iii) imposing hydraulic gradient and static
loading at the top and bottom boundaries; (iv) no-flux and no-slip conditions are im-
posed on the lateral boundaries; (v) measure the settlement deformation. Fig. 3 intro-
duces boundary conditions of pressure; the walls are rigid and impermeable; Voronoi
diagram; and display the pressure field with ܰ ൌ ͷͲͲͲ.

Fig. 3. Voronoi diagram and boundary conditions of force and pressure [10]
Numerical model of hydro–mechanical coupling DEM … 749

The main parameters of the numerical test:ܰ ൌ ͷͲͲͲ, static loading ߪ ൌ ͷͲͲͲܲܽ
and hydraulic conductivity݇ ൌ ͵‫݉ ଻ିͲͳݔ‬Τ‫ ݏ‬. The evolution of settlement in space
and time obtained from the simulation is shown on Fig. 4(left), where: ܵsettlement in
time, ܵ௙௜௡ the final settlement, and ܶ௩ a non-dimensional time parameter. The evolu-
tion of strain normalized to its final value ߝ Τߝ௙௜௡ . The numerical result was in good
DJUHHPHQWZLWK7HU]DJKL¶VDQDO\WLFDOVROXWLRQLQWHUPVRIHYROXWLRQRIWKHVHWWOHPHQW
in time. For the saturated compression test, the final strain is ߝ௙௜௡ ൌ ͲǤͲͺΨ, the
oedometric modulus is ‫ ܧ‬ൎ ͷǤͻ‫ܽܲܯ‬, and hydraulic conductivity ݇ ൎ ͹‫ିͲͳݔ‬ହ ݉Τ‫ ݏ‬.
The evolution of the settlement is found to be in good agreement with the analytical
solution (see Fig. 4(right)).

Fig. 4. Comparisons between the numerical results of settlement and 7HU]DJKL¶VDQDO\WLFDO

solution.

The evolution of pore pressure in space, ‫ݖ‬Τ‫ ܪ‬and in time, ܶ௩ is shown on Fig. 5. The
pore pressures results are normalized ܲ Τܲ ௠௔௫ to the external stress value ܲ௠௔௫ . The
evolution in space and time of the pore pressure is found to be in good agreement
with the analytical solution of Terzaghi (see Fig. 5).

Fig. 5. &RPSDULVRQVEHWZHHQWKHSRUHSUHVVXUHDQG7HU]DJKL¶VDQDO\WLFDO solution

750 A. T. Tong and N.H.P . Luong

4 Conclusion

The DEM-PFV model was applied to the simulation of the settlement of soil saturated
in embankment due to static loading. The evolution of settlement and pore pressure in
space and time were analysed and discussed. The final strain could be measured is
ߝ௙௜௡ ൌ ͲǤͲͺΨ for the saturated compression test. In this study, a pore-scale
hydromechanical model for geomaterials was presented for reliable and efficient
computations of the contact forces exchange between fluid phase and solid grains.
The pore-network model that was developped is now a relevant approach to interprete
the more physical-complex phenomena that take place at the micro scale. Current
works focus on applications of the model to problems concerning internal erosion and
transport of fines and clogging phenomena, and hydrofracturing.

References
1. Bonilla R.R.O.: Numerical simulation of undrained granular media. PhD Thesis, Universi-
ty of Waterloo (2004).
2. Catalano E., Chareyre B., Cortis A., Barthélemy E.: A Pore-Scale Hydro-Mechanical Cou-
pled Model for Geomaterials Investigation of internal erosion processes using a coupled
DEM-Fluid method. In: Particles 2011, II International Conference on Particle-Based
Methods, Fundamentals and Applications, 2628 October (2011).
3. DiRenzo A., DiMaio F.P.: Comparison of contact-force models for the simulation of colli-
sions in DEM-based granular flow codes. Chemical Engineering Science 59(3), 525541
(2004).
4. Jing L., Ma Y, Fang Z.: Modeling of fluid flow and solid deformation for fractured rocks
with discontinuous deformation analysis (DDA) method. International Journal of Rock
Mechanics and Mining Sciences 38(3), 343355 (2001).
5. Cundall P.A., Strack O.D.L.: A discrete numerical model for granular assemblies.
Geotechnique 29, 4765 (1979).
6. Bryant S., Blunt M.: Prediction of relative permeability in simple porous media. Physical
Review A 46(4), 20042011 (1992).
7. Chareyre B., Cortis A., Catalano E., Barthélemy E.: Pore-Scale Modeling of Viscous Flow
and Induced Forces in Dense Sphere Packings. Transport in Porous Media 92(2), 473493
(2012).
8. Tong A.-T., Catalano E., and Chareyre B.: Pore-Scale Flow Simulations: Model Predic-
tions Compared with Experiments on Bi-Dispersed Granular Assemblies. Oil & Gas Sci-
ence and Technology ± 5HYXHG¶,)3(QHUJLHVQRXYHOOHV±752 (2012).
9. Tong A.-T.: Modélisationmicromécanique des couplageshydromécaniqueset des
PpFDQLVPHVG¶pURVLRQLQWHUQHGDQVOHVRXYUDJHVK\GUDXOLTXHVhttp://www.theses.fr (2014).
10. Tong Anh Tuan and Bruno Chareyre.: Micromechanical modelling of internal erosion by
suffusion using DEM-PFV coupled model. In: Vietnam ± Japan Workshop on Estuaries
Coasts and Rivers, Hoi An, Vietnam, 7-8 Sept (2015).
11. Ma J., Wu K., Jiang Z.: An implementation for lattice Boltzmann simulation in low-
porosity porous media. Physical Review E 81(5) (2010).
Abnormal pore pressure and fracture pressure prediction
for Miocene reservoir rock, field X in Vietnam

Nguyen Van Hung

Petrovietnam University, Vietnam

Faculty of petroleum, Department petroleum engineering, Vietnam
hungnv@pvu.edu.vn

Abstract. In order to drill a well safely and economically, it is necessary to

know the pore pressure and fracture pressure so that the mud density can be op-
timized to provide sufficient overbalance. In areas where exploration and pro-
duction histories are established, offset data sets can be used to provide detailed
profiles of expected pressure for those wells about to be drilled. Seismic data,
log information and direct pressure measurements, production testing can be
used. This article focuses on abnormal pore pressure and fracture pressure pre-
diction by using modern methods and industry accepted concepts. Relationships
between petroleum geology and drilling engineering are interpreted to give the
accurate estimations of Miocene rock reservoir, field X in Vietnam. The results
show that there is an abnormal pressure zone appearing from the depth of
4900ft that selecting a favorable mud density is a good solution to ensure the
success and safety of the well drilling process.

Keywords: Pore pressure, abnormal pressure, fracture pressure, Leak-off test.

1 Introduction

The pore formation pressure and fracture pressure are two of the most crucial parame-
ters to build geomechanical Earth Models (GEM) for wellbore stability analysis in-
cluding: mud density selection, reservoir compaction. From this application, two pa-
rameters are also used to evaluate the true in-situ stresses helping the construction of
casing points selected [1, 2]. An inaccurate formation pressure and fracture pressure
prediction could lead to blowout, wellbore instability, which will cause huge econom-
ic losses, delay the drilling cycle and even affect the process of oilfield development.
In many instances, formation pressure is encountered that is greater than the normal
pressure or hydrostatic pressure at a given depth. The term abnormal pore pressure is
used to describe formation pressures that are greater than normal, in this case it is
overpressure. Abnormally low formation pressures also are encountered, and the term
subnormal formation pressure is used to describe these pressures. In both planning
and drilling a well, it is important to establish the fluid pressure, especially the engi-
neer must first determine whether abnormal pressures will be present. If they will be,
the depth at which the fluid pressures depart from normal and the magnitude of the
pressures must be estimated also. However, it is difficult for establishing a method of

© Springer Nature Singapore Pte Ltd. 2020 751

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_119
752 N. V. Hung

accurately providing this information when it is needed most urgently. In general, the
formation pore pressure data should be collected and available as early as possible.
But the cost of direct measurement is very expensive, and the measurement is possi-
ble only after the formation has been drilled. Thus, the drilling engineer generally is
forced to depend on indirect estimates of formation pressure [3].
Fracture pressure is the pressure in the wellbore at which a formation will crack. A
formation will fracture when the pressure in the borehole exceeds the least of the
stresses within the rock structure. Normally, these fractures will propagate in a direc-
tion perpendicular to the least principal stress [4]. The role of evaluating this pressure
is just as important as knowledge of how the formation pore pressure varies with
depth when planning and drilling a well. It is of necessity to carry out breakdown tests
to verify the fracture pressure of a specific zone such as oil and gas reservoir. The
pressure behavior during a test is recorded to determine fracture pressure by using
surface leak-off pressure [5].

2 Literature review

Lots of methods for detecting and anticipating abnormal formation pressure are based
on the fact that formations with abnormal pressure also tend to be less compacted and
have a high porosity than similar formations with normal pressure at the same burial
depth. There are three techniques for detecting and estimating abnormal formation
pressure: (1) predictive methods, (2) methods applicable while drilling, and (3) verifi-
cation methods [2]. On the one hand, for wildcat wells only seismic data may be
available and pressure information is obtained by a predictive method. On the other
hand, for development wells emphasis is placed on data from previous drilling experi-
ences in the area. Both type of wells, pressure information must be updated constantly
during drilling and checked again before casing is set, using various formation evalua-
tion methods.
Geoscientists, such as reservoir and processing geophysicists, are more involved in
the application of processes to do pore pressure prediction based on seismic in 3-D. In
this process, deriving accurate possible velocities from seismic processing is one of
the key factors in predicting reliable pressure. Thus, the information of pore pressure
is extremely valuable in trying to predict the pressure cube. Besides extracting the
forecast pressure profiles at future drilling locations from these volumes, they have an
interpretation value, for instance, providing alternative well locations, see how the
pressure regime is distributed [6].
Hottman and Johnson [7] presented one of the first empirical relationships between
formation pressure and formation resistivity and acoustic travel time/velocity. They
indicated that porosity decreases as a function of depth from analyzing acoustic travel
time in Miocene and Oligocene shales in Upper Texas and Southern Louisiana Gulf
Coast [7, 8]. This technique has received wide acceptance even though the prediction
charts were based only on data concerning Tertiary age sediments in the Gulf Coast
area. Later on, many empirical equations for pore pressure prediction were presented
on resistivity, sonic transit time (interval velocity) and other well logging data.
Abnormal pore pressure and fracture … 753

The main aim of this work is to find the most suitable method among those men-
tioned in literature review section to calculate the pore pressure and fracture pressure
in two case studies that have interval transit time computed from seismic measure-
ment.

3 Data review

This study illustrates the estimation of pore pressure gradient and fracture pressure
gradient from seismic data, the average acoustic velocity as a function of depth can be
determined from interval transit time.

Table 1: Average interval transit time data computed from seismic records in proposed X-well,
Miocene reservoir rock, field X, Vietnam.

Depth inter- Time Depth interval Average

Dav (ft) Time(10-6s/ft)
val (ft) (106s/ft) (ft) depth (ft)
1500-2500 2000 155 7500-8500 8000 121
2500-3500 3000 130 8500-9500 9000 126
3500-4500 4000 110 9500-10500 10000 124
4500-5500 5000 95 10500-11500 11000 125
5500-6500 6000 110 11500-12500 12000 124
6500-7500 7000 116

The average interval transit time data was computed from seismic records obtained
in normally pressured sediments in South Texas Frio Trend reservoir and X-well, Mio-
cene reservoir rock, field X, Vietnam occurring in the upper Miocene trend, in the [3]
and table 1 respectively, which are also plotted in Figure 1. The normal hydrostatic
formation pore pressure determined from the salinity data for South Texas Frio Trend
is 0.465psi/ft, and the overburden pressure gradient is around 1psi/ft as reported in
[3]. In addition, with proposed X-well, Miocene reservoir rock, field X, Vietnam, the
pore pressure gradient and overburden pressure gradient were approximately
0.586psi/ft, and 0.954psi/ft respectively as reported in [9].

4 Result and discussion

The figure 1 show the seismic transit travel time vs. depth in semi log papers for X-
well, Miocene reservoir rock, field X, Vietnam. Most methods for detecting abnormal
formation pressure are based on the fact that formations with abnormal pressure also
tend to be less compacted and have a higher porosity or interval transit time than simi-
lar formations with normal pressure at the same burial depth [3]. Thus, any measure-
ment that reflects changes in porosity or interval transit time also can be used to detect
abnormal pressure. It can be seen in the figure 1 that there is a significant decrease of
interval transit time with an increased depth of burial and compaction. The departure
from the normal pressure trend means a transition into abnormal pressure that is
754 N. V. Hung

8000ft for South Texas Frio Trend reservoir, and 4900ft for proposed X-well, Miocene
reservoir rock, field X, Vietnam. In reality, the casing must be set in the well at the
departure of transition zone before drilling next section.
Table 2: Pore pressure gradient and fracture pressure gradient calculation for the proposed
X-well, Miocene reservoir rock, field X, Vietnam.

Depth interval Time Pf prediction Pff gradient

Dav (ft) Dav (ft)
(ft) (10-6s/ft) (psi/ft) (psi/ft)

1500-2500 2000 155 2000 0.586 0.709

2500-3500 3000 130 3000 0.586 0.709
3500-4500 4000 110 4000 0.586 0.709
4500-5500 5000 95 4800 0.601 0.718
5500-6500 6000 110 3900 0.715 0.795
6500-7500 7000 116 3250 0.783 0.840
7500-8500 8000 121 3300 0.802 0.853
8500-9500 9000 126 3300 0.819 0.864
9500-10500 10000 124 3300 0.833 0.873
10500-11500 11000 125 3300 0.844 0.880
11500-12500 12000 124 3400 0.850 0.884

The table 2 and figure 2 show the final result of pore pressure gradient and fracture
pressure gradient. It is clear to say that there is very narrow mud window for both
case studies in the abnormal pressure zone as shown in figure 2. This result has a
sufficient effect on the well planning and the casing set within the well planning is
8000ft and 4900ft for South Texas Frio Trend reservoir and X-well, Miocene reservoir
rock, field X, Vietnam respectively.
After detecting abnormal formation pressure zone, the equivalent depth was deter-
mined with the assumption that formations having the same value of interval transit
time are under the same vertical effective matrix stress. For instance, in the figure 1,
at 9.000ft the interval transit time has a value of 125x10-6s/ft that is also same value
at 2100ft on the normally hydrostatic pressure line. The new values of overburden
pressure, pore pressure and effective stress can be estimated from the equivalent depth
value in the table 2. From this explanation, the overburden pressure, effective stress
and pore pressure gradient at the equivalent depth of 9000ft are 2100psi, 1123.5psi,
0.875psi and 4579.2psi, 1766.1psi, 0.691psi/ft for South Texas Frio Trend reservoir
and X-well, Miocene reservoir rock, field X, Vietnam respectively. The method to
determine formation fracture pressure gradient based on available data. In this study,
Hubbert and Willis method are used for pore pressure gradient calculation and also
fracture gradient, FG=(OBG+2Pf)/3D [10].
Abnormal pore pressure and fracture … 755

Fig 1. Average interval transit time data computed
from seismic records in proposed X-well, Mio-
cene reservoir rock, field X, Vietnam.

Fig 2. Pore pressure gradient and frac-
ture gradient of the proposed X-well,
Miocene reservoir rock, field X, Vietnam.

For the recommendation of mud weight, it can be noted that mud weight should be
selected between 0.465psi/ft – 0.643psi/ft for normally pressure zone and 0.875psi/ft-
0.938psi/ft for South Texas Frio Trend reservoir. Similarly, the figure for proposed X-
well, Miocene reservoir rock, field X, Vietnam is 0.586psi/ft-0.709psi/ft, and
0.601psi/ft-0.884psi/ft. However, it is clear that the predicting and estimating for-
mation pore pressure and fracture pressure from seismic data is more difficult in older
sedimentary basins that generally have a much more complex lithology. The change
in depth characterized by a major change in lithology should be verified while drilling
and direct measurement. To detect the change of formation pressure, drilling parame-
ters related to bit performance need monitor continuously and record by surface in-
struments. In addition, many variables associated with the drilling fluid and rock
fragments being circulated from the well also required monitored carefully and logged
using special mud logging equipment and personnel.

5 Conclusions

In conclusion, the pore pressure and fracture pressure can be measured indirectly and
directly. For a proposed well such as a well cat drilling, planning a well for drilling
operation has only seismic data and therefore, the method “equivalent matrix stress”
should be applied. Attempts in this study were made to verify the validity of equiva-
lent matrix stress by comparing the empirically derived normal compaction trend line
to an independently based prediction of interval transit time versus depth for normally
compacted sediments. The attempts did not provide confirmation of the validity of the
method used. In addition, it was observed that:
756 N. V. Hung

The hydraulic pressure gradient of mud column in wellbore prediction were

generally between 0.465psi/ft – 0.643psi/ft for normally pressure zone and
0.875psi/ft-0.938psi/ft for South Texas Frio Trend reservoir. Similarly, the
results for proposed X-well, Miocene reservoir rock, field X, Vietnam are
0.586psi/ft-0.709psi/ft, and 0.601psi/ft-0.884psi/ft.
Recommendation for casing set within the well planning is 8000ft and 4900ft
for South Texas Frio Trend reservoir and Miocene reservoir rock, field X,
Vietnam, respectively where the abnormal pressure would be appeared.
However, the pore pressure and fracture pressure predictions are uncertain because
of limited existing data. In fact, the petrophysical data is needed for the verification
and validation of the considered method. Moreover, realistic pore pressure is also
required to make reliable fracture gradient calculations.

Acknowledgments

This work is funded by Petrovietnam University under grant code GV1901, we

would like to thank to the board of manager of PVU for the support.

References

1. Bernt Aadnoy, Reza Looyeh: Petroleum rock mechanics – Drilling operations and well de-
sign. Gulf professional publishing, p.376 (2011).
2. Nguyen Van Hung, Luong Hai Linh, Truong Minh Hoang, Nguyen Huu Truong, Nguyen
Viet Khoi Nguyen, Bui Tu An: Application of a geomechanical model to wellbore stability
analysis-A case study X-well, Bach Ho field in Vietnam, VSOE conference. Lecture notes
in Civil engineering, Vol. 18. Springer, Singapore (2018).
3. Jr. Adam T. Bourgoyne, Keith K. Millheim, Martin E. Chenevert, Jr, F. S. Young: Applied
drilling engineering. SPE, p. 502 (1986).
4. F. Khoshnaw, P. Jaf, S. Farkha: Pore, abnormal formation and fracture pressure prediction.
WIT Transctions on ecology and the environment, Vol 186 (2014).
5. M.E. Chenevert: How to run casing and open hole pressure tests. Oil and gas journal
(1978).
6. Bikas Kumar, Sri Niwas and Bikram K. Mangaraj: Pore pressure prediction from well logs
and seismic data. 9th Biennial international conference & exposition on petroleum geo-
physics. Hyderabad, P005 (2012).
7. C.E. Hottmann, R.K. Johnson: Estimation of formation pressure from log derived shale
properties, SPE 1110 JPT 17 (717-722) (1965).
8. Jincai Zhang: Pore prediction from well logs – methods, modifications, and new approach-
es, Earth science reviews 108 (50-63) (2011).
9. Nguyen Thi Thanh Binh, Tomochika Tokunaga, Neil R. Goulty, Hoang Phuoc Son, Mai
Van Binh: Stress state in the Cuu Long and Nam Con Son basins, offshore Vietnam, Ma-
rine and petroleum geology 28 (973-979) (2011).
10. M.K. Hubber, D.G. Willis: Mechanics of hydraulic fracturing. Journal of American associ-
ation of petroleum geologists 12: 239-257 (1957).
 Study the influence of adherence edge to steel strip and
soil interaction in Mechanically Stabilized Earth Wall
with a self-made strip

CHAU Truong-Linh1 and NGUYEN Thu-Ha2

1,2
University of Science and Technology, The University of Da-Nang;
email: chau-linh@dut.udn.vn; ntha@dut.udn.vn

Abstract: Nowadays, high embankments are popular in many transportation

networks. A method of Mechanically Stabilized Earth Wall (MSE wall) has
been used for increasing the stability of backfilling soil in road embankments
soil because of its economic, technical and landscape advantages. This paper
presents a series of experiments on different categories of adherence edge to
demonstrate the effective of edge on strain, stress-strain, and displacement of
MSE wall. A FLAC software also was used to verify the experimental results.
The study could provide the instructions for the design, construction and project
management of MSE wall. It can also be used for suggestion of soil
improvement solutions, backfilling material selections to increasing the stability
of MSE wall.

Keywords: Mechanically Stabilized Earth Wall, backfill material, adherence

edge, physical modelling, numerical modelling

1 Introduction

Currently, all kinds of steel strip used for retaining walls in Vietnam are imported at
very high prices which leads to increase in the cost of construction [1], [2]. It is clear
that using the steel strips could save the costs and take advantage of the production
capacity of the domestic enterprises. The research team proposed the self-made steel
strip reinforcement for MSE wall. However, the internal stability of this wall depends
on the interaction between soil and the reinforcing elements. This paper presents the
study on the effect of adherence edge to soil and strip interaction in MSE wall with
self-made reinforcing strips.

2 Literature review on the soil and steel strips interaction in

MSE wall

The reinforcement has an important role of creating horizontal pressure from the
reinforced soil. The pressure creates a larger cohesive force within the soil [3], [4],

© Springer Nature Singapore Pte Ltd. 2020 757

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_120
758 T.-L. Chau and T.-H. Nguyen

[5]. The reaction of soil and reinforcing strips mainly depend on two parameters,
included soil and reinforcing strips friction and tensile strength of the strips.
When the soil is compressed vertically, if there is no steel strip reinforcement, the
soil will be damaged because of the free horizontal expansion. In other hands, using
strip reinforcement and assumption that between the soil and strip reinforcement is
sufficient anchorage during compression process. Then, the soil and reinforcing strips
will combine to resist the force. In this case, the soil is considered to be triaxial
compression with limited horizontal expansion. The horizontal pressure value is the
force acting on the soil through the friction between the soil and reinforcement, which
is denoted as μ (μ*) or f (f*) [2], [6], [7].
The coefficient of friction (μ) between the surface of the building material with the
sand and silt soil is usually by (0.5-0.6) times the internal friction coefficient of the
soil [7]. During the deformation process under shear force, the soil may be expanded,
the effective pressure on the reinforcement surface will change depending on the
interaction between soil and reinforcing strip. So, the friction coefficient determined
by experiment is used: μ*=F/(2b.L.σv), where: L is the reinforcement length; F is the
tension load in the streel strip reinforcement; b is the equivalent width of the surface
in contact with the soil; σv is the vertical distribution pressure.

3 Study the influence of adherence edge on the interaction

between soil and steel strips

3.1 Physical model

A physical model in a laboratory with a 1/1 scale which is similar to the actual size of
the construction was designed (Fig.1a). Normally, for reinforced concrete slabs are
used two layers of steel strips with the size of 1.5x1.5 m. In this experimental model,
one layer of steel strips was set up between two layers of soil, in which each soil layer
thickness is 37.5 cm (considering with 1/2 reinforced concrete wall and wall surface
around 75 cm in height). The model box is made with the size of 150x75x75 cm, and
the soil block behind the wall is simulated with the size 120x75x75 cm. A steel frame
was make for the loading process by hydraulic jacks.

a. b. c.
Fig.1. a. Experimental model of MSE wall; b. Steel strip 2; c. Steel strip 3
Study the influence of adherence edge to steel strip … 759

The strip is CT5 Φ10 steel with elastic modulus E = 210000 MPa, consisting of
three parallel bars with the length of 110 cm and spacing range about 30 cm. Three
horizontal bars were placed 40 cm apart. The distance from the anchor to the wall to
the first horizontal bar is 20 cm. The adherence edge (CT5 Φ10) are located at the
intersection and the height of the adherence edge is 5 cm. “Steel strip 1”: Φ10 no
adherence edge; “Steel strip 2”: Φ10 with three adherence edges (Fig.1b); “Steel strip
3”: Φ10 with five adherence edges (Fig.1c).The vertical bars connected to a hardwood
panel (assumed to be a retaining wall) with the size 75x75x5 cm (Fig.2a and Fig.2b).

a. b.
Fig.2. a. The structure of the wall; b. Steel strip 3 and strain gause in the soil

a. b. c. d.
Fig.3. a. Prepare soil samples to reach humidity W0; b. Compacted soil; c. Installation of
displacement transducers; d. Check the signal, connect the straingase to the computer

The steps to build the experimental model are as follows: The embankments were
taken at the mines in Da Nang city; Test of mechanical and physical parameters of
soil; Prepare soil samples to reach humidity W0 (Fig.3a); Backfill soil layers with
12.5cm each layer (after compaction); Installation of steel reinforcement (strain gause
have been installed at the locations shown in Fig.3b) after compaction of the third
layer (37,5cm thick); Connect the reinforcement in the wall by the bolts; Continue to
fill soil layers ebove the steel strips; Install three transducer in three positions: the
bottom, the centre and the top of the wall (Fig.3c and Fig.3d); Continue to install
hydraulic jack system and observe deformation results displacement.

3.2 The results obtained on experimental models

The coefficient of friction between soil and steel strip reinforcement changes over
strip length (Fig.4a). When pressure is increased, the coefficient of friction decreases
due to the increase of tension force in the strip and the movement between soil and
the strips. The coefficient of friction values has been changed significantly at the
location of adherence edge, it means that the adherence edge increases the friction
between the soil and steel strip reinforcement.
760 T.-L. Chau and T.-H. Nguyen

At the same pressure level and in the same location as the strip reinforcement in
the wall, "steel strip 3" has the highest friction value; while "steel strip 1" has the
lowest friction. This result shows the effectiveness of using the adherence edge in the
strip reinforcement layout (Fig.4b).

a. b.
Fig.4. a. The coefficient of friction in steel strip 3; b. Coefficient of friction when changing
types of steel strip

The form of the tension distribution chart in this research result is similar to the
tension chart studied by Chau [2], and it is also consistent with the friction chart in the
steel strip reinforcement under pressure levels (Fig.5a, Fig.5b).

a. b.
Fig.5. a. Tension distribution in steel strip 3; b. Tension distribution in the layers of Chau [2]

The tension in the strip reached the ultimate value at the location of 20 cm from the
wall and it decreases at the farther locations from the wall. In the area of adherence
edge, the tension in the strips changes (Fig.5a). Therefore, it can be said that the
application of adherence edge increases the friction between the soil and the steel
strip, leading to the increase of the anchoring capacity of the strips.
The increase of tension in the steel strips at adherence edge could be evaluated as:
Adherence edge 0 (Fig.6a): It is the starting point of mobilising the friction force
between the soil and the reinforcement and passive friction of the soil; Adherence
edge 1 (Fig.6a): The rate of increase in traction in the reinforcement at adherence
edge number 1 is the largest then decreases in the next adherence edge. It can
conclude that in the first adherence edge, the value of traction increased rapidly,
although the interface area between the soil and the steel strips is not mobilised.
Forward to the wall, the interface area increases, the value of tension in the strips
increases, the change of force at the edge locations is high but that increasing ratio is
decreased.
 Study the influence of adherence edge to steel strip … 761

3.3 Numerical models

While experimental models were built in the laboratory, the numerical models were
performed on FLAC software with the same geometric dimensions, physical and
mechanical characteristics, material characteristics of soil and of reinforcement in the
experimental models. The simulation was used to check the agreement of the data on
the physical model (Fig.6a). The results of the numerical simulation will provide the
values of stress, deformation, displacement of soil mass and wall, tension distribution
chart in the steel strip reinforcement.

a. b. c.
Fig.6. a. Simulation of MSE wall with FLAC 2D ("steel strip 3"); b. The displacement of the
wall surface in FLAC 2D; c. Results measured from the experimental model

a. b.
Fig.7. Traction chart in the steel strip reinforcement in FLAC 2D: a. Steel strip 1; b. Steel strip 3

The displacement of the retaining wall gradually increases from the bottom to the
top of the wall. Figure 6b shows the displacement of the wall in the case of using soil
from Luoi Meo soil mine. The measurement result from the experimental model (Fig.
6c) shows that when using the soil of Luoi Meo soil mine, the displacement of the
wall is smaller than that of Hoa Ninh soil mine. In addition, the numerical model
results by FLAC are consistent with the experimental model.
The tension distributions (Fig.7a and Fig.7b) are good agreement with the research
results from the experimental models and previous research of Chau [2] and Murray
and Farrar [8]. When applying five adherence edges, the tension value increases
markedly at the front and back position of the adherence edge. Moreover, at the
position of 20 cm from the wall, the tension value increases about three times in
compared to the case of using no adherence edge.

4 Conclusion

The adherence edge is highly influential to the interaction between soil and steel strip
reinforcement in MSE wall. When using adherence edge, the friction between the soil
762 T.-L. Chau and T.-H. Nguyen

and reinforcing strips increases, and the anchoring capacity of the strips, leading to
increase of internal stability for the wall.
From the relationship between the number of adherence edges, the position of
adherence edge with the coefficient of friction between soil and reinforcing strips
shows that the embankment component contributes an important role in the soil-strip
interaction. We can propose solutions to improve the soil by changing the grading
component, so that the interaction between the strip and soil interface is highest.
In addition, other research results on deformation, displacement, limit load are
considered very important. During the exploitation process, it should be set up a plan
for the tracking, included wall displacement monitoring, extracting core samples to
check corrosion rate, assess road surface condition, surface drainage system, and so
on to provide the solutions to fix the damage if the wall shows the displacement and
deformation signs exceeding the permitted level.
This is a new research direction, it could provide the foundation knowledge for
design, construction and exploitation management consultants to provide a higher
quality of MSE wall.

References
1. T-L Chau, H-S Tran, N.T-T Hồ, T-N. Nguyen: Research the effect of soil moisture to
stabilize internally the mechanically stabilized earth wall using steel
reinforcement. Journal of Transport ISSN 2354-0818. No 12/2015. pp.77-80, (2015)
2. T-L Chau, A. Corfdir, E. Bourgeois : Corrosion des armatures sur le comportement des
murs en terre armée - Effect of reinforcement corrosion on the behavior of earth walls
reinforced by steel elements (Soustitre: Scénarios de corrosion des armatures métalliques
et les dégradations du mur en terre armée). ISBN 978-3-8417-2710-7. Éditions
Universitaires Européennes (EUE), (2016)
3. Les ouvrages en terre armesée, recommendation et rèfgles de l’art. pp.54, 58, 72, (1979)
4. BS 8006: 2010: Code of practiced soils and other fills.
5. 22TCN 272-2005: Standard bridge design.
6. H.H. Duong: Design and construction of mechanically stabilized earth wall. Construction
Publisher, (2012)
7. LCPC, ed. : Guide technique - Recommandations pour l’inspection détaillée, le suivi et le
diagnostic des ouvrages de soutènement en remblai renforcé par des éléments métalliques.
pp.102. LCPC, (2003)
8. R.T.Murray, D.M. Farrar: Temperature distribution in reinforced soil retaining wall.
Geotextiles and Geomembranes, pp.159, 160, (1988)
 Formation of Rust Scales on Mild Steel Pile in Brackish
Water of Phu My Port - Vietnam

Vinh-Dat Vuong1,2,3, Anh Quang Vu1, Marina Kawai4, Kazuhito Tsujimoto4,

Thang Van Le1,2 and Nguyen Danh Thao5
1
Department of Energy Materials, Faculty of Materials Technology, HCMUT, VNU-HCM,
Vietnam
2
Material Technologies Laboratory, HCMUT, VNU-HCM, Vietnam
3
Graduate University of Science and Technology, VAST, Vietnam
4
JFE Steel Corporation, Japan
5
Department of Port and Coastal Engineering, Faculty of Civil Engineering, HCMUT,
VNU-HCM, Vietnam
ndthao@hcmut.edu.vn

Abstract. Corrosion test of mild steel (graded JIS-STK400) was investigated in

brackish water environment of Phu My industrial port (Ba Ria - Vung Tau Prov-
ince, Vietnam). Steel sample was exposed in 3 years. X-ray diffraction (XRD)
results show that formed rust powder was composed of α-FeOOH, Fe3O4 and α-
Fe2O3 in which the amount of first component is larger than second and third one.
It demonstrates that natural corrosion products evolved from low to high oxida-
tion state of iron by presence of dissolved oxygen in brackish environment. The
mechanism of phase conversion is revealed by metallographic of rust layer cross
section. α-Fe2O3 (white phase) was gathered at bank of cracking paths where was
favouring diffusion of oxygen and water into inner layer. After contacting Fe3O4
(dark gray phase), corrosive agents converted it to α-FeOOH and α-Fe2O3. The
phase conversion changes the weight density of materials and creates the porous
tunnels in rust scales. The growth of micro tunnels to form the tunnel network
allowed internal stress and cracks of rust scales evolved.

Keywords: rust peeling off, rust formation, metallography, brackish water, in-
dustrial port.

1 Introduction

Mild steel, the most widely used constructional metal, always confronts to a natural
phenomenon - corrosion. When steel surface contacts with corrosion agents of environ-
ment, iron is slowly converted to various oxidation state. Steel rust, the common name
of corrosion products, grows on steel surface by both chemical and electrochemical
processes. Most of studies existed in literatures and agreed that electrochemical corro-
sion is the role in real-life structures and building when chemical corrosion is specially

© Springer Nature Singapore Pte Ltd. 2020 763

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_121
764 V.-D. Vuong et al.

shown in industrial or technical applications. Composition of electrochemical rust de-

pends on environmental conditions, such as time of wetness (TOW), temperature, sa-
linity, dissolved oxygen… [1, 2].
Generally, Payer (1995) [3] reported that a thin (few nanometers) FeO/Fe(OH)2 layer
rapidly formed on steel surface when corrosive gas species are not present. This layer
is the natural protective barrier of mild steel to further corrosion. When relative humid-
ity of environment reached to a critical level, a thin adsorbed water layer formed on
surface and became electrolyte. This electrolyte changed protective layer on steel sur-
face to active layer which became an inner corrosion product layer with hundreds of
nanometers, lately. After that, reaction of inner layer with atmospheric agents produced
higher oxidation state of iron and formed a thicker outer layer with several micrometers
of thickness or greater. The outer layer was porous and permitted the penetration of
water and gases into under layers where corrosion process repeated continuously.

2 Experiment

2.1 Materials and Test Sample Preparation

Mild steel graded STK400 following JIS G3444. Composition of test sample was given
in Table 1. Test sample was design in O-pipe pile with L (length) = 12 m, OD (outside
diameter) = 165 mm, T (substrate thickness) = 9 mm (Fig. 1). Materials are products of
J-Spiral Steel Pipe Co. Ltd. (JSP).
Table 1. Composition of test sample.

Element C Si Mn P S Fe
%wt. 0.068 0.034 0.025 < 0.03 < 0.02 Balance

2.2 Field exposure test

Field exposure test was started at January 2014 at Phu My port (Ba Ria - Vung Tau
Province, Vietnam), which is an industrial port with complicated brackish environment.
Steel pipes were used for this test and planned for long term test up to 15 years. Fig. 1a
showed the scheme of installation. After 36 months, rust layer at lowest water level
(LWL) was collected and characterization by XRD at National Key Laboratory of Pol-
ymer and Composite Materials (PCKLab) - VNU-HCM, and metallography at Material
Technologies Laboratory (MTLab) - VNU-HCM.
Formation of Rust Scales on Mild Steel Pile … 765

Fig. 1. (a) Design of test sample and installation scheme of field exposure test; (b) X-ray diffrac-
tion pattern of rust powder and references; (c) formation and conversion of corrosion product of
carbon steel in aerobic and anaerobic environments by King (2011) [4].

3 Results and Discussion

3.1 Composition of rust powder

Fig. 1b showed X-ray diffraction of rust powder and references exported from Crystal
Impact Match!. XRD pattern was analyzed by relative intensity ratio (RIR) method.
Formed rust powder was composed of α-FeOOH (entry 96-9003078), Fe3O4 (entry 96-
900-6319) and α-Fe2O3 (entry 96-591-0083). Weight percentage of phases are related
to intensity of quantity peaks. XRD results indicated that presented amount of α-
FeOOH was larger than Fe3O4 and α-Fe2O3. Results of King (2011) [4] showed that
natural corrosion products evolved from low to high oxidation state of iron follow re-
action scheme of Fig. 1c.
At LWL, the inner corrosion product layer Fe(OH)2 was converted to α-FeOOH
faster than to Fe3O4 causing by dissolved O2 from atmosphere to water. In submerged
brackish water, absorption of O2 from outside and diffusion of Fe2+ from inner layer
generated an outer corrosion product layer through a dynamic equilibrium state of α-
FeOOH and Fe3O4. The α-FeOOH at outer layer was converted to α-Fe2O3 under drying
hours of days when Fe3O4 in inner layer was isolated with O2. The results are compa-
rable with results of other research site [2, 5-8].
766 V.-D. Vuong et al.

3.2 Metallographic of Rust Cross Section

Fig. 2 showed metallographic of rust layer cross section with different magnifications.
At 500x magnification (Fig. 2d, e, h, i, j), three phases were presented: Fe3O4 (dark
gray), α-FeOOH (light gray) and α-Fe2O3 (white), which had largest area. The white
phase α-Fe2O3 was gathered at bank of cracking paths where was absorbing tunnel of
oxygen and water into inner layer. After contacting dark gray phase Fe3O4, corrosive
agents converted that phase to α-FeOOH and α-Fe2O3. The transformation gradient of
Fe3O4 - (α-FeOOH) - (α-Fe2O3) was showed in Fig. 2d (arrow shapes) with a thin layer
of dark gray Fe3O4 (curve shape) separated two gradients. Phase conversion is concur-
rent with decrease of molecular weight from 231.5 g.mol-1 of Fe3O4 to 88.9 and 156.7
g.mol-1 of FeOOH and Fe2O3 respectively. These processes caused the decrease of
weight density and increase of porosity of rust. Thus, micro tunnels were created and
lengthened then connected together to evolve to a network. The growth of tunnel width
and network closeness increased internal stress and caused cracks of rust scales.

Fig. 2. Metallographic of rust cross section, unit: µm.

Formation of Rust Scales on Mild Steel Pile … 767

4 Conclusions

Results of XRD and metallography nominated phase conversion of corrosion products

at LWL of Phu My port. The mechanism of phase conversion is related to absorbing
and isolating process of dissolved O2 between water and rust layers. Metallographic
image also presented that growth of absorbing tunnel network is the cause of rust po-
rosity and cracks. Sequentially, appearance of micro tunnels and their width expansion
allow corrosive agents attack directly to inner layer. When length of cracks increases
and creates rust scales, the forming process can repeat cyclically until rust scales peels
off from steel surface.

Acknowledgment. The group of authors thank to JFE Steel Corporation for coopera-
tion in recent and future works.

References
1. Rodriguez, M. A.: Anticipated Degradation Modes of Metallic Engineered, The Minerals,
Metals & Materials Society 66(3), 503-525 (2014).
2. De la Fuente, D., Alcantara, J., Chico, B., Diaz, I., Jimenez, J. A., Morcillo, M.: Character-
isation of rust surfaces formed on mild steel exposed to marine atmospheres using XRD and
SEM/Micro-Raman techniques, Corrosion Science 110, 253-264 (2016).
3. Payer, J. H., Ball, G., Rickett B. I., Kim, H. S.: Role of transport properties in corrosion
product growth, Materials Science and Engineering A198, 91-102 (1995).
4. King, F., Padovani, C.: Review of the corrosion performance of selected canister materials
for disposal of UK HLW and/or spent fuel, Corrosion Engineering, Science and Technology
46(2), 82-90 (2011).
5. Li, Q. X., Wang, Z. Y., Han, W., Han, E. H.: Characterization of the rust formed on weath-
ering steel exposed to Qinghai salt lake atmosphere, Corrosion Science 50, 365-371 (2008).
6. Castle, J. E., Mann, G. M. W.: The mechanism of formation of a porous oxide film on steel,
Corrosion Science 6, 253-262 (1966).
7. Castle, J. E., Masterson, H. G.: The role of diffusion in the oxidation of mild steel in high
temperature aqueous solutions, Corrosion Science 6, 93-104 (1996).
8. Cindra Fonseca, M. P., Bastos, I. N., Caytuero, A., Baggio Saitovitch, E. M.: Rust formed
on cannons of XVIII century under two environment conditions, Corrosion Science 49,
1949-1962 (2009).
 Modelling Stress Distribution Around Boreholes

Khanh Do Quang1,*, Thanh Nguyen Thi Tam2, Phuc Kieu1, Nhan Vo

Huynh1, Quang Hoang Trong1
1
Faculty of Geology and Petroleum, Ho Chi Minh University of Technology (HCMUT),
Vietnam National University ± Ho Chi Minh City (VNU-HCM), Vietnam
2
Faculty of Heat-Refrigeration Engineering, Industrial University of Ho Chi Minh City
(IUH), Ho Chi Minh City, Vietnam
*Corresponding author: dqkhanh@hcmut.edu.vn

Abstract
The stress distribution around boreholes plays an important role for wellbore
problems. This study aims to model the stress distribution around boreholes of
offshore petroleum wells. The stress model around boreholes, which is associated
with the in-situ stresses, rock properties as well as the wellbore pressure and
configuration, is developed. The new approach uses transformation formula of a full
stress tensor including its orientations and magnitudes. A calculating program for
the stress analysis of wellbores (SAoWB), which is written in Matlab language, has
described and calculated all components of the stress tensor at the wellbore wall as
well as around boreholes. In this study, case studies are considered using the
program SAoWB based on the new approach. The field case studies are applied for
boreholes of the studied wells at Cuu Long basin, offshore Vietnam. The obtaining
results from our program SAoWB are in good agreement with the failure
observations from high solution image logs of the studied wellbore as well as coring
data. They are the bases to study its implications which benefit for offshore
petroleum industry activities such as wellbore stability, optimmum drilling wellbore
trajectories for well planning, drilling mud selection, sanding prediction and control,
hydraulic fracturing, etc. Modelling the stress distribution around boreholes from the
program SAoWB not only enhances knownledge of in-situ stress tensors but also
impacts on well problems, especially for offshore petroleum wells.

Keywords: In-Situ Stress, Stress Distribution, Stress Model, Boreholes, Failure

Observations.

1 Introduction

The rock surrounding boreholes must support the stresses previously carried by the
removed material. There will be the stress redistribution and concentration in the
vicinity of boreholes [1]. This can also result in wellbore failures if induced stresses
around boreholes overcome the formation rock strength. Consequently, knowledge of
the stresses around boreholes is essential for wellbore problems. Wellbore failures

© Springer Nature Singapore Pte Ltd. 2020 769

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_122
770 K. D. Quang et al.

may be compressive failures known as borehole breakouts (BOs) and/or tensile

failures as drilling-induced tensile fractures (DITFs) at the wellbore wall [2,3,4].
Today, many petroleum wells have complex trajectories, either horizontal or highly
deviated from vertical axis. It is necessary to understand the stresses around an
arbitrarily deviated wellbore as well as the factors controlling the occurrence of
compressive and tensile failures in this well with the arbitrary orientation. Therefore,
the stress model around boreholes is developed to describe fully and exactly stress
components for the three dimensional (3D) space.
In this work, the new approach uses transformation formula of a full stress tensor
including its orientations and magnitudes. Case studies are presented using the
program for the stress analysis of wellbores (SAoWB) based on the new approach.
The field case studies are applied for boreholes at different interest depths of the
studied petroleum wells at Cuu Long basin, offshore Vietnam. The obtaining results
from our program SAoWB are in good agreement with the failure observations from
high solution image logs of the studied wells.

2 Stress Model Around Boreholes

The calculation of the stresses around an arbitrarily inclined well requires that the far
field in-situ stress tensor is transformed into the borehole coordinate system. In this
coordinate system, the stress tensor may no longer be represented by the principal
stress magnitudes and directions for a vertical wellbore. The shear stress components
may be non-zero and the transformed stress tensor must be represented. The
transformed stress tensor is required to calculate the stress concentration around
boreholes.
In a deviated well, the principal stresses acting in the
vicinity of the wellbore wall are generally not
aligned with the wellbore axis (Fig. 1).

Fig. 1. An arbitrarily deviated wellbore with the

RULHQWDWLRQV RI WKH FLUXPIHUHQWLDO ıࣄࣄ), axial (ızz), radial
(ırr), minimum (ıtmin) and maximum (ıtmax) stresses, where
Ȧis the angle between ߬tmax and the wellbore axis [3].

To consider failures in an arbitrary well, we define three coordinate systems (Fig. 2)

as following:
- A geographic coordinate system, X, Y and Z oriented north, east and vertical
(down);
- A stress coordinate system, xs, ys and zs corresponding to the orientations of the far-
field principal stresses S1, S2 and S3;
- A wellbore coordinate system xb, yb and zb where xb is radial, pointing to the bottom
well, zb is down along the well axis and yb is orthogonal in a right-hand coordinate
system.
Modelling Stress Distribution Around Boreholes 771

Fig. 2. Three coordinate systems used to transform for

an arbitrarily deviated wellbore with WKH D]LPXWK į DQG
LQFOLQDWLRQij[3].

To most easily visualize wellbore failures we will

always look down deviated wells and evaluate
ZHOOERUHIDLOXUHVDVDIXQFWLRQRIDQJOHࣄIURPWKH
bottom of the well in a clockwise direction. We
also consider stress variations as a position
IXQFWLRQ RI SRVLWLRQ RI DQJOH ࣄ DURXQG WKH
wellbore going clockwise from the bottom.

Coordinate Transforms
In general, the principal stress tensor can be written in terms of the vertical stress Sv,
the maximum horizontal stress SHmax and the minimum horizontal stress Shmin as:
ܵ‫ݔܽ݉ܪ‬ ͲǤͲ ͲǤͲ
Ss = ൭ ͲǤͲ ݄ܵ݉݅݊ ͲǤͲ൱ (1)
ͲǤͲ ͲǤͲ ܵ‫ݒ‬
To rotate these stresses into a wellbore coordinate system (x b, yb, zb), we need two
coordinate transforms in succession. Firstly, we need to know how to transform the
stress field into a geographic coordinate system. We use the transform as:
‫ݔ‬௦ ܺ
൭ ‫ݕ‬௦ ൱= ܴ௦ ൭ܻ ൱ (2)
‫ݖ‬௦ ܼ
Mathematically, the matrix Rs required to transform the stress tensor into the
geographic coordinate system will be:
ܿ‫ߙݏ݋‬Ǥ ܿ‫ߚݏ݋‬ ‫ߙ݊݅ݏ‬Ǥ ܿ‫ߚݏ݋‬ െ‫ߚ݊݅ݏ‬
Rs = ൭ܿ‫ߙݏ݋‬Ǥ ‫ߚ݊݅ݏ‬Ǥ ܿ‫ ߛݏ݋‬െ ‫ߙ݊݅ݏ‬Ǥ ܿ‫ߛݏ݋‬ ‫ߙ݊݅ݏ‬Ǥ ‫ߚ݊݅ݏ‬Ǥ ‫ ߛ݊݅ݏ‬൅ ܿ‫ߙݏ݋‬Ǥ ܿ‫ߛݏ݋‬ ܿ‫ߚݏ݋‬Ǥ ‫ ߛ݊݅ݏ‬൱ (3)
ܿ‫ߙݏ݋‬Ǥ ‫ߚ݊݅ݏ‬Ǥ ܿ‫ ߛݏ݋‬൅ ‫ߙ݊݅ݏ‬Ǥ ‫ߛ݊݅ݏ‬ ‫ߙ݊݅ݏ‬Ǥ ‫ߚ݊݅ݏ‬Ǥ ܿ‫ ߛݏ݋‬െ ܿ‫ߙݏ݋‬Ǥ ‫ߛ݊݅ݏ‬ ܿ‫ߚݏ݋‬Ǥ ܿ‫ߛݏ݋‬
wherHĮ, ȕ and Ȗ are defined as Euler angles.
Next, to transform the stress field from the geographic coordinate system to the
wellbore system, we use the transform as:
‫ݔ‬௕ ܺ
൭‫ݕ‬௕ ൱= ܴ௕ ൭ܻ ൱ (4)
‫ݖ‬௕ ܼ
The matrix Rb required to transform the stress tensor in the geographic coordinate
system into the borehole coordinate system is:
െܿ‫ݏ݋‬ɁǤ ܿ‫ݏ݋‬ɔ െ‫݊݅ݏ‬ɁǤ ܿ‫ݏ݋‬ɔ ‫݊݅ݏ‬ɔ
Rb = ൭ ‫݊݅ݏ‬Ɂ െܿ‫ݏ݋‬Ɂ ͲǤͲ ൱ (5)
ܿ‫ݏ݋‬ɁǤ ‫݊݅ݏ‬ɔ െ‫݊݅ݏ‬ɁǤ ‫݊݅ݏ‬ɔ ܿ‫ݏ݋‬ɔ
where į is azimuth of the borehole horizontal projection and ij is inclination angle.
The stress tensors Sg and Sb in the geographic and wellbore coordinate systems
respectively can be defined as:
Sg = RST SS RS (6)
Sb = Rb (RST SS RS) RbT (7)
When considering effective stresses, the effective stress tensor becomes:
ıij = Sb,ij ± įijPp (8)
where Sb,ij is the i, j component of the stress tensor Sb and įij is the Kronecker.
772 K. D. Quang et al.

Mathematically, the general solution of effective stresses around an arbitrarily

inclined well will be described in terms of a cylindrical coordinate system as:
ܴଶ ܴଶ
ߪ௭௭ ൌ ߪଷଷ െ ʹᖳሺߪଵଵ െ ߪଶଶ ሻ ܿ‫ ߠʹݏ݋‬െ Ͷᖳߪଵଶ ଶ ‫ߠʹ݊݅ݏ‬
‫ݎ‬ଶ మ ‫ݎ‬
ଵ ோ ଵ ோర ோమ
ıࣄࣄ  ൫ߪଵଵ ൅ ߪଶଶ െ ʹܲ௣ ൯ ቀͳ ൅ మ ቁ െ ሺߪଵଵ െ ߪଶଶ ሻ ቀͳ ൅ ͵ ర ቁ …‘• ʹߠ െ οܲ మ  (9)
ଶ ௥ ଶ ௥ ௥
ଵ ோ ଵ మ ோమ ோర ோ మ
ıUU 
ଶ
൫ߪଵଵ ൅ ߪଶଶ െ ʹܲ௣ ൯ ቀͳ െ ௥మ ቁ ൅ ଶ ሺߪଵଵ െ ߪଶଶ ሻ ቀͳ െ Ͷ ௥మ ൅ ͵ ௥ర ቁ …‘• ʹߠ ൅ οܲ ௥మ 
ଵ ோమ ோర
߬Uϴ  ሺߪଵଵ െ ߪଶଶ ሻ ቀͳ ൅ ʹ െ ͵ ర ቁ •‹ ʹߠ
ଶ ௥మ ௥
whereοܲis the difference between the mud pressure Pm and the pore pressure Pp.
Therefore, the effective stresses at the wellbore wall (r=R) will become:
ߪ௭௭ ൌ ߪଷଷ െ ʹᖳሺߪଵଵ െ ߪଶଶ ሻ െ Ͷᖳߪଵଶ
ߪϴϴ ൌ ߪଵଵ ൅ ߪଶଶ െ ሺߪଵଵ െ ߪଶଶ ሻ …‘• ʹߠ െ οܲ 
߬ϴz ʹሺߪଶଷ ܿ‫ ߠݏ݋‬െ ߪଵଷ ‫ߠ݊݅ݏ‬ሻ (10)
ߪ୰௥ ൌ οܲ 
For a borehole arbitrarily inclined with respect to the principal stresses, ߬ఏ௭ is non-
zero In this case, the three principal stresses at the wellbore wall can be calculated
using:
ଵ ଶ
ߪ௧௠௔௫ ൌ ቀߪ௭௭ ൅ ߪ௾௾ ൅ ඥሺߪ௭௭ െ ߪ௾௾ ሻଶ ൅ Ͷ߬௾௭ ቁ (11)
ଶ
ଵ ଶ
ߪ௧௠௜௡ ൌ  ቀߪ௭௭ ൅ ߪ௾௾ െ ඥሺߪ௭௭ െ ߪ௾௾ ሻଶ ൅ Ͷ߬௾௭ ቁ (12)
ଶ
ߪ୰௥ ൌ οܲ (13)
where ߪ௧௠௔௫ and ߪ௧௠௜௡ are the maximum and minimum effective stresses in the plane
tangential to the wellbore wall.
Moreover, the angle Ȧ between ߪ௧௠௔௫ and the wellbore axis also is defined by:
ఛഇ೥
–ƒ ʹ߱ ൌ (14)
ఙ೥೥ ିఙ೷೷
Applying these formula and theory, a calculating program for the stress analysis of
wellbores (SAoWB) written in Matlab language has described and calculated all
components of the stress tensor at the wellbore wall as well as around boreholes.

3 Case studies

The observation of BOs and/or DITFs at the depths in the basement reservoir of
White Tiger field, Cuu Long basin, offshore Vietnam combined with Sv from density
logs, Shmin from hydraulic fracturing tests, and pore pressure Pp from DSTs, etc. can
indicate that the far field in-situ stress tensors at these interest depths in the basement
reservoir of White Tiger field are shown in Table 1.
Table 1. The full tensor of in-situ stress at the basement depths at White Tiger field.
Stress At 3900m: At 4100m: At 4300m:
Components BOs but no DITFs BOs but no DITFs No BOs but DITFs
Azimuth of SHmax,0N 1540N 1540N 1540N
Sv, MPa 87.57 92.55 97.56
Shmin, MPa 56.51 59.41 62.31
Pp, MPa 40.37 42.44 44.51
SHmax, MPa 87.00 91.50 99.00
Modelling Stress Distribution Around Boreholes 773

From the full tensors of in-situ stress at the interest depths, we input these data in the
program SAoWB to calculate in detail the stresses around boreholes to check and
constrain with the available information on the failure images from wellbores.
At the depth of 3900 m, the obtaining results from the program SAoWB are shown in
Fig. 3. We can see that the highest possibility of the BO occurrence is along the
direction of Shmin due to the concentration of the compressive stresses. If the
compressive stress over the rock strength, BOs will appear. Therefore, under the full
stress tensor at the depth of 3900 m, the occurrence of BOs with the predicted width
of 320 is fairly clear because the granite rock strength measured (110 MPa) is lower
than the rock strength required (120 MPa) to prevent the occurrence of BOs.
However, the occurrence of DITFs is almost impossible because Shmin is still positive
at the wellbore wall.

Fig. 3. Stress distribution at the basement depth of 3900 m at White Tiger field.
Similarly, the obtaining results at the depth of 4100 m from the program SAoWB are
shown in Fig. 4. The occurrence of BOs is more clear because the granite rock
strength measured (110 MPa) is lower than the rock strength required (130 MPa) to
prevent the occurrence of BOs but the predicted BO width can increase to 500. DITFs
also can not occur because Shmin does still not reach zero at the wellbore wall.

Fig. 4. Stress distribution at the basement depth of 4100m at White Tiger field.
774 K. D. Quang et al.

However, at the depth of 4300 m the obtaining results from the program SAoWB
are rather different (shown in Fig. 5). Although the highest possibility of the BO
occurrence is also along the direction of Shmin due to the concentration of the
compressive stresses, but the occurrence of BOs is impossible. This is because the
basement rock strength increased to 155 MPa due to the disapperance of the
weathered basement rock from the depth of 4300 m and the rock strength required
(148 MPa) to prevent the occurrence of BOs is still lower than the rock strength
measured (155 MPa). However, at this depth the occurrence of DITFs along the
direction of SHmax occurs because SHmax may reach zero at the wellbore wall.

Fig. 5. Stress distribution at the basement depth of 4300m at White Tiger field.

4 Conclusions

Modelling stress distribution around boreholes has presented fully, quickly and
precisely and been associated with the in-situ stresses, rock properties, etc. The
obtaining results for the field case studies are in good agreement with failure
observations from image logs of studied wells.
It helps to understand better the stress state, to establish more exactly the reservoir
geomechanics model as well as to solve well problems such as wellbore stability,
optimizing drilling wellbore trajectory, sanding, etc.

References
1. Fjaer, E., Holt, R.M., Horsrud, P., Raaen, A.M., and Risnes, R.: Petroleum related rock
mechanics. 2nd", Elsevier Ed., Developments in Petroleum Science 53, 491 p., (2008).
2. Khanh, D.Q.: PhD's thesis_Characterizing the full in-situ stress tensor and its applications
for petroleum activities, Dept. of Energy and Recourses Eng., CNU, Korea, (2013).
3. Peska, P. and Zoback, M.D.: Compressive and tensile failure of inclined wellbore and
determination of in-situ stress and rock strength" Jour. of Geophys. Res., 791-812, (1995).
4. Zoback, M.D.: Reservoir Geomechanics, Cambridge University Press, 449 p., (2010).
 A mathematical model for fault activation by water
injection

Thanh Son Nguyen1, Jonny Rutqvist2, Bastian Graupner3 and Yves Gugliemi2
1
Canadian Nuclear Safety Commission, Ottawa, Ontario, Canada
2
Lawrence Berkeley National Laboratory, Berkeley, California, USA
3
Swiss Federal Nuclear Safety Inspectorate, Brugg, Switzerland
Son.nguyen@canada.ca

Abstract. Faults in the host rock that might exist in the vicinity of deep geologi-
cal repositories for radioactive waste, constitute potential enhanced pathways for
radionuclide migration. Several processes might trigger pore pressure increases
in the faults leading to fault slip and induced seismicity, and increase the faults’
permeability. In this research, we developed a mathematical model to simulate
fault activation during an experiment of controlled water injection in a fault at
the Mont-Terri Underground Research Laboratory in Switzerland.

Keywords: fault activation, geological disposal, poromechanics.

1 Introduction

Subsurface fluid injection as practiced in some industrial operations induces excess

pore pressure that can be transmitted to faults, potentially leading to increased seismic-
ity [1]. The response of the rock formations and the faults to pore fluid pressure varia-
tions are governed by the classical laws of poromechanics. With pore fluid pressure
increases, the normal effective stress across the fault decreases, leading to a loss of
shear resistance. In this research, the authors focused on the potential for fault re-acti-
vation in the vicinity of geological disposal repositories for nuclear waste. Such repos-
itories rely on a system of multiple engineered and natural barriers to contain and isolate
the waste for hundreds of thousands to millions of years in the future. Faults that might
exist in the vicinity of a deep geological repository (DGR) for high level radioactive
waste, are potential enhanced pathways for radionuclide migration. Their hydro-me-
chanical characteristics need to be understood, and be taken into account in the safety
assessment and design of the DGR. In the very long time frame associated with the
DGR, several factors can trigger pore pressure increases across the faults, such as the
heat generated from the waste and glacial meltwater recharge after a future glaciation
period. The faults can then potentially slip and generate seismic events. Mathematical
models of fault re-activation have been proposed [2] based on the classical theory of
poromechanics and the classical law of Coulomb’s. Application of the above theoretical
framework for the prediction of fault behaviour is not a simple endeavour due to com-
plexities associated with the architecture of the fault and uncertainties in the in-situ

© Springer Nature Singapore Pte Ltd. 2020 775

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_123
776 T. S. Nguyen et al.

stress field. Furthermore, not all physico-chemical processes associated with the evo-
lution of the fault permeability and the conditions for fault re-activation are thoroughly
understood. In order to better understand the complex relationship between the above
factors, controlled induced fault slip in-situ experiments were performed [3]. In this
paper, we focus on the development of a mathematical model to simulate such experi-
ments performed in a natural fault in Opalinus Clay, at the Mont-Terri Underground
Research Laboratory (URL), Switzerland.

2 Mathematical modelling of injection experiment

2.1 Experimental setting

a) Geological section through Main and secondary faults b) Packed-off section

Fig. 1. Fault Injection experiments at the Mont Terri URL

The Mont Terri URL [4] is situated at a depth of approximately 300m below surface in
Opalinus Clay, an argillaceous rock formation being considered as a potential host for-
mation for the deep geological disposal of radioactive waste in Switzerland. The fault
injection experiments being performed at the URL are illustrated in Figure 1. Several
injection tests were performed in packed-off intervals of vertical boreholes that inter-
sect the Main Fault core and the secondary faults (Fig. 1a). The equipment used for the
tests consisted of surface equipment for conducting the tests and acquiring the data, and
a unique High-Pressure Pulse Probe borehole deformation tool (Fig. 1b). The packed-
off interval is pressurized (by water) and the probe measures deformation in both nor-
mal and shear displacement modes at a high sampling rate (~500 Hz). Nano-seismic
A mathematical model for fault activation … 777

monitoring arrays were also installed proximal to the experiment to determine source
locations of slip events. A total of five injection tests were conducted beneath, within
and above the Main Fault during the two measurement campaigns of June and Octo-
ber/November 2015. In this paper we focus on the injection test performed at the
packed-off section at 37.2 m depth of borehole BSF2 that crosses a secondary fault.
The secondary fault contains twelve striated planes most of them oriented N030°-to-
N060° dipping 50-to-70°SE. During the test, the injection pressure was increased step-
by-step in the chamber with an engine pump. Flowrate, injection chamber pressure and
borehole displacement variations were monitored.

2.2 Mathematical model

The governing equations of the model are derived from the theory of poromechanics:
௡ ଵି௡ డ௣ ఑ డ௣
൬ + ൰ െ ߘ ൫ߘ‫ ݌‬+ ߩ௙ ࢍ൯ െ ߙ =0 (1)
௄೑ ௄ೞ డ௧ ఓ డ௧

‫(׏‬ોᇱ + ߙ‫݌‬۷) = 0 (2)

In the above equations, p is the pore fluid pressure, ોᇱ is the effective stress tensor, N is
the permeability tensor, D is the Biot’s coefficient , Kf and Ks are the bulk moduli of
the pore fluid and the solid phase, respectively, P is the viscosity of the pore fluid, I is
the identity tensor, g is the acceleration of gravity and t is time.
The rock mass is assumed to be isotropic and elastic. The fault on the other hand is
assumed to be elasto-plastic. In the elastic domain, the fault is assumed to be trans-
versely isotropic, with two principal directions in the fault plane, and one principal di-
rection perpendicular to the plane. In the plastic domain, an isotropic Drucker-Prager
yield criterion is used to indicate the onset of fault slip and irreversible deformations.
It is assumed that when fault slip occurs, the permeability along the fault plane in-
creases, however the permeability in the normal direction to the fault remains un-
changed. The fault is conceptualized as a zone of finite thickness, with a series of par-
allel fracture planes oriented in the strike direction, separated at a distance s. The equiv-
alent permeability of the fault in the strike direction is then given by the cubic law:
య
௕೓
‫= צܭ‬ (3)
ଵଶ௦

In equation (3), bh is the hydraulic aperture of each individual fracture. It is assumed

that the permeability of the fault in the perpendicular direction ߢୄ to the fault remains
constant and equal to the one of the rock mass. Due to effective stress change, it is
assumed that the hydraulic aperture of each fracture would change according to:
ܾ௛ = ܾ௛௜ + οܾ௛௘ + ‫ܣ‬οܾ௛௣ (4)
778 T. S. Nguyen et al.

In equation (4), bhi is the initial hydraulic aperture and A is a damage enhancement
factor. The change in aperture is triggered by the change of normal effective stress
across the fault, induced by the change in pore fluid pressure. In this model, when the
stress state in the fault has not reached the Drucker-Prager criterion, the deformation is
elastic and induces an elastic aperture change 'bhe. When the Drucker-Prager criterion
is reached, permanent deformation occurs, inducing a plastic aperture change 'bhp.
Therefore, mechanical deformation results in changes in the permeability according to
equations (3) an (4). The authors also introduce a damage enhancement factor, by as-
suming damage following fault slip can enhance the interconnectivity between the in-
dividual fractures, resulting in a more drastic increase of the fault permeability.

Fig. 2. Finite element model for the secondary fault injection experiment

The governing equations of the model, with applicable boundary and initial conditions,
were solved by the finite element method. The geometry of the finite element model
and the boundary conditions are shown in Figure 2. We considered a block
20mx10mx20m, where the yz plane is plane of symmetry with the injection point at its
centre. The secondary fault is included as a finite thickness feature, striking N045o and
dipping at 65oSE.

2.3 Modelling results

In the injection chamber, the pressure is increased in a stepwise manner as shown in
Figure 3. As a result of the pressure increase, the fault opens resulting in displacement
in the vertical direction, and in the east and west directions. The simulated results com-
pare reasonably well in trends and absolute values with the relative displacements rec-
orded by the anchors. The fault movement increases in a stepwise manner, reflecting
the increase in injected water pressure. The horizontal relative displacements are larger
than the vertical one. The model correctly predicts that finding, albeit with an over
prediction of the vertical displacement as compared to the experimental results. The
model also correctly predicts a sharp increase in both vertical and horizontal displace-
A mathematical model for fault activation … 779

ment at the highest injection pressure (approximately 6.5 MPa), resulting in fault fail-
ure. In the last step of the injection sequence, the pressure is decreased to approximately
3.2 MPa. The model correctly shows that permanent deformation remains, albeit with
an overestimation as compared to the measured values. At the highest injection pres-
sure, fault failure leads to a drastic increase in the permeability, resulting in a sudden
spike in injection flow rate. Figure 4 compares the modelling results with the experi-
mental data for the injection rate. In that figure the spikes observed from the experiment
that occur before 400 s are due to the elastic deformation of the test chamber and do
not reflect actual water ingress into the fault. Consistent with the experimental results,
the model shows that flow into the fault is insignificant until approximately 400 s when
the injected pressure increases to over 6 MPa. The model correctly shows this behav-
iour, albeit with an underprediction of the flow rate.

Fig. 3. Modelling results for displacements compared to experimental data

780 T. S. Nguyen et al.

Fig. 4. Modelled injection flow rate compared to experimental data

3 Conclusion

In this research we developed a mathematical model for fault re-activation due to pore
pressure increase, based on the classical theory of poromechanics and Coulomb’s fric-
tion. A relationship for fault permeability evolution is proposed, taking into account
fault movement before and after fault slip. The model correctly simulates the main pro-
cesses that prevailed in a water injection experiment in a secondary fault in Opalinus
Clay. Further investigations on the effects of fault heterogeneity, the energy release
from fault slip and the simulation of the re-activation of the Main Fault are in progress.

References
1. McGarr A et al: Coping with earthquakes induced by fluid injection. Science 347(6224):
830-831 (2015).
2. Rutqvist J et al: Fault activation and induced seismicity in geological carbon storage - Les-
son learned from recent modeling studies. J. Rock Mech. and Geot. Eng. 8:789-804 (2016).
3. Guglielmi Y et al: Underground Research Laboratories for Conducting Fault Activation Ex-
periments In Shales. American Rock Mechanics Association, 49th U.S. Rock Mechan-
ics/Geomechanics Symposium (2015).
4. Thury M and Bossart P: The Mont Terri rock laboratory, a new international researchproject
in a Mesosoic shale formation, in Switzerland. Eng.Geol. 52(3-4):347-359 (1999).
 A model for soil-structure interaction - Application to
small modular reactors

L. Cui1, M. Fall2, N. Orbovic3, and T.S. Nguyen3*

1
Lakehead University, Thunder Bay, Canada
2
University of Ottawa, Ottawa, Canada
3
Canadian Nuclear Safety Commission, Ottawa, Canada

* email: son.nguyen@canada.ca

Abstract. Canada and other countries are contemplating the construction of

small modular reactors (SMR) for the generation of electricity. SMRs would be
built at a smaller scale than traditional nuclear reactors with anticipated lower
up-front capital costs and enhanced safety features. The SMRs would need to be
seismically qualified, and shown to withstand the design basis earthquake with-
out loss of containment. In order to assess the seismic qualification of SMRs, a
dynamic coupled elastoplastic-hydraulic model for soil behaviors under seismic
loadings is developed. The governing equations of the model are based on con-
servation of momentum for the porous skeleton, and conservation of water
PDVV3RUHZDWHUIORZLVDVVXPHGWRIROORZ'DUF\¶VODZZKLOHWKHVROLGVNHOe-
ton is assumed to be elasto-plastic, with the adoption of the modified Cam-Clay
model to simulate its stress-strain behavior. The model is tested with the simu-
lation of dynamic triaxial tests and a shaking table experiment. The results show
that the model can capture (1) the development of permanent deformation in
soil, (2) the shear-induced volume change (including both contraction and ex-
pansion), (3) the generation and dissipation of excess pore water under the dy-
namic loading, and (4) the strain hardening and softening behavior of soil under
complex stress paths. Finally, the model is used for the scoping analysis of the
seismic response of a hypothetical SMRs embedded in sandy soils.

Keywords: Soil-structure Interaction, Elastoplasticity, Excess Pore Water Pres-

sure, Small Modular Reactors

1 Introduction

Nuclear power generation was established since the 1950s [1] and is an important
contributor to the generation of electricity in many countries. Due to the high capital
cost of large nuclear reactors and the need to service small electricity grids, there are
plans to develop small modular reactors (SMRs) with power smaller than 300 MWe
[2]. SMRs could effectively reduce the amount of work on-site, making them simpler
and faster to construct, especially in remote areas [3]. The operation and maintenance
costs of SMRs would also be relatively low. Due to the above considerations, in re-

© Springer Nature Singapore Pte Ltd. 2020 781

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_124
782 L. Cui et al.

cent years SMRs have received substantial attention in Canada and other countries.
The eventual construction and operation of SMRs in Canada require licensing from
the Canadian Nuclear Safety Commission (CNSC) &DQDGD¶V QXFOHDU regulator. In
support of a licence application to the CNSC, the applicant must demonstrate that the
proposed activities ensure the protection of the public and the environment. In partic-
ular, the SMRs would need to be seismically qualified, and shown to withstand the
design basis earthquake without loss of containment. In order to assess the seismic
qualification of SMRs, a dynamic coupled elasto-plastic-hydraulic model for soil
behaviors under seismic loadings was developed. The model was validated against a
cyclic triaxial test and a shaking table test. The model was then used to perform a
scoping analysis of a hypothetical SMR embedded in sandy soils.

2 Formulation of Dynamic Coupled Elastoplastic-hydraulic

Model

Based on the volume-averaging method, the mass balance equation of combined solid
and liquid phases in a soil can be derived to describe the hydraulic process. In addi-
tion, the mechanical process can be captured by the momentum balance equation.
wU U wU wH
I w  1  I  w s  ’ ˜ IU w v rw   U w v (1)
wt U s wt wt
’˜ ı  ª¬1  I  Us  IUw º¼ g ª¬1  I  Us  IUw º¼ u
&
& (2)
where ‫ ׋‬is porosity, ȡ is the density, ı is the total stress tensor, v is the water flow
velocity, İv is the volumetric strain of soil skeleton, g is the acceleration of gravity,
and u is the displacement field. B\ LQFRUSRUDWLQJ WKH JUDYLW\ HIIHFW 'DUF\¶V ODZ LV
used to calculate the pore water flow in the saturated soil.
To describe the development of permanent deformation in soil, the Modified Cam-
Clay (MCC) Model was incorporated into the coupled soil-pore water model.
F q 2 pc2  M 2 1  pcc 0 pc  0 (3)
where q is the deviatoric stress, p' is the mean effective stress, pc' is the initial
preconsolidation pressure, M is the slope of critical state line. I1 and J2 respectively
denote the first stress invariant and second invariant of the deviatoric stress.
To capture the strain-rate-dependence in soil, the Perzyna type of viscoplastic formu-
lation is adopted in the current study.
wF
İ&vp K I ( F ) (4)
wı
where Ș is evolutive fluidity parameter for soil, ½¾is the Macaulay bracket, F is the
yield function (equation (3)) and ‫(׋‬F) is a dimensionless scaling function defined as
‫(׋‬F)=F/F0 with F0 as the normalizing constant with the same unit as F.

2
A model for soil-structure interaction … 783

3 Validation of the Coupled Elastoplastic-hydraulic Model

To validate the developed elastoplastic-hydraulic model, the modeling results are

compared with the measured data from a cyclic triaxial test and a shaking table test.

3.1 Case study 1 - Cyclic triaxial test

To study the dynamic behavior of sandy soil, a series of stress-controlled undrained

cyclic triaxial tests were conducted [4]. The development of excess pore water pres-
sure (PWP) is shown in Fig. 1a. From this figure, it can be observed that the cyclic
stress can significantly affect the excess PWP. Specifically, excess PWP can develop
during the cyclic loading and dissipate during the stress release process. However, the
excess PWP did not completely dissipate at the end of the test. This is due to the de-
velopment of permanent deformation in the soil sample under the cyclic loadings. The
development of irrecoverable deformation in the soil is clearly demonstrated in Fig
1b. Through the comparison between predicted results and measured data, it can be
found that the developed model is able to adequately capture the development of de-
formation and associated excess PWP.

a) Data-Measured Data-Model b) Data-Measured Data-Model

120 5
Excess PWP (kPa)

Axial strain (%)

90 0
60
-5
30

0 -10
0 1 2 3 4 5 6 0 2 4 6
Time (s) Time (s)

Fig. 1. Comparison of numerical prediction and experimental data of a) excess pore water pres-
sure and b) axial strain.

3.2 Case study 2 - Shaking table test

To further demonstrate the predictive capability of the coupled elastoplastic-hydraulic

model, a shaking table test with a loose sandy soil [5] is considered in this study. The
input motion at the bottom of the shaking table is presented in Fig. 2a. The variation
of pore water pressure at the monitoring points are shown in Fig. 2b. From this figure,
it can be observed that (1) the initial hydrostatic pressure increases with the depth of
the soil deposit (the monitoring point with the elevation of 0.04m has the largest depth
with respect to the top surface of soil deposit); (2) the generation of excess PWP in-
creases with the duration of the input base motion, then start to dissipate after the
input motion ceases; (3) the excess PWP does not fully dissipate at the end of the test.
As shown in Fig. 2b, the predicted results are in good agreement with the experi-
784 L. Cui et al.

mental data at different monitoring points. Therefore, based on the results obtained
from this study, it can be confirmed that the developed model can capture the main
processes (development of plastic strain, and excess pore water pressure) that prevail
in loose sandy soils under dynamic loadings.

Data-0.04m Data-0.08m Data-0.12m

Model-0.04m Model-0.08m Model-0.12m
0,3
60

Excess PWP (kPa)

0,2
Acceleration (g)

50
0,1 40
0 30
0 5 10 15 20 20
-0,1
-0,2 10
0
-0,3
Time (s) 0 5 10 15 20
Time (s)

Fig. 2. a) horizontal input motion and b) comparison of predicted results and experimental data
of shaking table test.

4 Scoping Analysis of a hypothetical SMR embedded in sand

In this study, the elastoplastic-hydraulic model was implemented into a commercial

finite element analysis program (COMSOL Multiphysics) to perform the seismic
analysis of a hypothetical SMR embedded in sand. The horizontal acceleration from
the 1988 Saguenay earthquake ( Fig. 3) was adopted to simulate the base motion. The
1988 Saguenay earthquake is one of the strongest earthquakes in the 20th century that
occurred in Eastern North America. It has a moment magnitude of 5.9. Fig. 4 shows
the geometry of the numerical model. To simplify the seismic analysis, it is assumed
that the sandy soil and bedrock are isotropic and homogenous materials. Moreover,
compared with the surrounding soil, the SMR and bedrock are relatively rigid and
thus are assumed to behave elastically under seismic loadings. In addition, only half
of the soil and SMR structure system was simulated. For the applied boundary condi-
tions, a low-reflecting boundary was used at the far end (vertical sidewalls) of the soil
deposit, and the bottom bedrock was set as a zero-normal displacement boundary
condition. The ground surface was set as a free boundary. The horizontal acceleration
was applied to the bedrock domain. For the hydraulic process, a zero pressure was
applied at the top surface of soil, assuming that the phreatic surface is at ground
surface, and no flow boundary conditions were applied to sidewalls and bottom sur-
face.

4
A model for soil-structure interaction … 785

Fig. 3. Input horizontal acceleration on the bedrock

Fig. 4. Geometry model adopted by the finite element analysis

Three monitoring points were selected in the soil domain. Points 1 and 2 are at the
same horizontal distance (5m) from the SMR but at different depths (point 1 with a
depth of 7.5m and point 2 with a depth of 10m). Point 3 is located at a depth of 7.5m
(which is the same as point 1) and a horizontal distance of 10 m from the SMR struc-
ture. To assess the liquefaction potential of the soil, the pore water pressure ratio was
calculated and presented in Fig.5. The pore water pressure ratio is defined as the ratio
between the excess pore water pressure and the effective confining pressure. The soil
may liquefy when the pore water pressure ratio is in the range of 0.7 to 1.0 [6]. It can
be observed that the pore pressure ratio is higher for the monitoring points which are
closer to the SMR structure and ground surface. Consequently, soil liquefaction may
take place at shallow depths in the vicinity of the hypothetical SMR, as a result of a
combination of factors (such as thick loose sand layer, phreatic surface at ground
surface, and strong earthquake). Liquefied sand has higher mobility and negligible
shear strength and could impact the structural integrity of the SMR and amplify the
soil motion to different components of the SMR. Therefore, the influence of soil con-
ditions on the structural integrity and containment capability of a proposed SMR must
be taken into account in the seismic analysis and design of the facility.
786 L. Cui et al.

1 Point 1 Point 2 Point 3

0,8

PWP ratio 0,6

0,4

0,2

0
0 5 10 15 20 25 30 35 40 45 50
Time (s)

Fig. 5. Residual pore water pressure ratio in the sandy soil foundation

5 Conclusions

A dynamic coupled elastoplastic-hydraulic model was developed to assess soil be-

havior under seismic loadings. The model was validated against the experimental
data of cyclic triaxial test and shaking test on soils. The good agreement between
predicted results and measured data clearly demonstrates the capability of the de-
veloped model to capture the coupled mechanical-hydraulic processes that occur in
soils under seismic loading. The model would thus be a useful tool for the seismic
analysis of structures, such as SMRs, taking into account the influence of soil
foundation behavior and its interaction with groundwater.

References
1. Carless TS, Talabi SM, Fischbeck PS.: Risk and regulatory considerations for small modu-
lar reactor emergency planning zones based on passive decontamination potential. Energy
167,740-756 (2019).
2. Markou G, Genco F.: Seismic assessment of small modular reactors: NuScale case study
for the 8.8 Mw earthquake in Chile. Nucl Eng Des. 342,176-204 (2019).
3. Hidayatullah H, Susyadi S, Subki MH.: Design and technology development for small
modular reactors±Safety expectations, prospects and impediments of their deployment.
Prog Nucl Energ. 79,127-135 (2015).
4. Ural N, Gunduz Z. Behavior of nonplastic silty soils under cyclic loading. The Scientific
World Journal. (2014).
5. Taboada VM, Dobry R. Experimental results and numerical predictions of model. In: Pro-
ceedings of International Conference on the Verification of Numerical Procedures for the
analysis of soil liquefaction problems. California, United States; (1993).
6. Saeid AL, Mahmoud Y, Arya AL.: A study on the liquefaction risk in seismic design of
foundations, Geomech. and Eng. 11(6), 805-820 (2016).

6
Full-scale Pullout Testing of Ground Anchors to Evaluate
the Applicability of French Design Practice TA95 for
Vietnam

Nguyen Ngoc Thanh1 and Nguyen Phuong Duy2

1 Hanoi Architecture University, Hanoi, Vietnam
2
University of Transports and Communications, Hanoi, Vietnam
nnthanhxd@gmail.com

Abstract. Ground anchors have been playing an important role in reinforcing

dykes, underground, stabilizing structures of earth retaining walls, subway sta-
tions, and anchoring abutments of bridge or in sea port, etc. In Vietnam, the de-
sign of ground anchors is generally based on the international design guidelines,
in particular the Eurocode. As there is a diversity of current design practices in
different countries, it is necessary to carefully verify and validate an international
design of ground anchor before adopting it to Vietnam. This paper aims at pre-
senting a series of full-scale pullout tests of ground anchors recently performed
in several projects in Hanoi and Ha Long city, Vietnam. The testing results in
terms of the ultimate load-holding capacity, the pullout performance and the skin
friction are discussed and compared with the French design practice TA95 in
order to evaluate as well as to improve its applicability in Vietnam.

Keywords: Ground anchor, trial testing, post grouting.

1 Introduction

Over the last two decades, in Vietnam, the application of the ground anchors used as
an alternative to the bracing retaining system in the deep excavation has become more
and more popular. The first temporary ground anchors application in Hanoi was rec-
orded in 1997 at Vietcombank Tower project in Hoan Kiem district. The ground an-
chors were combined with diaphragm wall for the construction of these underground
basements. Since then, we have observed many successful applications of ground an-
chors to deep excavations in Hanoi such as: National Assembly Building project, EVN
tower project, ViettinBank tower project, Goldmark city project, HH1 Nam Cuong pro-
ject, Eurowindow river park project and FPT tower project, etc.
Through several consulting and research projects, it has been demonstrated that the
ground anchor system is an efficient solution which could help to not only reduce the
cost of king-post and bracing system but also improve the construction schedule on site,
especially for large excavation construction. However, the ground anchor design in Vi-
etnam has been generally relied on the international design standards and codes of prac-
tice such as AASHTO [1], Eurocode [2, 3] and TA95 [4]…. Note that the developments

© Springer Nature Singapore Pte Ltd. 2020 787

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_125
788 N. N. Thanh and N. P. Duy

of the design methods recommended in these international guidelines have been mainly
based on observations from field of anchor tests and long-term behaviors experimental
data reflecting the regional practices and local experiences. Therefore, there is potential
diversity of existing design practices used in different countries, which could make it
difficult and even impossible to apply them to the soil condition in Vietnam. Unfortu-
nately, very few reports on the full-scale case study of ground anchors performed in
Vietnam have been published. In this context, it is highly important to verify and vali-
date the applicability of the design approaches developed in other countries, before ap-
plying them to the domestic design, by comparing with local full-scale measurement
data from site experiences and trial tests of ground anchors. This is the main objective
of the research work presented in this paper.

2 Ground anchor design and empirical approaches

For ground anchor design in Vietnam, three most used codes of calculation are the
British Standard Code of practice for Ground Anchorages BS 8081:1989 (BS);
AASHTO LRFD 2012 for bridge design specification (AASHTO); and the Ground An-
chors – Recommendations for the design, calculation, execution and control by Comité
Français de la Mécanique des Sols et des Travaux de Fondations (TA95).
The ground anchor holding capacity can be generally determined by the following
formulation:

Tu = S.D.Lb. qs (1)

where Tu is the ground anchor ultimate bearing capacity; Lb is the effective bond length
or the grouting anchors length; D is the effective diameter of the bond length and qs is
the ultimate unit skin friction at ground/grout interface.
It is important to note that the following assumptions have been used in Equation
(1): (a) The applied load is transferred from the fixed anchor to the ground as a uni-
formly distributed stress acting over the entire perimeter of the bond length; (b) failure
takes place by sliding at the grout/soils interface or by shearing adjacent to soils/grout
interface in the weaker region; (c) there are no discontinuities or inherent weakness
planes along where failure can be included; and (d) there are no local de-bonding at the
grout/soils interface. This means that the ultimate load-holding capacity of anchorage
is depended not only on the bond length, Lb, of the anchorage but also on the effective
anchor bond diameter, D, and the ultimate skin friction, qs, at grout/soils interface.
These parameters are the complex functions of soils condition and the installation tech-
niques (drilling and grouting).
For final design, the codes of calculation recommend to perform a pullout tests on
sacrificial anchors in each soil unit in order to establish anchor length and capacities
that are consistent with the chosen method of anchor installation. Proof tests conducted
on every production anchor are then required a safety factor on the design load to verify
capacity. However, for preliminary design, the load holding capacities of the anchors
could be estimated either based on the preliminary pullout load tests, previous experi-
ence or using published soils/grout bond guidelines.
Full-scale Pullout Testing of Ground Anchors … 789

For cohesion-less soils, the estimation in the AASHTO guidelines is based on the
ground conditions, density, borehole diameter, bond length and grouting pressure. The
post grouting method for creating the fixed anchor bond length is suggested in order to
increase the bearing capacity by 20-50%. For specific soils conditions, the experimental
ultimate unit bond stress, qs, is given in regarding to the SPT numbers.
As for the BS code, the design method for load holding capacity of fixed anchor
installation in cohesionless soils is solely based on the experiences from field tests. The
approach has considered neither the effects of high pressure on the soils/grout skin fric-
tion improvement nor the increase of effective diameter.
On the other hand, in the French Design Practice TA95, the design method of anchor
bearing capacity takes into account the bond length, the unit skin friction on the
grout/soils interface as well as the effective diameter of the bond length. This diameter
is determined by experiments satisfying a certain conditions of the injection methods
including the selective repetitive injection under high pressure (IRS) and the injection
global and unique under low pressure (IGU). The effective fixed anchor bond length
can be estimated as follows:

D Į'd (2)

where D is the effective bond length diameter; Dd is the drill hole diameter; and ĮLVDQ
experimental reference value depending on the grout injection technique, grouting pres-
sure and soils condition.

3 Full-scale trial tests of ground anchors

As previously stated, before any mass production and to verify the preliminary design
of bearing capacity, it is recommended to perform the full-scale trial tests. We present
here the details of the trial tests, which have been recently carried out at the FPT Com-
plex Tower project, Dich Vong ward, Cau Giay district, Hanoi. The ground anchors
have the bond length of 12 m embedded in the medium hard clay layer (NSPT/30 = 6-10
for first layer and NSPT/30 = 8-12 for remaining layers). The pullout tests were performed
after both grouting stages for the first layer of trial ground anchor.

Fig. 1. Load-displacement curves (1st layer of ground anchor).

In order to confirm the performance of the grouting method and the effective bond
length, in considering the “Proof load” value of Pp = 1.25 × Design load = 478.5 kN,
790 N. N. Thanh and N. P. Duy

trial anchors were first incrementally loaded to that value and then additional cycles of
unload-reloading were carried on. Fig. 1 shows the relationship between measured dis-
placement and tension in the anchor comparing to the upper limit and lower limit elon-
gation line according to BS 8081 standards. The obtained results are satisfied and well
agreed with the requirement of BS8081 standards, which confirm the effective bond
length and the performance of the grouting injection method.
For the ultimate load-holding capacity and the pullout performance of the trial an-
chorages, the applied loads were progressively increased and creep behaviors were also
observed. After reaching the loading step and the stabilized pressure gage, the pressure
was then maintained constant during the observation time (30’). The jacking pressure
was adjusted whenever it decreased more than 1% in according to the TA95. The failure
occurred when the maximum cumulative creep displacements are more than 1 mm.
Fig. 2 shows the load-displacement curves of the pullout performance of both trial
anchorage layers. The holding capacity of the first layer of trial anchorage is consider-
ably increased after post-grouting stage (1.7 times), proving the efficiency of the grout-
ing technique. This value is slightly lower in comparison to the working coefficient Į
proposed by TA-95 for clayey soils. After post-grouting stage, the first layer of trial
anchorage recorded an ultimate holding capacity of 65 tons and the second layer ground
anchor of trial anchorage recorded a load capacity of 76.6 tons. Table 1 summaries the
calculated results of the ultimate skin friction based on the recorded load holding of the
trial anchorages. Note that these tests assumed the effect of post-grout injection in-
creased the apparent bond anchorage diameter solely.

Fig. 2. Load-displacement curves of pullout performance of both trial anchorage layers.

Table 1. Skin friction calculated based on ultimate load holding of trial anchors.
Dd Lb Ultimate load
D (m) NSPT/30 Skin friction
No GA layer (m) (m) (kN)
qs (kPa)
Tu = Į.Dd.Lb.qs
1 First layer 0.134 0.228 12 6 - 10 650 75.66
Second
2 0.134 0.228 12 8 - 12 765.6 89.11
layer
Full-scale Pullout Testing of Ground Anchors … 791

4 Skin friction and standard penetration resistance

Table 2 summaries the field experimental data assessment of the pullout full-scale trial
tests carried out within six recent projects in Vietnam.

Table 2. Ground anchors trial tests in recent projects in Vietnam.

Average
TT Projects Name Ground description Lb/ Incli.
SPT(N30)
1 FPT Tower, HN Clayey/low plasticity 8 12m/ 35º
2 FPT Tower, HN Clayey/low plasticity 10 12m/ 35º
3 Goldmark city, HN Clayey/low plasticity 12 12m/ 35º
4 Goldmark city, HN Clayey/low plasticity 20 12m/ 35º
5 Lideco tower, QN Clayey/low plasticity 30 12m/ 35º
6 Lideco tower, QN Clayey/low plasticity 40 12m/ 35º
7 ViettinBank tower, HN Sandy/very dense 20 12m/ 35º
8 ViettinBank tower, HN Sandy/very dense 30 12m/ 35º
9 Eurowindow Res., HN Sandy/medium dense 17 12m/ 35º
10 Eurowindow Res., HN Sandy/very dense 26 14m/ 35º
11 HH1 Nam Cuong, HN Sandy/medium dense 10 12m/ 35º
12 HH1 Nam Cuong, HN Sandy/ medium dense 13 12m/ 35º

Fig. 3 and Fig. 4 illustrate the estimations of ultimate skin friction, qs, obtained from
trial tests compared with the values proposed by TA95 for the grouting technique in
clayed and sandy soils, respectively. It is observed from these figures that with high
pressure post-grouting injection technique, the experimental curve of ultimate friction
grout/ground surface with SPT (N30) is similar in shape to TA95's proposal for sandy
soil but lower in case of installation in clayey soil.

Fig. 3. Relationship between the skin friction qs and SPT (N30) in clay.
792 N. N. Thanh and N. P. Duy

Fig. 4. Relationship between skin friction qs and SPT (N30) in sand.

5 Conclusions

The paper presents in detail the full-scale ground anchors trial tests carried-out within
the FPT project in Vietnam and discusses calculated results in terms of the ultimate
bearing capacity and the performance of the installation method. The test results could
also provide important and useful data for evaluating WKHZRUNLQJFRHIILFLHQWĮDQGWKH
ultimate skin friction of soils/grout interface based on the in-situ SPT values.
The analyses of the pullout full-scale trial testing results of ground anchors, which
were performed within six recent projects in Vietnam, indicate that for sandy soils, the
ultimate skin friction values obtained from the domestic testing sites agree very well
with the empirical prediction recommended by the French design practice TA95. For
the case of clayey soil of Hanoi, the use of the empirical method of the French Design
Practice TA95 could result in significant overestimation of the ultimate skin friction.
Therefore, it is concluded that the empirical curve recommended in TA-95 for evalua-
tion of the ultimate skin friction in clay/grout should be used with care for the domestic
soil conditions.

References
1. AASHTO LRFD-Bridge Design Specifications (2012).
2. BS 8081:1989. British Standard Code of Practice for Ground anchorages (1989).
3. BS EN 1537:2000 Execution of special geotechnical work-Ground anchors, BSI (2000).
4. TA 95. Tirants d’ancrage- Recommandations concernant la conception, le calcul, l’exécu-
tion et le contrôle. Comité Français de la Méchanique des Sols et des Travaux de Fonda-
tions, Eyrolles –France (1995).
 Durability Evaluation of a Geothermal Grout

Monnot Pascal 1, Poinclou Christophe2, Blanc Philippe1, Rojo Amandine3,

Molez Laurent4, Lanos Christophe3(*)
1
BRGM, Direction des Geo-ressources, Orléans, France
2 ANTEAGROUP, Olivet, France
3
Univ Rennes, LGCGM EA 3913, Rennes, France,
4 Univ Rennes, INSA, LGCGM EA 3913, Rennes, France,

(*) christophe.lanos@univ-rennes1.fr

Abstract. The durability of a geothermal grout used to seal vertical geothermal

probe is evaluated. The study integrates an experimental approach and geochem-
ical models simulations. The accelerated ageing of the geothermal grout is eval-
uated in laboratory subjecting the grout to realistic geochemical and thermal ag-
gressions to which the grout can be subjected. The parameters related to the ther-
mal conductivity and the permeability are particularly monitored. An experi-
mental setting is developed to reproduce the operating conditions of a geothermic
probe under realistic containment conditions. The experimental results are the
input of the geochemical degradation models leading to the simulation of long
time scale degradation around the probe. Conclusions relating to the service life
rise from this study.

Keywords: geothermal grout, thermal conductivity, permeability, geochemical

model, freeze thaw cycles.

1 Introduction

The durability of geothermal grouts is a major issue, because the construction of vertical
geothermal probes (VGP) is confronted with the difficulty of "continuously and perma-
nently" sealing the equipment (probe loops) in the boreholes drilled in the surrounding
ground. In such a context, the main objectives of good cementation are to ensure the
protection of the environment by avoiding surface infiltration or the connection of sev-
eral aquifers, to ensure the balance of mechanical stresses in the well, but also to opti-
mize the transfer of energy (hot and/or cold) between the subsoil and the heat ex-
changer. The particularity of the thermomechanical behavior of the grout in a geother-
mal probe is mainly related to the thermal stresses due to the operation of the heat pump
(in hot or cold cycle), as well as to the peaks of operation at maximum power which
can lead, in the case of an unsuitable design of the heat exchanger, to the freezing of
the grout and the ground.
To guarantee, in the long term (at least for the operating time of the heat pump, if
possible more), the proper sealing of the structure and the thermal performance of the
geothermal installation, the grout must be able to preserve its heat transfer capacity

© Springer Nature Singapore Pte Ltd. 2020 793

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_126
794 M. Pascal et al.

(thermal conductivity), as well as its initial mechanical properties and physico-chemical

stability. The main objective of the work carried out within the framework of the
DURACIM project is to characterize, at lab scale, the accelerated ageing of the geo-
thermal grout, by reproducing the operating conditions of a geothermal probe and the
geochemical stresses to which the grout can be subjected. The observed physical phe-
nomena and the appearance of possible pathologies were coupled with a geochemical
model to extrapolate, over the long term, the durability of the grout.

2 Experimental approach and model of degradation process

The methodology used to assess the durability of geothermal grouts is based on the
development of two grout characterization devices: the interpretation of thermome-
chanical data in relation to accelerated aging of samples in the laboratory and the de-
velopment of two geochemical models to extrapolate accelerated aging over a 25-year
period. On the experimental level, the project is structured around two major axes:
- The development of a coupled device, allowing to reproduce, on a reduced scale, the
operating context of an VGP under conditions close to those of a borehole at a depth of
~200 m. This device makes it possible to recreate the grout containment conditions in
a steel oedometric cell, under thermal, hydraulic and constant mechanical stress.
- The evaluation of the degradation conditions of geothermal grout samples by dissoci-
ating the different alteration mechanisms (decoupled device). This device allows the
grout samples to be immersed in a thermo-regulated bath, filled with more or less ag-
gressive solutions, under atmospheric stress. 3 types of solutions are used: tape water,
water saturated with gypsum, water with sulfuric acid to ensure a sulfur content close
to the higher content quoted in French groundwater.
The study was carried out on a grout produced by an industrial partner (Heidelberg
Cement). The commercial product chosen is Thermocem, which is representative of the
products available on the market. Grout durability is assessed by analyzing the evolu-
tion of two performance indicators:
- Thermal conductivity (λ), a parameter that influences the ability to transfer energy
(hot - cold) from the ground to the heat transfer fluid of the geothermal probe, and thus
the performance of the geothermal installation;
- Water permeability (K), a parameter that influences the grout's ability to prevent
possible contamination from the surface into groundwater, but also to avoid connecting
possible aquifers, and thus protect the environment.
The analysis and monitoring of the evolution of these two parameters is carried out
on grout samples that have undergone thermal stress, freeze-thaw cycles and chemical
aggression. As references, a number of studies have studied the influence of grout com-
position on its thermal conductivity: relative contents of pure cement, bentonite, water
and quartz [1][2][3][4], even graphite [5][6]. These studies show that a grout mainly
composed of sand, cement and water, with a very high sand/water mass ratio (between
2.0 and 2.4), has a thermal conductivity of about 2.3 W.K-1.m-1. The thermal conduc-
Durability Evaluation of a Geothermal Grout 795

tivity of this material is not very sensitive to water content (desiccation), unlike ben-
tonites or cements [7]. [3] Measured the hydraulic conductivity of a sample of this grout
contained in a polyethylene (PE) tube and found it to be between 10 -9 and 10-7 m.s-1.
In parallel with the development of the two experimental devices, two geochemical
models are being developed [8]:
- The first, called the "grout model", is a geochemical model of grout alteration with-
out chemical element transport. This model takes into account the influence of temper-
ature as well as the different chemical solutions in contact with the grout. The model
provides the mineralogical composition and the composition of the solution obtained
after a theoretically infinite time. This model makes it possible to locate the changes in
the parameters analysed as a function of time.
- The second geochemical model, called the "alteration model", allows the produc-
tion of calculations with different hydrogeological context scenarios. Based on the re-
sults obtained on the test pilot (coupled device), the first geochemical model (grout
model) was completed by integrating a transport component, in order to simulate the
percolation of the solution and the consequences on the chemical stability of the slurry.
Taking into account the diffusion/advection of the percolation solution makes it possi-
ble to introduce a kinetic component into the simulation. The calculations carried out
make it possible to evaluate the consistency between the physical parameters measured
and the evolution of the chemical and mineralogical parameters.

3 Results

At fresh state, the density and the viscosity of the slurry are controlled. The fresh den-
sity is 1460 kg.m-3 and the Marsh flow time is 60 s. The water to binder ratio is 0.8.
The slurry is casted to shape blocks 210 x 350 x 250 mm3. The setting time is about
20 h. The blocks are sealed and placed in room at 100% relative humidity. Some meas-
urements made on the reference hydrated product highlight the homogeneity of the
samples core drilled in the top and bottom of cast blocks and the reproducibility of the
measurements performed on 100 mm and 50 mm diameter samples. The hydrated prod-
uct has a low density (in a saturated state): 1580 kg.m-3. The propagation velocity of
the compression and shear waves is 2340 m.s-1 and 1660 m.s-1 respectively. The Poisson
coefficient thus obtained is 0.31°±°0.04. The elastic moduli then calculated are in the
order of 5000 MPa. These results are in the same order than those obtained by acoustic
resonance measurements. The thermal dilation coefficient is measured: 25.10-6 K-1.
This value appears high and remains eight time lower than the thermal dilation coeffi-
cient of HDPE used as pipe for the probe. Several tests were performed in the coupled
device. In order to reproduce the operating conditions of a VGP, during a test, different
temperature cycles were performed, which could lead to freezing and thawing of the
grout. During the test, a vertical load is applied to the grout and the upstream and down-
stream fluid pressures are controlled to generate a vertical gradient of fluid pressure.
Temperatures are measured at different points in the grout section. The material studied
is of very good quality and no alteration in its performance (from the perspective of a
796 M. Pascal et al.

geothermal application) could be observed. In the decoupled device, different degrada-

tion conditions are studied: in the first case, the grout samples are exposed to different
aggressive solutions (water, gypsum water, acid water) under isothermal temperature
conditions; in the second case, in addition to the aggressive solutions, the samples are
exposed to freeze/thaw cycles. Sample density and acoustical velocity are not specifi-
cally changed during the freeze thaw cycles or during immersion in aggressive solu-
tions. Thermal conductivity is measured after different exposure periods or freeze/thaw
cycles. The thermal conductivity is measured under submerged conditions by the hot
wire method in transient conditions. The standard deviation remains less than 10%. The
results indicate that there is very little evolution, whether in the presence of freeze/thaw
cycles or not (Figure 1). The thermal conductivity remains higher than 1.5 W.m-1.K-1
under freeze/thaw conditions, and around 2 W.m-1.K-1 under isothermal conditions
whatever the type of aggressive solution. This value is in the same order than the data
in the literature and complies with the current standard [9]. Water permeability is meas-
ured initially and then after different time periods. The measurement is carried out in a
triaxial cell under a pressure gradient in steady state (800 kPa confinement and 10 m
hydraulic gradient). The results indicate in all cases that the permeability remains below
10-9 m.s-1. Same types of results are obtained after freeze/thaw cycles. There is no clear
evolution with time or with type of immersion, as clogging and unclogging phenomena
certainly occur due to the precipitation and dissolution of chemical species in the po-
rosity during the phases of curing/alteration. Mechanical tests are carried out at constant
displacement speed (2.5°mm.min-1) on samples after immersion in aggressive solu-
tions. Reference compressive strength at 28 days is 6 MPa but the mean value obtained
during this study is 4 MPa. The performances appears quite not influenced by the type
of solution used for immersion. The most interesting behavior quoted during the test is
presented on figure 2 in the case of a cyclic compression test. The calculated stiffness
is different between the beginning of the test and before the mechanical cycles per-
formed. We observe an evolution of stiffness from one series of cycles to another (in-
crease). However, the calculated stiffness is identical for a given series. This phenom-
enon is not observable for the third cycle presented because the material is already se-
verely damaged. A phenomenon of material compaction is then observed. To complete
the study, a creep test is performed at a stress of 2.5°MPa (50 % of compressive
strength) on a specimen with a slenderness of 3 over a period of 100 minutes. The creep
deformation reaches 600°μm.m-1, which corresponds to a specific creep of 240°μm.m-
1
.MPa-1. This is certainly due to the migration of water into the material and compaction
of the material: horizontal compaction bands are observed at the end of the test. The
ductility of the grout favors the sealing of potential cracks. Geochemical models work
with data evaluated from experimental developments (mineralogy, porosity, diffusion,
diffusion coefficient, etc.). The models are in agreement with the small changes ob-
served experimentally. SEM additional profiles performed on hydrated samples for 1
year seem to be in overall agreement with the model predictions. In addition, the model
shows an agreement between the compositions of the solutions of the conservation
baths acquired after one month and after one year of hydration, which supports the
approach. From the point of view of the results of geochemical models, we can see that
the only geochemical modeling applied to the evolution of thermal conductivity (λ)
Durability Evaluation of a Geothermal Grout 797

predicts an increase at the interface with the solution. But it also predicts, after 30 years
(figure 3), a total disappearance of the hydraulic binder, in an area whose extension
depends on the envisaged scenario (1 cm to 4 cm).

Fig. 1. Evolution of thermal conductivity - left: after each freeze-thaw cycle for the three storage
media - right: at each maturity for the three storage media. Solution type: blue = water, or-
ange = gypsum water, grey = acid water, yellow = x10 stronger acid water.

Fig. 2. Left: Cyclic compression test on 28 days old specimen with a slenderness of 1 – right:
Creep test on 28 days old specimen with a slenderness of 3.

Fig. 3. Results obtained with the "alteration model". Left: Solid calcium concentration in grout
versus probe pipe – Right: distribution of mineral composition (in cm3.L-1) after 100 years.

4 Conclusions and perspectives

The study showed that the experimental devices were suitable for studying the durabil-
ity of vertical geothermal probes. The coupled and decoupled devices allow to study
798 M. Pascal et al.

the phenomena of alteration of geothermal grouts with a good reproducibility of meas-

urements. The material studied was of very good quality and no deterioration in its
performance (from the point of view of geothermal application) could be observed. An
intrinsic variability in the material may have masked the changes in the studied prop-
erties. However, geochemical models indicate possible alteration of the hydraulic
binder in 30 years close to the probe pipe.
In terms of prospects, the following work is planned:
- Continue experiments on Thermocem grout over longer periods and/or a greater num-
ber of freeze-thaw cycles to identify the durability limits of this material;
- Apply the protocols established during this research project to different types of grout
(geothermal or not), compositions and sources;
- Evolve protocols for standardization: identify the main parameters that influence the
durability of a grout, possibly reduce protocols accordingly and propose a standard op-
erating procedure routinely usable.

Acknowlegments

Duracim project was funded in the frame of the ADEME-BRGM-Université Rennes

1 agreement n°1405C0037.

References
1. Tang, A-M., Cui, Y-J. et Le, T-T.: A study on the thermal conductivity of compacted ben-
tonites. Applied Clay Science, Vol. 41, pp. 181-189 (2007).
2. Park, M., et al. : Applicability of cement-based grout for ground heat exchanger considering
heating-cooling cycles. SCIENCE CHINA - Technological Sciences, Vol. 54, 7, pp. 1661–
1667 (2011).
3. Allan, M. L.: Materials characterization of superplasticized cement±sand grout. Cement and
Concrete Research. Vol. 30, pp. 937-942 (2000).
4. Lee, C., et al. : Characteristics of thermally enhanced bentonite grouts for geothermal heat
exchanger in South Korea. SCIENCE CHINA Technological Sciences, Vol. 53, 1, pp. 123-
128 (2010).
5. Jobmann, M. et Buntebarth, G.: Influence of graphite and quartz addition on the thermo–
physical properties of bentonite for sealing heat-generating radioactive waste. Applied Clay
Science, Vol. 44, pp. 206-210 (2009).
6. Delaleux, F., et al. Enhancement of geothermal borehole heat exchangers performances by
improvement of bentonite grouts conductivity. Applied Thermal Engineering. , Vol. 33-34,
pp. 92-99 (2012).
7. Perry, R.L.et Smith, M.D.: Borehole grouting: field studies and thermal performance testing.
ASHRAE Transactions, Vol. 105 (2009).
8. Blanc, P., Bourbon, X., Lassin, A., Gaucher, E.C.: Chemical model for cement-based mate-
rials: Thermodynamic data assessment for phases other than C–S–H. Cement and Concrete
Research 40:1360–1374 (2010).
9. Standard NF X 10-970 « Forage d’eau et de géothermie – Sonde géothermique verticale
(échangeur vertical en U avec liquide caloporteur en circuit fermé) » (2011).
 $VVHVVPHQWRQWKHSHUIRUPDQFHRI(3%7%0
LQWKHFRQVWUXFWLRQRISLORWPHWUROLQHLQ+R&KL0LQKFLW\
9LHWQDP

7KX+DQJ75$1DQG0DQK7XDQ+2$1*
)DFXOW\RI&LYLOHQJLQHHULQJ8QLYHUVLW\RI7UDQVSRUWDQG&RPPXQLFDWLRQV+DQRL9LHWQDP

6KLPL]X±0DHGD-RLQW2SHUDWLRQ&3%/LQH+&0FLW\8UEDQ5DLOZD\3URMHFW9LHWQDP
WUDQWKXKDQJXWF#JPDLOFRP

$EVWUDFW/LQH%HQ7KDQK±6XRL7LHQRIWKH+R&KL0LQKFLW\8UEDQ5DLO

ZD\ 3URMHFW LV WKH SLORW PHWUR LQ 9LHWQDP DOVR WKH ILUVW WUDQVSRUW WXQQHO FRQ
VWUXFWHGE\WKHHDUWKSUHVVXUHEDODQFHGVLQJOHVKLHOGWXQQHOERULQJPDFKLQH(3%
7%0 RI ZKLFK WKH RXWHU GLDPHWHU LV P 7ZR SDUDOOHO PHWHUWXQQHOV
:HVW %RXQG DQG (DVW %RXQG KDYH EHHQ FRQVWUXFWHG IURP 0D\  7KH\
PDLQO\ORFDWHLQORRVHVDQGOD\HUVZLWKVPDOO637YDOXHVDQGWKHVPDOORYHUEXU
GHQKHLJKWVWRPRIWKHIRUPHUDQGWRPRIWKHODWWHU(3%7%0
KDYHSURYHQWKHVXLWDELOLW\IRUWKHPHWUROLQHLQVXFKJHRORJLFDOFRQGLWLRQVDQG
WKHQDUURZFRQVWUXFWLRQVLWHLQXUEDQDUHDVRI+R&KL0LQKFLW\,QWKLVSDSHUIRU
WKHDVVHVVPHQWRI(3%7%0GXULQJWKHWXQQHOFRQVWUXFWLRQVRPHPDLQRSHUDWLRQ
SDUDPHWHUVRIWKHPDFKLQHLQHDFKSHUIRUPDQFHURXWLQHKDYHEHHQDQDO\VHGDQG
WKH LQIOXHQFH RI WKH JHRORJLFDO FRQGLWLRQV DORQJ WKH DOLJQPHQW WR WKH PDFKLQH
MRXUQH\ZDVLQLWLDOO\VWXGLHG7KHVWXG\¶VJRDOLVWKHEHWWHUXQGHUVWDQGLQJRIWKH
(3%7%0SHUIRUPDQFHLQWKHILUVW7%0PHWUROLQHLQ9LHWQDPDVDQH[SHULHQFH
OHDUQLQJIRUWKHQH[WRQHVZKLFKKDYHEHHQVFKHGXOHGWREHVRRQDSSHDUHGLQ+R
&KL0LQKFLW\DQG+DQRL

.H\ZRUGV(3%7%0SHUIRUPDQFHDVVHVVPHQWPHWUR+R&KL0LQKFLW\

 ,QWURGXFWLRQ

/LQHIURP%HQ7KDQKWR6XRL7LHQLVWKHSLORWPHWUROLQHRIWKH+R&KL0LQK+&0
FLW\8UEDQ5DLOZD\&RQVWUXFWLRQ3URMHFW,WLVNPLQWRWDOOHQJWKXQGHUJURXQG
NLORPHWUHV DQG HOHYDWHG NLORPHWUHVFRQVLVWV RI WKUHH XQGHUJURXQG VWDWLRQV DQG
HOHYHQ HOHYDWHG VWDWLRQV7KH OLQHFRQQHFWVWKH FLW\ FHQWUHGLVWULFW ZLWK WKHFLW\¶V
QRUWKHDVWHUQJDWHGLVWULFW7KHSURMHFWLVH[SHFWHGWREHFRPSOHWHGVRRQWRIDFHZLWK
WKHLQFUHDVLQJGHPDQGRIWKHXUEDQWUDQVSRUWLQIUDVWUXFWXUHWRSURPRWHWKHXVHRISXEOLF
WUDQVSRUWDWLRQV\VWHPE\DPRGHUQDQGFRQYHQLHQFHFRQQHFWLRQWRPLWLJDWHWKHXVHRI
SULYDWHYHKLFOHVWUDIILFMDPDQGSROOXWLRQLQ+&0FLW\7KH(DVW%RXQG7XQQHO(%
7XQQHOFRQQHFWVWKHQGXQGHUJURXQGIORRURI%D6RQ6WDWLRQDQGWKHWKXQGHUJURXQG
IORRURI2SHUD+RXVH6WDWLRQZKLOHWKH:HVW%RXQG7XQQHO:%7XQQHOOLQNVWKHQG
XQGHUJURXQGIORRUVRIWZRVWDWLRQV7KHERUHGWXQQHOLVPLQLQQHUGLDPHWHUP

© Springer Nature Singapore Pte Ltd. 2020 799

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_127
800 T.-H. Tran and M.-T. Hoang

LQRXWHUGLDPHWHUXVLQJWKHPWKLFNSUHFDVWVHJPHQWDOFRQFUHWHOLQLQJDQGWKHRYHU
EXUGHQKHLJKWYDULHVIURPWRP7KHFRQVWUXFWLRQSDFNDJH%ZDVH[HFXWHGIRUWKH
P XQGHUJURXQG VHJPHQW ZKHUH WKH WZLQ P 7%0 WXQQHO VWD\V EHWZHHQ WKH
2SHUD+RXVHVWDWLRQDQG%D6RQVWDWLRQ'LIIHUHQWFRQVWUXFWLRQPHWKRGVKDYHEHHQDS
SOLHGLQWKLVSDFNDJHDQGWKHGHWDLOVFDQEHVHHQLQ)LJXUH


)LJ&RQVWUXFWLRQPHWKRGVRI&3%RI/LQH+&0&LW\XUEDQUDLOZD\QHWZRUN>@

7KHVKLHOGWXQQHOERULQJPDFKLQH7%0XVLQJWKHHDUWKSUHVVXUHEDODQFLQJ(3%LV

WKHLGHDOL]HGWXQQHOOLQJPDFKLQHZKHQSURYLGLQJDWWKHVDPHWLPHWKHIDFHVWDELOLW\DQG
WKHSURWHFWLRQRIWKHVXUURXQGLQJJURXQGE\FRQWUROOLQJWKHHDUWKSUHVVXUHWKURXJKWKH
FRRUGLQDWLRQRIWKHGULYLQJVSHHGDQGWKHPXFNUHPRYLQJUDWH>@(DFKZRUNLQJURXWLQH
RIWKHPDFKLQHLQYROYHVWKHH[FDYDWLRQE\WKHVSRNHW\SHFXWWHUKHDGZLWKWKHHOHFWULF
LQYHUWHUPRWRUGULYLQJWKHIDFHVWDELOL]DWLRQE\WKHK\GUDXOLFWKUXVWMDFNDQGWKHHDUWK
SUHVVXUHEDODQFLQJE\VOXGJHSXPSWKHDGYDQFHE\WKHK\GUDXOLFWKUXVWMDFNWKHPXFN
LQJE\WKHVKDIWW\SHVFUHZFRQYH\RUWKHWXQQHOOLQLQJHUHFWLRQE\WKHULQJJHDUW\SH
VHJPHQWHUHFWRUHWF>@7KH7%0ZRUNZDVVWDUWHGLQ0D\$IWHUILQLVKLQJWKH
(% 7XQQHO IURP %D6RQ WR 2SHUD +RXVH 6WDWLRQ WKH PDFKLQH ZDV GLVPDQWOHG DQG
WUDQVSRUWHGEDFNWRPDQLIHVWWKHVDPHMRXUQH\RIWKH:%7XQQHO8SWRWKHHQGRI
PRUHWKDQRIWKHFRQVWUXFWLRQZRUNKDVEHHQFRPSOHWHG7KHSURMHFWLVVFKHGXOHG
WREHILQLVKHGDWWKHHQGRI,WLVWKHILUVWWUDQVSRUWSURMHFWLQ9LHWQDPZKHUHWKH
7%0KDVEHHQXVHG,QWKLVSDSHUWKH7%0¶VSHUIRUPDQFHDORQJWKHDOLJQPHQWZDV
VWXGLHG6RPHW\SLFDORSHUDWLRQVRIWKHPDFKLQHZHUHDQDO\VHGDQGWKHLQIOXHQFHRIWKH
JHRORJLFDOFRQGLWLRQVWRWKHPDFKLQHMRXUQH\ZDVLQLWLDOH[DPLQHG

 *HRWHFKQLFDOFRQGLWLRQVRIWKHERUHGWXQQHOV

7KHXQGHUJURXQGRIWKHWZRP7%0WXQQHOVZDVW\SLFDOIRUWKHVRIWVRLOZLWKXQ
GHUJURXQGZDWHURI+&0FLW\:KLOHWKH:%7XQQHOVWD\VDPRQJWZRDOOXYLXPOD\HUV
$VDQG$VWKH(%7XQQHOFRQWDFWVSUHGRPLQDQWO\ZLWKWKH$VOD\HU%RWKWXQQHOV
SDVVVRPHORFDWHGOLTXHILDEOH]RQHVRIWKH$VOD\HU$VLPSOLILHGJHRORJLFDOPDSDORQJ
WKHDOLJQPHQWZDVSUHVHQWHGLQ)LJXUH
Assessment on the performance of EPB-TBM … 801


)LJ6LPSOLILHGJHRORJLFDOSURILOHRIWKH7%0WXQQHOVRI/LQH>@

7KHGHWDLOHGJHRWHFKQLFDOSDUDPHWHUVRIWKHXQGHUJURXQGFDQEHIRXQGDW7DEOH

7DEOH*HRWHFKQLFDOSDUDPHWHUVRIWKHERUHGWXQQHOXQGHUJURXQG>@

6RLOOD\HU 6XEOD\HU 7KLFNQHVV %XONGHQVLW\ 637YDOXH

PP N1P EORZVFP
)LOO)  ·  ·
$OOXYLXP &OD\OD\HU$F ·  
 6LOW\ILQHVDQGOD\HU$V ·  
 6DQGOD\HU$V ·  
'LOXYLXP +DUGFOD\H\VLOW'F ·  
 'HQVHVLOW\VDQG'V   
5DQJHGIURPPWRPEHORZWKHJURXQGOHYHOWKHJURXQGZDWHUWDEOHLVDOZD\V
DERYHWKHWZLQWXQQHOV7KHXQGHUJURXQGZDWHULVS+QHXWUDODQGWKHDYHUDJHSHUFHQW
E\ZHLJKWRIFKORULGHDQGVXOSKDWHLVDQGUHVSHFWLYHO\>@,WFDQEH
XQGHUVWRRGDVDORZDJJUHVVLYHFRUURVLYHHQYLURQPHQWWRWKHSUHFDVWUHLQIRUFHGWXQQHO
VHJPHQWIROORZLQJWKHFXUUHQW9LHWQDPHVHVSHFLILFDWLRQ>@

 3HUIRUPDQFHRI(3%7%0

7KH UHSHDWHG RSHUDWLRQ MRXUQH\ RI WKH(3%7%0 RI /LQH  ZDVSURJUDPPHG RI WKH
PH[FDYDWLRQIROORZLQJE\WKHOLQLQJVHJPHQWHUHFWLRQIRUHDFKWXQQHO¶VIDFH(DFK
ERUHGWXQQHOFRQWDLQVRISUHFDVWUHLQIRUFHGOLQLQJVHJPHQWVHTXLYDOHQWZLWK
ULQJV)RUWKHDVVHVVPHQWRIWKH(3%7%0¶VSHUIRUPDQFH GXULQJWKHFRQVWUXFWLRQRI
WKH WZR ERUHG WXQQHOVVRPH RI WKHW\SLFDO RSHUDWLRQDO IHDWXUHV RI WKH PDFKLQH ZHUH
VWXGLHG)LUVWO\WKHORFDWLRQHUURURIWKHWXQQHO¶VFHQWUHDORQJWKHDOLJQPHQWZDVDQD
O\VHG7KHVSDWLDOGHYLDWLRQRIWKHWXQQHO¶VFHQWUHSRLQWZDVFDOFXODWHGE\
మ
 ' ൌ ඥሺܺ െ ܺ଴ ሻଶ ൅ ሺܻ െ ܻ଴ ሻଶ ൅ ሺܼ െ ܼ଴ ሻଶ 

ZKHUH;<=ZDVWKHGHVLJQHGFRRUGLQDWHDQG;<=ZDVWKHDFWXDOFRRUGLQDWH

7KHWHQGHQF\RIWKHVSDWLDOGHYLDWLRQRIWZRWXQQHOVZDVVNHWFKHGLQ)LJXUH
802 T.-H. Tran and M.-T. Hoang


)LJ6SDWLDOGHYLDWLRQRIWKHFHQWUHRIWZRERUHGWXQQHOVGXULQJWKH(3%7%0MRXUQH\

6LQFHWKHJLYHQPD[LPDOWROHUDQFHPDUJLQZDVPP>@WKHSRVLWLRQDFFXUDF\RIWKH
:% DQG (% 7XQQHO ZDV IRXQG WR EH VDWLVILHG $Q LVRODWHG P WKLFN OD\HU RI ORZ
SODVWLFLW\ILUPVDQG\FOD\$FVWD\LQJRQWKHFURZQRIWKH:%WXQQHOZDVFRQVLGHUHG
DVWKHFDXVHIRUWKHODUJHVWFRRUGLQDWHHUURURIWKH:%7XQQHOIURP5LQJFKDLQDJH
WR5LQJFKDLQDJH
6HFRQGO\WKHSHQHWUDWLRQUDWH35PKZDVVWXGLHGE\DSSO\LQJWKHHTXDWLRQ>@

 35 SH1

ZKHUHSHPPLVWKHSHQHWUDWLRQGHSWKSHUUHYROXWLRQRIWKHFXWWHUKHDGDQG1USPLV

WKHFXWWHUKHDGURWDWLRQUDWH1 USP>@7KHFRUUHODWLRQEHWZHHQWKHSHQHWUDWLRQ
UDWH 35 DQGWKHWXQQHOOHQJWK RIWKH(%7XQQHOZDVVNHWFKHG DQGWKHDYHUDJH YDOXH
35DYHUDJH PKZDVIRXQG


)LJ&RUUHODWLRQEHWZHHQWKH7%0¶VSHQHWUDWLRQUDWHDQGWKHOHQJWKRIWKH(%7XQQHO

7KLUGO\WKHWRWDOWLPHIRUPERUHGWXQQHO7KZDVGHWHUPLQHGDVWKHIUDFWLRQRIWKH
WLPHGXUDWLRQIRURQH7%0ULQJRYHUWKHREWDLQHGWXQQHOOHQJWK+HQFHWKHFRUUHODWLRQ
EHWZHHQWKDWSDUDPHWHUDQGWKH(%7XQQHO¶VOHQJWKZDVREWDLQHG7ZRDYHUDJHYDOXHV
RI7IRUWKH(%7XQQHOZHUHFDOFXODWHGD8VLQJDOOGDWD7DYHUDJH KE6NLSSLQJ
DOOYDOXHV7LODUJHUWKDQWLPHV7DYHUDJH7FRUUHFWHGDYHUDJH K6LQFHWKHWLPHUHTXLUHG
IRUH[FDYDWLQJPWXQQHOZDVGHWHUPLQHGDVWKHLQYHUVLRQRIWKHSHQHWUDWLRQUDWH>@
WKHPHDQYDOXHRIKZDVIRXQG7KHFXUYHDQGWKHPHQWLRQHGYDOXHVZHUHVNHWFKHG
LQ)LJXUH
Assessment on the performance of EPB-TBM … 803


)LJ&RUUHODWLRQEHWZHHQWKHWRWDOWLPHIRUPERUHGWXQQHODQGWKHOHQJWKRIWKH(%7XQQHO

7KHWLPHIRUWKHH[FDYDWLRQZDVDQGRIWKHWRWDOFRQVWUXFWLRQWLPH
HTXLYDOHQW ZLWK 7DYHUDJH DQG 7FRUUHFWHG DYHUDJH UHVSHFWLYHO\ &RPSDUHGZLWKWKH GHVLJQHG
YDOXHRI>@WKHPDFKLQH¶VUHDOWLPHSHUIRUPDQFHZDVIRXQGUDWLRQDO
)RXUWKO\ WKHHIIHFW RIWKHJHRORJLFDOFRQGLWLRQDOWHUDWLRQWRWKH 7%0¶VSHUIRUPDQFH
ZDVH[DPLQHG ,WLVUHFDOOHGWKDWWKH(% 7XQQHO VWD\V PDLQO\ LQ WKH $V OD\HU WKHQ
JUDGXDOO\FKDQJHVWRWKH$VOD\HUILQDOO\IDOOVLQWRWKHOLTXHILDEOH]RQH&RQVLGHULQJ
WKH VXUURXQGJURXQGOD\HUFRQWDFWHGGLUHFWO\ZLWKWKHWXQQHOWKH WXQQHOZDVGLYLGHG
LQWRWZRSDUWVD4XDVLKRPRJHQRXVFKDLQDJHIURPWRE+HWHURJHQH
RXV FKDLQDJH IURP  WR  +HQFH WKH WXQQHO VHJPHQW IURP FKDLQDJH
ULQJWRFKDLQDJHULQJZDVH[SHFWHGWRZHOOFRYHUWKH
JHRORJLFDOWUDQVIHUULQJ]RQH7KHDPRQJYDULRXVPDFKLQH¶VPDLQSDUDPHWHUVWKHFDVH
RIWKHWKUXVWIRUFHZDVDQDO\VHG



D,QWKHH[DPLQHGVHJPHQW E,QVHSDUDWHGULQJVDORQJWKHVHJPHQW
)LJ9DULDWLRQRIWKH7%0¶VVFUHZURWDWLRQVSHHGRIWKH(%7XQQHO

,Q OHIW JUDSK WKH YDULDWLRQ RIWKHDYHUDJHWKUXVW IRUFH LQWKH H[DPLQHG VHJPHQW ZDV
SUHVHQWHGZKLOHLQWKHULJKWRQHWKHFRUUHODWLRQEHWZHHQWKHWKUXVWIRUFHDQGWKHGULYLQJ
OHQJWKLQVHSDUDWHGULQJVDORQJWKHVHJPHQWZDVVNHWFKHG7KH\ERWKLQGLFDWHGDGH
FUHDVHWHQGHQF\RIWKHWKUXVWIRUFHZKLOHPRYLQJIURPWKHTXDVLKRPRJHQRXV]RQHWR
WKHKHWHURJHQHRXVRQH+RZHYHUEHVLGHWKHVRLOOD\HUWUDQVIHUULQJWKHFRPSOLFDWHJH
RWHFKQLFDOFRQGLWLRQVFDXVHGE\WKHGLVWDQFHIURPWKHWXQQHOH\HEUHDNWKURXJKDQGIURP
WKHXQGHUJURXQGZDWHUWDEOHWKHORFDWLRQRIWKHOLTXHILDEOH]RQHWKHQHLJKERXULQJXU
EDQLQIUDVWUXFWXUHVFUHDWHGGLIILFXOWLHVIRUWKHFRPPHQWDU\RQWKHLQIOXHQFHWRWKHPHQ
WLRQHG SDUDPHWHU 'XH WR WKH VKRUWDJH RI DYDLODEOH LQSXW GDWD QR IXUWKHU FRUUHODWLRQ
804 T.-H. Tran and M.-T. Hoang

EHWZHHQWKHJHRORJLFDOFRQGLWLRQVDQGWKHWKUXVWIRUFHZDVIRXQGLQWKHVFRSHRIWKLV
VWXG\

 &RQFOXVLRQ

7KHSHUIRUPDQFHRIWKH(3%7%0GXULQJWKHFRQVWUXFWLRQRIWKHWZLQERUHGWXQQHORI
/LQH+&0FLW\XUEDQUDLOZD\QHWZRUNZDVDQDO\VHGEDVHGRQWKHUHFRUGHGGDWDRI
WKH PDFKLQH GXULQJ WKH RSHUDWLRQ 7KH DFFXUDF\ RI WKH ERULQJ ZRUN ZDV H[DPLQHG
WKURXJKWKHVSDWLDOGHYLDWLRQRIWKHFHQWUHSRLQWRIWZRWXQQHOV6RPHPDLQSHUIRUPDQFH
SDUDPHWHUVRIWKH(%7XQQHOZHUHDVVHVVHGDQGFRPSDUHGZLWKWKHGHVLJQHGYDOXHVDV
VKRZLQJLQ7DEOH

7DEOH7%0¶VSHUIRUPDQFHSDUDPHWHUVGXULQJWKHFRQVWUXFWLRQRIWKH(%7XQQHO

3DUDPHWHU $VVHVVHGYDOXH 'HVLJQHGYDOXH>@ 8QLW

$GYDQFHUDWH   PPPLQ
([FDYDWLRQWLPH   PLQULQJ
&\FOHWLPH   PLQULQJ
:LWKWKHDYHUDJHSHQHWUDWLRQUDWH35 PKDQGWKHDYHUDJHDGYDQFHUDWH$5 
PKWKHGHWHUPLQHG7%0XWLOL]DWLRQ8 ZDVDWWDLQHG7KHHIIHFWRIWKH
JHRORJLFDOFRQGLWLRQWRWKH7%0SHUIRUPDQFHKDVEHHQLQLWLDOO\IRXQGLQWKLVUHVHDUFK
$QRWKHUVWXG\LVDLPHGDWIXUWKHUH[DPLQLQJLW7KHFDVHVWXG\ KDVSURYLGHGDEHWWHU
XQGHUVWDQGLQJRIWKH7%0WHFKQRORJ\,WZDVDQHVVHQWLDOH[SHULHQFHIRUWKHXSFRPLQJ
XUEDQWUDQVSRUWSURMHFWVLQ9LHWQDPZLWKWKHQH[WDSSOLFDWLRQIRUWKHSLORWPHWUROLQH
LQ+DQRL

5HIHUHQFHV
 6KLPL]X0DHGD -RLQW 2SHUDWLRQ ,QWURGXFWLRQ YLGHR RI &RQWUDFW 3DFNDJH E &LYLO
8QGHUJURXQG6HFWLRQIURP.0WR.0+R&KL0LQKFLW\
 :RUNLQJ JURXS IRU VKLHOG WXQQHOV RI -6&( 6WDQGDUG VSHFLILFDWLRQV IRU WXQQHOOLQJ 
6KLHOGWXQQHOV-DSDQ6RFLHW\RI&LYLO(QJLQHHUV7RN\R
 6KLPL]X0DHGDMRLQWRSHUDWLRQ0HWKRGVWDWHPHQWIRUERUHGWXQQHOFRQVWUXFWLRQRI&RQWUDFW
3DFNDJHE+&0&LW\8UEDQ5DLOZD\&RQVWUXFWLRQ3URMHFW+R&KL0LQK&LW\
 6KLPL]X0DHGD-RLQW2SHUDWLRQ5HSRUWRQ*HRORJ\RI&RQWUDFW3DFNDJHE+&0&LW\
8UEDQ5DLOZD\&RQVWUXFWLRQ3URMHFW+R&KL0LQKFLW\
 6KLPL]X0DHGD-RLQW2SHUDWLRQ*HRWHFKQLFDOH[SODQDWLRQUHSRUWIRUERUHGWXQQHORI&RQ
WUDFW3DFNDJHE+&0&LW\8UEDQ5DLOZD\&RQVWUXFWLRQ3URMHFW+R&KL0LQKFLW\
 0LQLVWU\ RI 6FLHQFH DQG 7HFKQRORJ\ RI 9LHWQDP 7&91  $QWLFRUURVLRQ LQ
FRQFUHWH DQG UHLQIRUFHG FRQFUHWH VWUXFWXUH FRQVWUXFWLRQ  &ODVVLILFDWLRQ RI HURVLRQ
HQYLURQPHQW+DQRL,Q9LHWQDPHVH
 6KLPL]X0DHGD-RLQW2SHUDWLRQ7HFKQLFDOVSHFLILFDWLRQVRI&RQWUDFW3DFNDJHE +&0
&LW\8UEDQ5DLOZD\&RQVWUXFWLRQ3URMHFW+R&KL0LQKFLW\
 0RRQ + $ &RPSDUDWLYH 6WXG\ RQ )LYH $VVHVVPHQW 6\VWHPV RI 7%0 7XQQHOOLQJ
*HRV\VWHP(QJLQHHULQJ
 Reusability of muds dredged from lakes in Hanoi City
from its geotechnical engineering characteristics

Pham Tri Thuc1 and Phan Huy Dong2

1 Military Academy of Logistics, Hanoi, Vietnam
trithuchvhc@gmail.com
2 National University of Civil Engineering, Hanoi, Vietnam

phdong65@gmail.com

Abstract. This study is aimed at discussing on the typical geotechnical engi-

neering characteristics of sludge dredged from lakes in Hanoi city, which have
strong demand to be dredged for environmental urban. Based on which, the
possibility of recyclability as filling material will be discussed. Sludge samples
were collected from “West Lake” and “Hoan Kiem Lake”. A series of laborato-
ry tests for analyzing the typically physical and chemical characteristics has
been conducted. The X-ray diffraction tests (ASTM C114) and plasma spec-
trum analyzer tests (TCNB-ICP 01/04) were also carried out to investigate the
chemical components of the muds. The test results show that the chemical com-
ponents meet the National technical regulation on hazardous waste threshold
and they can be disposed as normal fill. Overview on the experiences of cement
treated soil in Japan reveals that the muds dredging from lakes in Hanoi City
have high possibility for recycling as filling material.

Keywords: Sludge; dredging; chemical analysis, cement treated soil;

1 Introduction

For sustainable development as well as environmental improvement, Hanoi City has

strong demand for dredging large amount of sludge sediment in lakes within the City,
such as “West Lake” and “Hoan Kiem Lake”, such as in West Lake a dredging pro-
ject in which about 1.3 million m3 of dredging mud has been launched recently.
So far most of the dredging muds has been dumped at the designated disposal sites.
However, the dumping will occupy a consideration land fill area (land fill area for
dumping dredged muds from West Lake is estimated at about 20 hectares) whereas
the land fund and cost for dumping such huge amount of dredged mud becomes a big
issue. It is also noted that a relatively huge amount of sludge has been illegally
dumped, which causes serious pollution and influences on the environmental hygiene
and landscape and becomes a terrible problem of the whole society.
It can be seen that the dredged muds can not be directly reused as filling material
since it has very high water content and very low strength, the most typical treatment
method is that the muds are premixed with a specific amount of among cement, lime,
paper sludge, etc. The hydradtion reaction between soil particles, free water in mud,

© Springer Nature Singapore Pte Ltd. 2020 805

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_128
806 P. T. Thuc and P. H. Dong

and cement will happened, as the results the soil mass can be archived specific
strength after solidification. The reuses of dredging muds from lakes, rivers, city sew-
er system, water treatment plants, etc, has been developed and become practical uses
in many countries round the world. Various methods for muds stabilization and treat-
ment have been developed over the world. The reused quantity of mud over the total
generated amount before 2009 is reported by Sheehan et al., 2009 [8] as follows:
Japan (90%), USA (20%-30%), Holand (23%), Ireland (20%)... At present, Japan has
a number of domestic trading companies that specializes only in treating and recy-
cling the sludge to become backfill material. However, the research works as well as
application have still inadequate in Vietnam circumstances.
In order to promote the reuse of dredged mud, sludge from lakes, water sewer sys-
tem, and water treatment plants, this paper presents the tests results for chemical and
physical analysis. In addition, the typical treatment methods for reuses of dredged
muds as filling material are also introduced.

2 Influencing factors for cement treated soil

The performance of cement treated soil is mostly observed based on the value of un-
confined compressive strength value. It has been demonstrated that the development
of soil strength is governed by several factors as: 1. Soil characteristics and condi-
tions; 2. Properties of additives and binders; 3. Mixing conditions and 4. Curing con-
ditions. In which, influences of soil properties includes physical, chemical and miner-
alogical characteristics of the soil; organic matter content in soil; Potential hydrogen
(pH) of pore water; water content in the soil. This is the inherent characteristic of the
soil and it represents the deposition of each soil type. Soil properties and conditions
are particularly important for clay [1], [2].
The mechanism for the development of soil strength after reuse is closely related to
the chemical reaction between soil and binder. As such, the chemical characteristics
have a strong impact on reusability. With the same mixing conditions, the same
amount of binder, the strength of the soil varies considerably with each soil type.
The unconfined compressive strength (qu) of cement treated soil using a similar
cement content (20% cement content in wet weight of mud) are summarized by
Kitazume et al. [2]. As shown in Fig.1, the qu values of 21 different soil types with
different specific gravity, water content, organic content and pH value. It can be
demonstrated that the water content, organic content and specially pH values existing
in original muds are the clearest factors that governs the development of strength of
treated soil.
Reusability of muds dredged from lakes in Hanoi City … 807

Fig. 1. The influence of soil type in cement stabilization (Nina et al., 1981 [2])

3 Charactezation of chemical properties of muds dredging from

lakes in Hanoi city.

3.1 Sampling method

With very large area, it is very difficult to describe the general characteristics of the
whole sludge in the lake bed. For a general study, sludge samples were collected in
different locations overall of the lakes. Sludge samples in the study included 2 types:
undisturbed and disturbed sample. The disturbed samples were taken using small
bucket within depth of 0 m to 0.5m from bottom of the lake. Whereas, the undisturbed
samples were collected by using a thin metal tubes at depth bellow 0.5m from bottom
of the lake.

Fig. 2. Location and image of sludge samples in Hoan Kiem Lake

3.2 Test results and discussions

A series of laboratory tests have been conducted to obtain the water content, particle
size distribution, organic content and pH value. Several test results are summarized in
Table 1 with notes that the sample numbered as “H” are collected from Hoan Kiem
Lake, whereas the ones named with letter “W” means muds from West Lake.
808 P. T. Thuc and P. H. Dong

Table 1. The typical physical properties of muds dredged from West and Hoan Kiem Lakes

No.
H01 H02 H03 H04 W01 W02 W03 W04 W05 W06 W07 W08 W09
Value
Water con-
76.8 61 65 65 211 203 216 204 195 202 209 212 199
tent, w (%)
Organic
7.05 4.5 5.8 5.6 19 22 11 20 15 18 21 22 16
content, (%)
pH 7.05 7.1 7.1 7.2 7.1 7.2 7.1 7.2 7.2 7.3 7.3 7.2 7.1

Fig. 3. Laboratory test for obtaining physical parameters of the muds

Water content of soil

The mud samples in the West Lake, in the depth of 0.5 m from the lake bottom give
very high water content as it varies from 195.4% to 212.3%. However, the test results
of undisturbed samples show that the water content varies between 61.2% to 76.8%.
This is an estimate of the moisture content of clay samples reported by Kitazume et
al. [2].
Thus, for the sludge layer with the surface of many organic compounds, high mois-
ture content, it is important to take measures to control moisture when recycling it.
The best moisture used for recycling sludge is the moisture near the plastic limit val-
ue.

Organic content
The test results for determination of organic matter content in West Lake and Hoan
Kiem Lake show that the content of organic compounds in the surface layer is quite
high, from 15.6% to 21.5%. Therefore, when recycling the sludge, it is necessary to
remove the surface layer. This sludge layer is expected to be about 0.5m to 1m in
Reusability of muds dredged from lakes in Hanoi City … 809

thickness. The organic matter content in depth below the surface is from 4.48% to
7.05%.

pH test
Since the pH effects on unconfined compressive strength. Most of soils with a pH
value below 5 show a smaller strength than those with a pH value over 5 for the same
binder content [2]. Test results for soil sample in West Lake and Hoan Kiem Lake
gives value pH > 7. This is the appropriate value for recycling sludge.

Grain size distribution

To determine the effect of grain size distribution on soil strength, Niina et al., 1977
tested the mixing of alluvial clay samples with sandy soil samples. The highest soil
stabilization can be achieved by containing 60% sand with a grain size distribution as
shown, regardless of cement content [2]. The results of experiments on soil samples at
Hoan Kiem Lake showed, sand grain content accounted for 24.9% to 38.2%. Mean-
while, sludge samples in the West Lake accounted for 40.7% to 58.2%. Such sand
content can give high strength of soil when recycling [2].

Fig. 4. The grain size distribution in sludge soil in West Lake, Hoan Kiem Lake

Heavy metal content.

In order to check contents of heavy metals for accessing the hazardous threshold, a
series of chemical test analysis has been conducted on PV7300 Plasma emission
spectroscopy according to TCNB-ICP 01/04 and X ray-diffraction test. The test
results are shown in Fig. 5.
Compared with the permissible limit of lead content (Pb) in sludge, the lead con-
tent (Pb) of the material samples varies from 59.3 ppm to 147.22 ppm, below the
allowable threshold. According to the national technical regulation on hazardous
thresholds for sludge from water treatment process
810 P. T. Thuc and P. H. Dong

Fig. 5. Result of testing the metal content in sludge soil in West Lake, Hoan Kiem Lake

4 Conclusion

The results of the study and analysis show that the muds samples collected at West
Lake and Hoan Kiem Lake have high possibility for treatment and reuse as filling
material.
Heavy metal contents are relatively below the threshold according to the national
technical regulation on hazardous thresholds for sludge from water treatment process
[3]. Water content in the soil less than the adverse limit of 200% [2]. Organic matter
content is from 4.48% to 21.5%. While organic content is considered unfavorable
when it is more than adverse limit 15% [2]. Experimental results of soil samples in
Ho Tay and Hoan Kiem shows pH value is > 7. This is the suitable value when
recycling sludge [2]. Sand grain content is which is in consitence with the study by
Babasaki et al., 1996a, 1996b [2], with sand grain content of 40%-60%, soil strength
will be higher.

References
1. Phan Huy Dong (2017), Study on some physical and chemical properties of sludge sam-
ples dredged in Hanoi and propose solutions for recycling sludge into materials, Vietnam
Construction Magazine, page 263-267.
2. M.Kitazume, The Pneumatic Flow Mixing Method, Tokyo Institute of Technology, Tokyo,
Japan, 2016.
3. QCVN 07:2009/BTNMT, National Technical Regulation on Hazardous Waste Thresholds,
Viet Nam, 2009.
 Numerical study of pile reinforced slope-A case at Khe
Cham coal preparation construction site project
(Vietnam)

Nguyen Van Dung1, Nguyen Phi Hung1 and Tan Do1*

1 Hanoi University of Mining and Geology, Duc Thang, Bac Tu Liem, Ha Noi, Vietnam
domanhtan@khoaxaydung.edu.vn

Abstract. Within the framework of Khe Cham coal preparation construction

site project, evaluation of slope stability and solution of slope reinforcement are
essential. A site investigation was performed first along a slope located in Khe
Cham coal preparation construction site. The slope reinforcement was recom-
mended due to a group of joints and cracks observed along slope surface. The
main purpose of this study is to confirm the stability condition of the unrein-
forced slope and optimize the pile reinforcement (i.e., pile position, pile spac-
ing). A numerical modeling software, Flac program, was used for this investiga-
tion. Soil properties using in the simulation model were obtained by the site in-
vestigation. Groundwater table, which was assumed to be in the top of the
slope, was also considered in the model. The safety factor based-optimization
analysis was performed with respect to pile position and pile spacing. Shear
strength reduction method was established to determine the safety factor of the
slope. As a result, the same conclusion as site investigation was found for the
unreinforced slope, that is, the slope should be reinforced for the long-term sta-
bility. In addition, the sensitivity of the slope safety factor on pile position and
pile spacing was discovered. Eventually, the optimization of pile reinforced
slope was chosen and proposed based on the sensitivity of safety factor.

Keywords: Numerical study, pile reinforced slope, Khe Cham project.

1 Introduction

Khe Cham coal preparation construction site is a key ongoing project of the coal in-
dustry planning to 2020. This project has been building up for the purpose of coal
preparation in Khe Cham and Cao Son areas before supplying to Mong Duong Ther-
mo-electric Center. During the construction, evaluation of slope stability and solution
of slope reinforcement are essential. With the subject of reinforcing slope, there have
been a number of theories suggested for the evaluation. Among them, the limit equi-
librium method (LEM), finite element method (FEM), and finite difference method
(FDM) were the most widely used approaches in the literature. The advantages of
using LEM are simple and easy ways for the analysis. However, the limit equilibrium
method has been claimed to be inaccurate due to many assumptions. On the other

© Springer Nature Singapore Pte Ltd. 2020 811

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_129
812 N. V. Dung et al.

hand, FEM and FDM are considered more accurate and over limit equilibrium meth-
ods without assumptions of shape or location of the failure surface, slice side forces
and their directions.
In this study, a Finite Difference Method program, Flac 2D, was adopted to simu-
late the pile reinforced slope. Soil properties with the groundwater table using in the
simulation model were obtained by the site investigation. A total of 16 sensitive simu-
lation case studies were performed to determine FOSs of slope with respect to the pile
positions (i.e., piles at the toe, middle, and top of slope) and pile spacing (i.e., 1 m, 2
m, 3 m, 5 m, and 8 m). The FOSs of slopes were determined based on the shear
strength reduction method coupling in Flac program. Eventually, an optimization
analysis for the actual slope was accomplished based on the determined FOSs.

2 Methodology

2.1 Site descriptions

The actual slope simulated in this study is located in Mong Duong, Cam Pha, Quang
Ninh, Vietnam (Fig. 1). The slope was constructed as part of Khe Cham coal prepara-
tion construction site. The slope consists of thick layers (up to 30 m) of sandy soil and
silty soil extending to the sandstone layer, which is beneath the surface ground of the
coal preparation construction site.

Fig. 1. Location of the actual slope (Khe Cham coal preparation construction site)

Fig. 2. Joints and cracks along the surface of the actual slope
Numerical study of pile reinforced slope … 813

The site investigation found some joints and cracks systems observed locally along
the slope surface (Fig. 2). In this sense, the stability of the actual slope could be in
doubt and the reinforcement should be executed before the further construction works
of the coal preparation construction site. Based on the site investigation and slope
condition (i.e., lateral movement-induce joints and cracks), the pile reinforced slope
(i.e., 45 m of pile length) was recommended. The pile reinforced slope method have
been widely applied for unstable slopes with lateral soil movements [1-4]. In this
reinforce system, the piles are fixed firmly and deeply into bedrock or a stable layer
[5].

2.2 Slope stability analysis using numerical modeling

In general, geotechnical engineers primarily conduct slope design based on the calcu-
lated factor of safety values. In this study, the shear strength reduction method was
established to determine the safety factor of the slope. This method can simulate the
coupled pile-slope interaction and then calculate FOSs. In this technique, cohesion (c)
and internal frictional angle (ࢥ) values of the slope materials are reduced until the
failure occurs and with the assumption that the failure mechanism of slope is directly
related to the development of shear strain. Totally, six blocks were generated for the
full slope model. The toe of the grid was pinned, and its lateral boundaries were sup-
ported by rollers as shown in Fig. 3. A relatively fine mesh was applied near the slope
and the mesh modeling became coarser further away from the slope. The row of piles
reinforcing slope was modeled by pile elements available in Flac 2D program. The
piles were all created with 10 segments per pile.

Fig. 3. Boundary condition and finite difference mesh of the actual slop

Table 1. Input parameters for numerical analysis

Colluvi Sandy Silty Sandst

Parameters Shale Unit
um soil soil one
Unit weight 20 2.65 2.62 2.72 2.67 kN/m3
Young’s modulus 21.6 34 30 20000 24000 MPa
Poisson’s ratio 0.26 0.25 0.26 0.25 0.22 -
Internal friction angle 30 37 28 38 40 degree
Cohesion 30 40 36.2 10000 13000 kPa
814 N. V. Dung et al.

3 Results and analysis

From the computed results with Flac 2D program, The factor of safety (FOS) of the
unreinforced slope was found to be 1.21, (i.e., lower than the critical FOS of 1.25,
which is the FOS requirement for a safe slope as reported in TCVN [6]. In this sense,
the same conclusion as site investigation was found for the unreinforced slope, that is,
the slope should be reinforced for the long-term stability (i.e., pile reinforcement was
recommended by the investor). The slip surface of the actual slope can be clearly seen
in Fig. 4.

Fig. 4. Slip surface of the unreinforced slope

The further works are to optimize the pile reinforcement for the actual slope with
respect to the pile positions (i.e., piles at the toe, piles at the middle, and piles at the
top of slope) and pile spacing (i.e., 1 m, 2 m, 3 m, 5 m, and 8 m). The pile location
plays a key role in the reinforced slope analysis. While Ito, Matsui [1], Poulos [2]
recommended the piles at the middle of the slope, Lee, Hull [7], Nian, Chen [8] had a
different suggestion as piles at the toe. However, the optimal pile position depends on
various factors of slope such as dimension, soil profiles, and the groundwater table.
Therefore, it is essential to investigate for the optimal pile position for every actual
slope. The safety factor based-optimization analysis (i.e., pile position) is shown in
Fig. 5. As a result, the safety factors of the actual slope were highest as the pile in-
stalls at the middle of the slope, regardless of the pile spacing. In the case of the pile
installed at the upper top of the slope, the safety factor was slightly higher than that of
the unreinforced slope. More interestingly, there was no change in the safety factor
(i.e., same FOS as that of the unreinforced slope) in the case as piles installs at the toe
of the slope. This observation can be explained by the lateral movement of piles as
shown in Fig. 6. A significant horizontal sliding of the piles can be observed when
piles are placed at the middle of the slope. This movement becomes smaller as piles
install at the upper top and toe of the slope. Based on these findings, the suggestion
was made, that is, the optimal pile position for the actual slope in this study is in the
middle.
Numerical study of pile reinforced slope … 815

Fig. 5. The safety factor based-optimization analysis for the reinforced slope

In addition, the effect of pile spacing on the safety factor of slope was discovered in
this study (Fig. 5). The safety factor decreased as the pile spacing increased (i.e., a
significant decrease in the case of pile at the middle and a slight decrease in the case
of pile at the upper top), except for the case pile at the toe. As to the recommended
case (i.e., piles at the middle of slope), FOSs reduced from 1.62 to 1.40; 1.34; 1.28;
and 1.25 as the pile spacing increased from 1 m to 2 m; 3 m; 5 m; and 8 m, respec-
tively. It should be noted that the safety factor of the slope was higher than the critical
FOS of 1.25 even in the case of 8 m pile spacing. However, the pile spacing of 5 m
(i.e., FOS=1.28) was recommended for the slope in this study.

(a) (b)

(c)
Fig. 6. Pile reinforced slope at the toe (a), upper top (b) and middle (c)

4 Conclusions

A 2D numerical study has been used to evaluate the stability condition of the unrein-
forced slope and optimize the pile reinforcement for the actual slope belonging to Khe
816 N. V. Dung et al.

Cham coal preparation construction site. On the basis results of this study, the follow-
ing conclusions can be drawn:
1. The factor of safety (FOS) of the unreinforced slope was found to be 1.21, lower
than the critical FOS of Vietnamese standard for slope stability TCVN 4054
(FOS>1.25). It is recommended that the actual slope should be reinforced for
long-term stability. This finding agreed with the site investigation.
2. The safety factor of the reinforced slope was sensitive to the pile position. The
target FOS could not be reached in the case of the pile at the toe. However, lo-
cating the piles at the upper top and especially at the middle represented a favor-
able condition for the slope. The safety factors of the slope in these cases im-
proved relatively.
3. The safety factor of the reinforced slope significantly depended on the pile spac-
ing, especially when piles are located at the middle of the slope. The safety fac-
tor decreased as the pile spacing increased, except for the case pile at the toe.
Based on the parametric study on pile spacing, the optimal pile spacing was
found to be 5 m to achieve the critical FOS of higher than 1.25.

Acknowledgment

This work was supported by Hanoi University of Mining and Geology and Dong Bac
corp-port company.

References
1. Ito, T., T. Matsui, and W.P. Hong, Design method for stabilizing piles against landslide—
one row of piles. Soils and Foundations, 1981. 21(1): p. 21-37.
2. Poulos, H.G., Design of reinforcing piles to increase slope stability. Canadian Geotech-
nical Journal, 1995. 32(5): p. 808-818.
3. Griffiths, D., H. Lin, and P. Cao, A Comparison of Numerical Algorithms in the Analysis
of Pile Reinforced Slopes. 2010. 175-183.
4. Guo, W.D. and H.Y. Qin, Thrust and bending moment of rigid piles subjected to moving
soil. Canadian Geotechnical Journal, 2010. 47(2): p. 180-196.
5. Yang, S., X. Ren, and J. Zhang, Study on embedded length of piles for slope reinforced
with one row of piles. Journal of Rock Mechanics and Geotechnical Engineering, 2011.
3(2): p. 167-178.
6. TCVN, V.s., Vietnamese standard for slope stability TCVN 4054. 2010, Vietnamese stand-
ards-TCVN.
7. Lee, C.Y., T.S. Hull, and H.G. Poulos, Simplified pile-slope stability analysis. Computers
and Geotechnics, 1995. 17(1): p. 1-16.
8. Nian, T.K., et al., Limit analysis of the stability of slopes reinforced with piles against
landslide in nonhomogeneous and anisotropic soils. Canadian Geotechnical Journal, 2008.
45(8): p. 1092-1103.
 Experimental Correlations for the Swelling Pressure of
Expansive Clays in the City of Tebessa, Algeria

Adel Djellali1, Behrooz Saghafi2, Mohamed Salah Laouar 3

1
Mining institute, Larbi Tebessi University, 12002 Tebessa, Algeria. E-mail:
adel.djellali@univ-tebessa.dz
2
Highway Engineer and Highways Innov8 Champion, Capita Real Estate and Infrastructure,
UK. Email: behrooz.saghafi@capita.co.uk
3
Civil Engineering Department, Larbi Tebessi University, 12002 Tebessa, Algeria. E-mail:
medsalah.laouar@univ-tebessa.dz

Abstract. This paper aims to describe the dependency of the swelling pressure (Ps) of Tebessa
clay soils to the conventional soil properties namely; Plasticity index (PI), dry density (Jd),
initial moisture content (W) and carbonate content (Ca). A statistical model, capable of obtain-
ing an indirect estimation of Ps based on these soil parameters, is provided. High correlation
coefficient (R2) of 0.93 demonstrated predictability of swelling, using multiple regression mod-
eling. Although the above-mentioned parameters play important roles in the swelling behavior
of the clayey soils, there is not any universally accepted, simple and quantitative method to
classify swelling pressure at present. The equation, proposed in this paper, helps the engineers
with evaluating soil pressure in practice. In addition, it can be popularized when environment
conditions are satisfied for application in field explorations and design of structures over ex-
pansive soils.

Keywords: Expansive Clays, Swelling pressure, geotechnical parameters, regression

analysis.

1. Introduction

The soils of Tebessa city in Algeria, North of Africa, are mainly composed by layers
of limestone, gravels and calcareous tufa. However, the middle part of the city pre-
sents extensive clay layers and marly clays with swelling behavior [1]. In this part of
city, the overall types of construction are essentially residential housing and flexible
pavements, that have suffered with cracking and premature loss of serviceability,
caused essentially by the swelling behavior of underlying clays. Swelling of clays
affects the structural stability of the overlying infrastructure due to volume change
associated with seasonal moisture content fluctuation in the underlying soils [2]. Var-
ious remedial measures like soil replacement [3], pre-wetting [4], moisture control [5]
and lime stabilization [6] have been practiced with varying degree of success. How-
ever, these techniques have a certain limitations regarding their adaptability matters.
Rational design of foundations of constructed facilities and pavements structure re-
quires prediction of the swelling characteristics of clays. If soil is suspected to be
expansive during preliminary site investigation, the swelling parameters are estimated
using the correlations available in literature. The empirical expressions relate the

© Springer Nature Singapore Pte Ltd. 2020 817

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_130
818 A. Djellali et al.

swelling parameters to the geotechnical parameters determined by identification tests

[7]. Estimation is sometimes essential due to the difficulties in measuring swelling
directly and sometimes desirable in order to ascertain the achieved results. The corre-
lations are generally either mechanistic-empirical or empirical. Komornik and David
[8], and Vijayvergiya and Ghazzaly [9] developed the earliest prediction equations in
which the swelling pressure was logarithmically related to liquid limit and dry unit
weight. In addition, the swell pressure can be estimated by using soft computing
methods [10], and with procedure based on diffuse double layer theory. Swell percent
can be also predicted from measured values of electrical conductivity and specific
gravity. [11]. This research work intends to evaluate the swelling behavior clayey soil
of Tebessa city. Statistical model is developed based on chemical and physical prop-
erties and correlated to predict the swelling pressure.

Spain Mediterranean Sea

N16
Algiers
Tunisia

Morocco Tebessa

Study N10

Area N82
ALGERIA

Hoggar TEBESSA

0 5 10 km

Tebessa Airport
0 200 400 km National Road

Fig.1. Location of the study area.

2. Material and methods

The problem of swelling soil and its relationship to soil characteristics can be studied
with various approaches. In this experimental work, 40 samples were selected as
representatives of the studied material. Tests were carried out on the core samples
obtained using hydraulic-feed rotary boring machine. The boreholes were made
within certain national roads of the city (N10, N16 and N82) and residential building
to get sound information about their underlying formations, where the active zone
varied from 1 to 5 m. Knowing swelling pressure at such points could help with
redesigning the existing and further structures. The samples extracted from various
depths ± 6 m at the deepest ± revealed that the soils of the area were of marly clays,
clayey silts or marls. The free swell odometer test is used to measure the swelling
pressure according to French standard XP P 94-091[12].The geotechnical properties
of tested samples are summarized in table 1.
 Experimental Correlations for the Swelling Pressure … 819

Table 1. Basic geotechnical and swelling characteristics of the collected material.

Standard
Properties R2(*) Minimum Maximum Mean deviation
Particles <0.08mm (%) -0,110 53,78 97.2 86,89 9,86
Water content w (%) 0,656 11,71 28,55 19,14 3,85
Dry density Jd (Mg/m3) -0,724 1,47 1,93 1,69 0,13
Liquid limit (%) 0,660 42 77 60 9,37
Plastic limit (%) 0,799 15 51 29,82 1,72
Plasticity index (%) 0,38 16 51 30,17 1,27
Methylene blue values
(cm3/g) -0,06 2,08 10,86 5,12 1,99
Carbonate content (%) -0,55 6,56 74,20 41,16 7,10
2
swelling pressure (kN/m ) 1 100 260 175,50 8,89
(*) Correlation coefficient of PS with other geotechnical parameters

The VXEVRLO¶V were classified as A-7-6 [13], where those soils are clays and marly
clays of high plasticity. The classification, based on Dakshanamanthy and Raman
(1973) chart, showed the soils were called inorganic clay with medium to high swell-
ing potential (Fig. 2).

70 Low Intermidiate Hight Very Hight Exteremly Hight

Ligne B
60
Plasticity Index (%)

Ip=0.9(LL-20)
50 Ligne A
Ip=0.73(w-20)
40
30
20
10
0
0 20 40 60 80 100 120
Liquid Limit (%)
Fig. 2. Subsoil classification of Tebessa city, based on Dakshanamurthy and Romana [14].

3. Statistical analysis
3.1. Correlation analysis
The main purpose of this work is to select soil parameters which influence the swell-
ing pressure (Ps) and to estimate (Ps) in a short time. Table 2 shows the correlation
coefficients relating (Ps) to the particles less than 0.08mm (C), dry density ( Jd),
water content (W), liquid limit (LL), plastic limit (PL), plasticity index (PI), Meth-
820 A. Djellali et al.

ylene blue values (MBV) and Carbonate content (Ca) respectively. It observed that
(PL) has a higher linear relationship to (Ps) with a coefficient of determination R2 of
0.799. It is slightly lower than (LL) with R2 = 0.660. These two parameters were
proved previously to have a good correlation with (Ps) [10]. The (MBV) and (Ca)
have a weak and moderate negative linear relationships to (Ps), with R2 of -0.06 and -
0.55 respectively, this negative correlation indicate that the increase of those two
parameters decreases the (Ps). Those parameters were also used by Türköz and Tosun
[15]. The R2 of (W) and (Jd)with (Ps) gives good correlations. Those two parameters
were considerably related to (Ps), or in other words, changes in those geotechnical
parameters create notable change in swelling behavior of Tebessa clay. Al-0H¶DPDU,
concludes similar results, and has explained that samples of higher density (17.8
Mg/m3) allow less water to enter within their structure due to their higher compact-
ness [16]. Therefore, such samples demonstrate a swelling lag at the beginning of
their swelling stage when compared to the samples of lower density and the (Ps) de-
veloped in advantage in further time.

3.2. Regression analysis

The indirect approaches are statistically based on empirical or semi-empirical expres-
sions that relate swell percent or swell pressure to a measured soil parameter such as
Atterberg limits, clay content, activity, linear shrinkage, moisture content, soil suction
or even combination of those [17, 18]. The direct techniques involve measurement of
heave based on a one-dimensional approach, making use of vertical swell values
measured during the wetting of specimens under an appropriate vertical load in an
oedometer [19]. The multiple regression analysis to correlate the measured (Ps) to the
soil properties revealed the following result:

Ps = 535.9739+1.52095 W -198.20894Jd -1.67244 IP -0.20003 Ca+H  

Where H is the mean-zero Gaussian random error term.

The coefficient of determination R2 for Equation (1) is 0.9327 and the corre-
sponding adjusted value is 0.9286. The t-statistics as well as the corresponding p-
values for the equations are shown in Table 2. Each of the variables in the equation is
significant at 5% level (i.e. D = 0.05) as the p-values are less than 0.05.

Table 2. Results of regression analysis relevant to Equation (1).

Variables Coefficient t-statistic p-value
Constant 92558.96 11.99 0.0010
W 831.55 18.10 0.0059
Jd 16386.41 159.56 0.0001
PI 10416.94 101.43 0.0001
Ca 505.84 4.94 0.0030
Experimental Correlations for the Swelling Pressure … 821

The accuracy of the predicted values of (Ps) against the corresponding measured val-
ues is shown in Fig. 3. The figure shows that the predicted values read well to the
measured values which verifies the achieved coefficient of correlation R2=0.93.

300
Measured values of swelling

250 y = 0,9327x + 11,589

R² = 0,9327
200
pressure
(kN/m2)

150

100

50

0
0 50 100 150 200 250 300
Predicted values of swelling pressure (kN/m2)

Fig.3. Measured versus predicted values of (Ps) for Tebessa clays.

4. Conclusions
An experimental investigation on the swelling characteristics of a clayey soil in
Tebessa city, Algeria, has been carried out. The samples derived from clayey deposit
have confirmed the material to be inorganic clay with medium to high swelling pres-
sure. A set of correlation analyses was made between (Ps) and different geotechnical
parameters. Based on data collected from the samples of subVRLO¶V, a simple mathe-
matical equation is suggested to model (Ps) according to significant properties of the
clay, i.e. (W), (Jd), (PI), and (Ca). The proposed equation presented a R2 of 0.93 when
compared to the measured values obtained from testing in laboratory. If (Ps) is
known, the pavement and building structures can be designed in a way that it applies
a surcharge equal to the (Ps) induced by the subsoil, so the structures life service is
prolonged and required less maintenance.

References
1. Djellali, A., Laour, M. S., & Houam, A.: Geotechnical Mapping of Clayey Subgrade Soils
Characteristics: A Case Study from Tebessa City (Algeria). In A. Kallel, Z. A. Erguler, Z.-
D. Cui, Al. Karrech, M. Karakus, P. Kulatilake, & S. K. Shukla (Eds.), Recent Advances
in Geo-Environmental Engineering, Geomechanics and Geotechnics, and Geohazards.
Cham: Springer International Publishing, 247±250 (2019).
2. Puppala, AJ., Manosuthkij, T., Nazarian, S., Hoyos, LR.: Threshold moisture content and
matric suction potentials in expansive clays prior to initiation of cracking in pavements.
Geotechnical Geological Journal, 48 :( 4), 519-531 (2011).
3. Jones, DE., Holtz, WG.: Expansive Soils - the hidden disaster. Civil Engineering ,
ASCE, 43(8), 49-51(1973).
4. Subba Rao, KS., Satyadas, GC.: Prewetted and Overloaded Expansive Soil. Proceedings of
Geotech 80, Bombay, 29-32 (1980).
822 A. Djellali et al.

5. Marienfeld, ML., Baker, TL.: Paving fabric interlayer as a pavement moisture barrier.
Transportation Research Circular E-C006, Transportation Research Board, Washington,
D.C (1999).
6. Thompson, MR., Robnett, QL.: Pressure Injected Lime for Treatment of Swelling Soils.
One of the 4 reports prepared for the 54th Annual meeting of the TRB, TRR-568, 24-34
(1976).
7. Yilmaz, I.: Indirect estimation of the swelling percent and a new classification of soils de-
pending on liquid limit and cation exchange capacity. Elsevier Scientific Publishing, Engi-
neering Geology 85, 295±3 (2006).
8. Komornik, A., David, D.: Prediction of swelling pressure of clays. Journal of Soil Mechan-
ics and Foundation Division, 95(SM1), ASCE, New York, 209±225(1969).
9. Vijayvergiya, VN., Ghazzaly, OI.: Prediction of swelling potential for natural clays. Pro-
ceeding of the 3rd international conference of expansive soils, Vol. 1, Haïfa, 227-
236(1973).
10. Ikizler, SB., Vekli, M., Dogan, E,. Aytekin, M., Kocabas, F.: Prediction of swelling pres-
sures of expansive soils using soft computing methods. Springer Scientific Publishing,
Neural Computing and Applications, (2012) doi : 10.1007/s00521-012-1254-1.
11. Bharat, TV., Sivapullaiah, PV., Allam, MM.: Novel procedure for the estimation of swell-
ing pressures of compacted bentonites based on diffuse double layer theory. Springer Sci-
entific Publishing, Environmental Earth Sciences, 70(1), 303-314 (2013).
12. $)1256ROV5HFRQQDLVVDQFHHWHVVDLV(VVDLGHJRQIOHPHQWjO¶RHGRPqWUH'pWHUPLQa-
tion des déformations par chargement de plusieurs éprouvettes. Normalisation Française
XP P 94-091 (1995).
13. AASHTO.: Standard Specifications for Transportation Materials and Methods of Sampling
and Testing: Part I. Specifications, 14th edn. American Association of State Highway and
Transportation Officials, 444 North Capitol St. N.W., Suite 225, Washington, DC, (1986).
14. Dakshanamurthy, V.; Romana, V.: A simplemethod of identifying an expansive soil. Soils
Found. 1(13), 97±104 (1973)
15. Türköz, M., Tosun, H.: The use of methylene blue test for predicting swell parameters of
natural clay soils. Journal of Scientific Research and Essays, 6(8), 1780-1792 (2011).
16. AL-0H¶DPDU, HF.: The Influence of dry density and applied loads on expansive soils. Ac-
ademic Scientific journals, Engineering and Development, Mustansyriah University, Iraq,
11(3), 120-130 (2007).
17. Seed, HB., Woodward, RJ., Lundgren, R.: Prediction of swelling potential for compacted
clays. Journal of the Soil Mechanics and Foundations Division, 88(3), American Society of
Civil Engineers, 53±87 (1962).
18. Van Der Merwe, DH.: The prediction of heave from the plasticity index and the percent-
age clay fraction of soils. Civil Engineers in South Africa, vol. 6, 103± 107 (1964).
19. McOmber, RM., Thompson, RW.: Verification of Depth of Wetting for Potential Heave
Calculations. Proceedings of Sessions on Unsaturated Soils at Geo-Denver 2000. The Geo-
Institute, ASCE, Denver, Colorado (2000).
 Design of Experiments (DOE) techniques To Predict
Swelling Pressure of Expansive Soils In Tebessa (Algeria)
Yacine Berrah1, Abderrahmane Boumezbeur1, Nouar Charef2 and Serhane Brahmi1
1
University of Tebessa, (Algeria)
2
Mohamed Cherif Messaadia University Souk Ahras (Algeria)
yacine.berrah@univ-tebessa.dz

Abstract. This research work aims to study the swelling behavior of clayey
soils in Tebessa area (Algeria) using the Design of experiments (DOE) method,
based on mechanical, physical and clay mineralogy test results. The effect of
different parameters such as dry unit weight, degree of saturation, water con-
tent, plasticity index, etc. on the swelling behavior of soil is evaluated and the
statistical contribution of each variable in the calculated swelling pressure is al-
so discussed. Besides, relationships between factors affecting the expansion
process have been determined. The swelling pressure generated within the soil,
which is useful in the design of foundations and civil engineering structures is
taken as output process in the screening design methodology. Optimization of
the parameters that affect the swelling behavior by Response Surface Method
(RSM) allows finding the best set of factor levels to establish the mathematical
model. Hence, the efficiency of this model is assessed by comparison of its out-
put results with those obtained from laboratory tests. All of these techniques al-
low choosing the appropriate model that can be used for all soil conditions.

Keywords: Swelling Soil; Design of experiments (DOE); Soil parameters; Re-

sponse Surface Method (RSM).

1 Introduction

Swelling soils cover wide areas in many parts of the world. Many studies have been
conducted on the behavior of expansive soils and concluded that the swelling charac-
teristic is generally governed by physical, mineralogical, mechanical parameters and
geological nature [1, 2, 3]. Enormous studies conducted to predict swelling pressure
and potential as parameters that describe the phenomenon and could be widely used in
design and sustainable of geotechnical infrastructure [4, 5, 6]. Swelling pressure is
defined in many ways and is dependent on the testing procedure [7]. The swelling is a
complicated phenomenon and parametric prediction cannot take into account all of
the effects. Many methods have been developed for estimating the swelling pressure
of clayey soil, which can be classified as direct or indirect methods [8, 9]. Note that
the indirect methods are generally based on tests and experiences. Correlation of the
basic soil mechanical parameters can provide quick and useful identification of swell-
ing soils. Tebessa area (Algeria) is the case study of the present research work, the
geological formation of the studied area are in major part fine deposits with high plas-
ticity; expansive soils are well identified from the different reports and expertise elab-

© Springer Nature Singapore Pte Ltd. 2020 823

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_131
824 Y. Berrah et al.

orated by laboratory and earth works establishment, especially at some profile were
damages are observed in different structures and buildings.
In this research, the concept of design of experiments (DOE) has been introduced to
study the swelling behavior of the clayey soils using about 90 samples identified and
tested in soil mechanics laboratory. It is worth nothing that the DOE method has been
used in many engineering fields, especially chemical and manufacturing engineering.
This method is an extremely powerful approach to experimentation and systematic
method to determine the relationship between factors affecting a process and the out-
put of that process [10, 11]. In other words, it is used to find cause-and-effect rela-
tionships. In this paper, the swelling pressure presents the output parameter as de-
pendent or independent; that affected by several of physical and mechanical parame-
ters.

2 Material and Methods

The general approach to conduct and planning the experiments analysis in Design of
Experiments methods is called the strategy of experimentation [11, 12]. This method
consists of selecting a baseline level for each factor, and then successively varying
each factor over its range with other factors held constant at the baseline level, it is
limited to analyzing the joint effect of factors on the response. The details of the
method can be found in any standard textbook on Design and Analysis of Experi-
ments such as the one by [13]. In this strategy, experiments are conducted by simulta-
neously varying six factors over two levels (namely low level and high level) see
Table 1. This case study presents an example of using the response surface for the
modeling of the swelling pressure ܲ௦ ሺƒሻ. For the presented implementation of DOE
technique, Design-Expert 10 software was employed to obtain the appropriate func-
tional equations. Among conventional DOE techniques, the response surface method-
ology RSM is widely used. Experiments based on RSM technique relate to the deter-
mination of response surface based on the following general equation:
࢟ ൌ ࡭૙ ൅ ࡭૚ ࢞૚ ൅ ‫ ڮ‬൅ ࡭࢑ ࢞࢑ ൅ ࡭૚૛ ࢞૚ ࢞૛ ൅ ࡭૚૜ ࢞૚ ࢞૜ ൅ ࡭૚૚ ࢞૛૚ ൅ ࡭࢑࢑ ࢞૛࢑ (1)

Where଴ , ୧ ,  ୧୨ are regression coefficients for intercept, linear, quadratic and inter-
action coefficients, respectively, and š୧ are independent input variables. RSM produc-
es an empirical polynomial model, which gives an approximation of the true response
surface over a factor region.

Definition of the Input Variables and the Output Responses

In the case study, the effects of six parameters namely; dry unit weight઻‫ ܌‬ሺ‫ۼܓ‬Ȁ‫ܕ‬૜ ሻ,
water content‫ܟ‬ሺΨሻ, plasticity index ۷‫ ܘ‬ሺΨሻ Saturation degree‫ܚ܁‬ሺΨሻ, the
preconsolidation pressure‫ ܋۾‬ሺ‫܉۾ܓ‬ሻ, the swelling coefficient ሺ۱‫ ܛ‬ሻon the swelling pres-
sure‫ ܛ۾‬ሺ‫܉۾ܓ‬ሻ which defines as the output response parameter. The first step of the
present case study is the tabulation of the levels for each factor (see Table 1).
Design of Experiments (DOE) techniques … 825

Table 1. Factors for response surface study

Factor Unit /RZOHYHOí High level (+1)
Ȗd ݇ܰȀ݉͵ 13.5 18.5
w % 12.71 28.55
Sr % 65 98.18
Ip % 20 56
Pc ݇ܲܽ 21.5 232
Cs - 0.005 0.307

3 Analysis of Results and Diagnostics of the Statistical

Properties of the Model
ANOVA is commonly used to summarize the test for significance of the regression
model and test for significance on individual model coefficients. In this case, coeffi-
cient of determination, ³$GMXVWHG 5-6TXDUHG´ DQG ³3UHGLFWHG 5-squarHG´ YDOXHV DUH
KLJKHU IRU ³/LQHDU´ PRGHO 7KLV PRGHO LV VXJJHVWHG IRU the actual analysis, and the
analysis of the experimental data was performed to identify statistical significance of
the parameters dry unit weight, degree of saturation, water content, plasticity index,
preconsolidation pressure and the swelling index on the measured swelling pressure
Ps. The model was developed for 95 % confidence level.

Table 2. Model Summary Statistics

Adjusted Predicted R-
Source Std. Dev. R-Squared R-Squared Squared PRESS
Linear 129.3788753 0.96465741 0.9279179 0.9038569 1448396.26 Suggested
2FI 133.1528248 0.34523123 0.11217794 -1.23979088 3578263.3
Quadratic 131.5451303 0.42593562 0.13348773 -1.69740331 4309339.48
Cubic + Aliased

Table .2 shows the regression statistics and the possible models proposed for the pre-
sent case study, the coefficient of determination is high and close to 1, namely R-
Squared equals to 0.96, which is desirable. The Predicted R-squared (0.92) is in rea-
sonable agreement with the Adjusted R-Squared (0.9). Hence, the suggested model as
shown in Table .2 is Linear. Other diagnostic plots may provide interesting infor-
mation in some situations. The residuals are examined using the normal probability
plots of the residuals and the plot of the residuals versus the predicted response. Re-
siduals versus predicted response should be randomly scattered without pattern as
shown in Fig.1 (a). Normal plot of residuals, shown in Fig.1 (b), should be in a
straight line. The residuals generally fall on a straight line implying that errors are
distributed normally. Nonlinear patterns, such as an S-shaped curve, indicate non-
normality in the error term, which may be corrected by a transformation.
826 Y. Berrah et al.

Residuals vs. Predicted

Design-Expert® Software Normal Plot of Residuals
R1
3.00
Color points by value of
R1: 99
Internally Studentized Residuals 670
95
1.50 60
90

Normal % Probability
80
70

0.00 50

30
20

10
-1.50
5

(a) 1
(b)
-3.00

54.86 204.91 354.96 505.02 655.07 -2.42 -1.26 -0.10 1.06 2.22

Predicted Internally Studentized Residuals

Fig. 1. (a) Residuals versus predicted response for swelling pressure
(b) Normal probability plot of residuals for swelling pressure

Fig. 2. Predicted response versus actual for swelling pressure

In order to determine the quality of the adopted model, it needs to be checked whether
points of predicted response versus actual values are randomly scattered along the 45°
line like in Fig. 2.

Final Equations and Models Graphs

For the analyzed example, the final equation in terms of actual factors was deter-
mined, which present the swelling pressure Ps from the input factors:
ࡼ࢙ ൌ ʹ͸ͲͶǤ͸ െ ͳͲͶͳʹǤ͵ ‫  כ‬െ ͳͳͶͶͲ ‫  כ‬൅ ͶͲ͵Ͷ ‫  כ‬െ ʹͷ͸ͺǤ͹ ‫  כ‬൅ ͳͶʹͲǤ͵ ‫  כ‬൅ ͳ͵͵Ǥ͸ ‫ כ‬ (2)

Figures 3 (a) and (b) show the response surfaces describing the swelling pressure Ps
dependence on the Dry unit weight (kN/m3), the plasticity index (%) and the degree
of saturation Sr (%) respectively. The Plasticity index (Ip), water content (w) and the
Design of Experiments (DOE) techniques … 827

preconsolidation pressure are also combined and correlated but with different signifi-
cance and are not discussed in the present case study.
Design-Expert® Software
Design-Expert® Software
R1
670 R1
670
60 25000 60 24000
X1 = A: A
X1 = A: A
X2 = C: C
Swelling pressure

15500
X2 = D: D

Swelling pressure
10750
Actual Factors Actual Factors
B: B = 20.63 6000
B: B = 20.63 -2500
D: D = 38.00 C: C = 81.59
E: E = 143.25 E: E = 143.25
-3500 -15750
F: G = 0.16 F: G = 0.16

-13000 -29000

(a) 91.00
84.50 (b)
56.00
47.00
1.89 1.89
78.00 1.75 38.00 1.75
Saturation degree71.50 1.62 Plasticity index 29.00 1.62
1.49 1.49
65.00 1.35 20.00 1.35
Dry unit weight Dry unit weight

Fig. 3. (a) Response surface 3D representing the Swelling pressure dependence on the Dry unit
weight (kN/m3) and the plasticity index (%)
(b) Response surface 3D representing the Swelling pressure dependence on the Dry unit
weight (kN/m3) and the degree of saturation (%)

Figures 4 (a) and (b) represent respectively the affecting factors for the swelling pres-
sure as the plasticity index, the saturation degree and the preconsolidation pressure.
Design-Expert® Software
Design-Expert® Software
R1
670 R1
670
60 7700 60 6100
X1 = C: C
X1 = C: C
X2 = E: E
Swelling pressure

5750 X2 = D: D
Swelling pressure

4225
Actual Factors Actual Factors
A: A = 1.62 3800
A: A = 1.62 2350
B: B = 20.63 B: B = 20.63
D: D = 38.00 E: E = 143.25
1850 475
F: G = 0.16 F: G = 0.16

-100 -1400

(a) 265.00
204.13 (b)
56.00
47.00
91.00 91.00
143.25 84.50 38.00 84.50
Preconsolidation pressure
82.38 78.00 Plasticity index 29.00 78.00
71.50 71.50
21.50 65.00 20.00 65.00
Saturation degree Saturation degree

Fig. 4. (a) Response surface 3D representing the Swelling pressure dependence on the Plasticty
index and the Saturation degree (%)
(b) Response surface 3D representing the Swelling pressure dependence on the
Preconsolidation pressure and the Saturation degree (%)

As presented by these figures of 3D surface it can be similarly optimizing the effect

parameters on the swelling pressure. Furthermore, based on the data from multifactor
RSM it is possible to obtain the numerical optimization of the process, i.e. the opti-
mum swell pressure conditions. In the Design-Expert setting it is allowed for all vari-
ables, including factors and propagation of error. The program restricts factor ranges
to factorial levels, plus one to minus one in coded values, the region for which this
experimental design provides the most precise predictions.
 828 Y. Berrah et al.

4 Conclusion
In the present work, all the tests were randomly realized at least twice (for repeatabil-
ity and reproducibility), to reduce the influence of the not assigned variables and the
randomness of responses. The present experimentation considered the potential ef-
fects of six parameters (dry unit weight, degree of saturation, water content, plasticity
index, preconsolidation pressure and the swelling index) on the output factor or the
predicted swelling pressure Ps. It was revealed that dry unit weight, plasticity index,
degree of saturation and the preconsolidation pressure have the greatest effect on the
swelling pressure of the clayey soil in Tebessa area. The effect of the swelling index
is considered non-significant compared to the other parameters. Based on the research
work presented in this paper, it is concluded that the application of one or more meth-
ods in the DOE planning and analysis are necessary and useful since they can help to
reduce the costs as well as the time needed for conducting the experiments.

References
1. Komornik, A., Wiseman, G., and Ben-<DDFRE < ³6WXGLHV RI ,Q-Situ moisture and Swelling Potential
3URILOHV´ Proceedings, 2nd International Research and Engineering Conference on Expansive Soils,
Texas A&M Univ. Press, College Station, TX, pp. 348±361. (1969).
2. )UHGOXQG ' * +DVDQ - 8 DQG )LOVRQ + / ³7KH 3UHGLFWLRQ RI 7RWDO +HDYH´ Proceedings of 4th
International Conference on Expansive Soils´$6&(DQG ,QWHUQDWLRQDO6RFLHW\ IRU6RLO0HFKDQLFVDQG
Foundation Engineering, Denver, pp. 1±17. , (1980).
3. Berrah, Y., Boumezbeur, A., Kherici, N.: Application of dimensional analysis and regression tools to
estimate swell pressure of expansive soil in Tebessa (Algeria). Bull. Eng. Geol. Environ. (2016).
4. 5DR60³,GHQWLILFDWLRQDQG&ODVVLILFDWLRQRI([SDQVLYH6RLOV´Exapnsive Soils²Recent Advances in
Characterization and Treatment, Taylor & Francis, pp. 15±24. (2006).
5. Jennings, J. E. B. and Knight, . ³7KH 3UHGLFWLRQ RI 7RWDO +HDYH IURP WKH 'RXEOH 2HGRPHWHU 7HVW´
Transact. S. African Inst. Civil Eng., Vol. 7, pp. 285±291. (1957).
6. Yilmaz I. Indirect estimation of the swelling percent and a new classification of soils depending on liquid
limit and cation exchange capacity. Engineering Geology 85, 295-301. (2006).
7. ASTM Standard D4546-³6WDQGDUG7HVW0HWKRGVIRU2QH-Dimensional Swell or Settlement Potential
RI &RKHVLYH 6RLOV´ Annual Book of ASTM Standards, Vol. 4, No. 8, ASTM International, West Con-
shohocken, PA, pp. 696±702, (1995).
8. Magnan, D., 1993. Caractérisation in situ des sols gonflants  O¶HVVDL ([SDQVRO Thèse de doctorat,
Université J. Fourier, Grenoble (France), p 190.
9. Vijayvergiya V.N., Ghazzaly, G., 1973. Prediction of swelling potential for natural clays. Proceedings,
3rd International Conference on Expansive Soil, Haifa, vol. I, pp. 227-236.
10. Kralik. J. J., A RSM approximation in probabilistic nonlinear analysis of fire resistance of technology
support structures, Advanced Materials Research, Volume 969, (2014).
11. Haldar A., Mahadevan. S., Probability, reliability and Statistical Methods in Engineering Design. John
Viley&Sons. New York, (2000).
12. Makadia AJ, Nanavati JI Optimisation of machining parameters for turning operations based on response
surface methodology. Measurement 46:1521±1529. doi:10.1016/j. measurement.2012.11.026, (2013).
13. Montgomery, D.C.: Introduction to statistical quality control, 7th edn. Wiley, New York, USA (2013)
 Analysis of Landslides In The Region Of Souk Ahras
(Zaarouria Sector) North- east Of Algeria Using
pseudo-static method
Nouar Charef1, Yacine Berrah2 and Abderrahmane Boumezbeur3
1
University of Annaba, Algeria.
2,3
University of Tebessa, Algeria
charefnouar@yahoo.fr

Abstract. The increasing development in the region of Souk Ahras (Algeria) is more
and more threatened by several landslides, i.e. mass movements of materials. Sustaina-
bility of structures and in particular roads are threatened by this widespread phenome-
non in the region. Landslides cause significant damage and casualties to people and
property. Practical observations of the lands in movement can establish a correlation
between geological, hydrogeological conditions and the landslide phenomenon. In this
study, the main objective is to combine geology with soil mechanics in order to study
causes of landslides in three important sectors (Mechroha, Zaarouria, Hammam Tassa)
all over the wilaya of Souk Ahras territory. In addition, we study particular landslide in
sector of (Zaarouria) by modeling the mechanical behavior of the field using numerical
calculations and discuss the influence of pore water pressure variation on the safety fac-
tor. Finally, the resulting model will experience the use of pseudo- static method to per-
ceive the influence of seismicity on safety factor.

Keywords: Modeling; Pseudo-static method; Pore water pressure; Cohesion;

Safety factor.

1 Introduction
This study examines the moving areas in the wilaya of SoukAhras in the NE Algeria.
The landslide problems in Souk Ahras region are only discovered when the event is
declared in view of traffic jam since the majority of landslides affect the roadways (in
national roads and paths of Wilaya). Then the public works department of the state
intervenes punctually to provide immediate solutions and generally put the road back
into operation with neither necessary investigation nor studies that can help to under-
stand the possible causes and find suitable solutions for the ground movements.
Moreover, it is also important that these solutions should also include at long term a
concept of methodology of regional treatment that addresses potential landslides and
prevents their future retrogression.
The choice of the three studied areas (Mechroha, Zaarouria, Hammam Tassa) is
based on importance of landslide event in the region as well as the analogy of the
hydroclimatological, geological and mechanical conditions. These informations can
help to better discover the probable causes of the landslides and to foresee a numeri-
cal model which can be generalized with precision of the parameters intervening in
such phenomenon.
Generally, changing groundwater levels are one of the main factors that triggers all
inventoried landslides. This observation has motivated us to focus in particular on the
influence of the pore water pressure change on the stability (safety factor) of slopes.
© Springer Nature Singapore Pte Ltd. 2020 829
C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_132
830 N. Charef et al.

Mapping this influence can predict the water level of the subsurface aquifer that can
trigger a landslide. This method applied to the selected sectors gave very similar re-
sults and has been presented in this study only that of the Zaarouria model at
landsliding point (pk102+ 000).
According to CRAAG (Research Center in Astrophysical Astrophysics and Geo-
physics) [1] seismicity in northern Algeria occurs permanently. The surveillance net-
work records almost 50 tremors a month. Nearly 90% of this activity is of low magni-
tude and occurs far from urban centers. This scientific truth has guided us towards the
study of landslide considering the earthquakes with the application of the pseudo-
static method on Zaarouria model to predict the seismic influence on the stability of
the terrain.

2 Study area and data used

2.1 Geological Setting

According to L.David [2] , the lithostratigraphy of SoukAhras area consists on sedi-

mentary formations of which the oldest age is Triassic. In lithology, we can find lime-
stone, clays, marls, sandstones, gravels and alluvium.
The region of SoukAhras is exposed to various types of landslide hazards. Alt-
hough all types of materials can be affected, whether unconsolidated (clay, marl,
sand, etc..) or consolidated, landslides are most often occurring in the unconsolidated
deposits and particularly in the marl, clay and shallow clay marl.

2.2 Hydrological and Hydrogeological Conditions

Landslides in the study area were accompanied by dripping water sources and sewage
waters downstream and upstream slope. The waters are the driving force movement of
land. Surface waters act by their mechanical and chemical action and are responsible
of surface erosion [3]. Those waters, when they infiltrate, change the hydrogeological
conditions by humidifying soil and increasing the pore pressure; this latter, in fact, is
the cause of outbreak of landslides consecutive to intense and continuous rainfall [4].

2.3 Types of Landslides

The dominant type of sliding in the region of Souk Ahras is Rotational Rock Slump.
It is very similar to the rotational landslide in its circular failure surface, but the verti-
cal section of the sliding surface is irregular [5] as shown in Fig.1. This slump is not
deep (<20m) and occurs in marls and clays.
Several factors may be responsible for these landslides:
- The moving or creating a road embankment (anthropic action)
- The slope (gravity action)
- Geological nature of soil (particularly swelling-shrinkage of the clay)
- Rain and exceptional winter rainstorms (increased interstitial pressure).
- Crue torrent and rapid river flood (surface erosion)
- Infiltration and drainage in the soil (decrease of soil strength).
Analysis of Landslides In The Region Of Souk … 831

- Intense temperature change (thermoclastie, gelifraction).

Fig.1 Overview of landsliding slope, PK102+ 000 - Zaarouria

2.4 Soil Characterization

Laboratory tests comprise: particle size analysis, dry density, natural water content,
Atterberg limits, straight shear tests, oedometer tests
In situ tests are presented by the dynamic penetrometer surveys along a main pro-
file located in the axis of sliding. Depending on the granulometry and Atterberg lim-
its, the analyze results of soil [6] show that the fine grains of unstable terrain are com-
posed by inorganic clays of illite dominance. The plasticity of these clays is generally
medium to high for some soils; Wl varies between 20% and 68%, while Ip varies
between 09% and 39%.

3 Methodology

3.1 Geotechnical Conditions at Zaarouria Sector

Definition of the geometric model

Definition of the fracture surface

Definition of mechanical parameters

Definition of the piezometric surface (H) Definition of

piezometric
level
Definition of safety factor F
Iteration

Back analysis

Fig. 2 Methodology used to calculate the safety factor.

832 N. Charef et al.

In the sector Zaarouria, Soil is named plastic marly clay with illite having cohesion of
0.25  C  0.40 bars and angle of friction 20° M  35° [6].
The finite element method used by the Geo-slope software (version 2004) is
adopted to treat the problem of landslides. The algorithm for calculating the safe-
ty factor is mentioned in Fig.2 that shows the different steps in calculating the
safety factor for a given profile with a direct method (classic) and reverse method
(Back-Analysis). The first iteration concerns, the variations of piezometric levels
and the second by the inverse method concerns variations of the piezometric level
and friction angle M.

4 Results and Discussion

4.1 Numerical Model of Sliding in Zaarouria (PK102+000)

In a classical approach, as listed in Table1, we deduce that the slope in natural
conditions (assuming a groundwater level at 3m depth), is in a state of stability
admissible (F t 1.3) by the method of Bishop. But by Janbu method, it is in unac-
ceptable state approaching the critical (unstable) condition )”
Table 1 Calculation results of F in terrain conditions.

Safety factor F Cohésion Friction Depth of water Hwater

Janbu Bishop C (KPa) Angle M (°) (m)
20 15
1.143 1.340 3
50 30

4.2 Evaluation of Stability with Variation of Piezometric Level

As indicated in Table 2, when the field is considered with no water, the slope is
stable. This stability becomes ineligible for a groundwater level at depth of 3 m.
For a completely water saturation of soil (case of intensive rainfall in winter) the
factor of safety F < 1 indicates an unstable state of the slope,

Table 2 Results of calculation of F for different depths of groundwater level (Zaarouria).

Safety factor
F C Friction angle M Depth of water
(KPa) (°) (m)
Janbu Bishop
without water-
1.387 1.616 20-50 15-30
table
1.247 1.316 20-50 15-30 3
0.895 1.032 20 - 50 15 - 30 0
Analysis of Landslides In The Region Of Souk … 833

4.3 Evaluation of Stability by Back Analysis

Assuming the groundwater level at 3 m and drawn in Fig.3, the use of Géoslope/W
program allows to declare a critical state of stability (F = 1) with the following ge-
otechnical characteristics: C = 15.5 kPa and M = 15 °
26

24 1.006

22

20

18

16

14

12

10

0
0 10 20 30 40 50

Fig. 3 critical slip surface considering &DQGijRIWKHVWXGLHGVRLO.

Mapping the variation of the safety factor is shown in Fig. 4. It can be seen from
the figure that for cohesionless soils(c = 0 kPa in rupture), slope failure can occur if
the soil has a friction angle of 23 ° and a groundwater level at 3m depth; if Hwater
exceed 3 m, the stability condition remains precarious with a safety factor of less than
the admissible (F < 1.3). Variation of F as a function of M (°) is presented in Table 3.

Table 3 Stability conditions at Hwater = 7 m and C = 0 by the method of Janbu (Zaarouria).

M (°) M d 20 ° 20 °< M d 25.5 ° M > 25.5 °
F Fd1 1 < F d 1.3 F > 1.3
stability state unstable critical stable

Fig. 4 Mapping the variation of F (Janbu) based and the piezometric level and the angle of
friction M for a cohesion (C = 0 kPa).
4.4 Influence of seismicity

SLOPE/W has the option to consider an undrained response when doing a pseudo-
static analysis. The values of the pseudo-static coefficient vary in significant ways
according to the authors and there is no rational approach to define this value. It is
defined either as a fixed value often close to 0.1 or 0.15 depending on the magnitude
of the earthquake or sometimes as a fraction of the maximum acceleration at the rock
or at the surface of the ground. This fraction varies typically between 0.33 and 0.75.
834 N. Charef et al.

[7] However, the choice of coefficients used in the slope stability analysis is very
subjective and lacks a clear rationale[8].
According to Algerian seismic standards in the Souk Ahras region, the horizontal
coefficients of acceleration load can vary between 0.16 and 0.2. Results of calculation
of safety factor (F) with pseudo-static analysis are shown in table 4.

Table 4 Results of calculation of F with pseudo-static analysis (Zaarouria).

F (h.coef = 0.16) F (h.coef = 0;2) F (Static analysis) C (KPa) M (°) Water depth (m)
Janbu Bishop Janbu Bishop Janbu Bishop
0.820 0.889 0.755 0.821 1.247 1.316 20-50 15-30 3
For an acceleration load even low, the safety factor is less than 1, hence the slope is
unstable.

5 Conclusion
This paper discusses a typology of landslides in the wilaya of Souk Ahras. In particu-
lar, the research study focuses on rotational landsliding that are most prevalent espe-
cially in the center and north of the Wilaya.
The stability calculation has highlighted the important influence of several parame-
ters and more particularly the piezometric level converted to a pore pressure.
For the considered Zaarouria slope, it is found that the stability becomes inadmissible
for a groundwater level at 3 m depth. If the groundwater level is below 7 m, it does
not give any influence on the variation of the safety factor. In fact, the slope stability
depends only on the friction angle. In the case of a friction soil, non-cohesive and
having a piezometric level Hwater t 7 m, stability is achieved only for values of M >
25.5 °. Cohesion plays the stabilizing role because the friction angle alone offers no
margin of safety even at considerable values.
The influence of seismicity is extremely serious on the stability of sloping ground
even for low values of acceleration load.
References
1. CRAAG homepage https://www.craag.dz/r_sismologique.php, last accessed 2019/04/30
2. L.David, Geological study of mountains of Medjerda, Thesis Sci. Paris. Publ. Serv. Geol.
Map Algeria Bull. 11 (1956) 189. (in French)
3. T.Lebourg, Geological and mechanical analysis of landslides in the moraines of central and
western Pyrenees, Ph.D. thesis, France, 2000, p. 361. (in French)
4. P. Aleotti, A warning system for rainfall-induced shallow failure, San Giuliano Milanese,
Italy, 2004
5. G. Colas, G. Pilot, Description and classification of landslides, Bull. Liaison Lab. P. et Ch.
special (1976) 21-30. (in French)
6. Laboratory reports of soil analysis in Zaarouria sector at kilometric point (pk102+000),
L.T.P.E Souk Ahras. (in French)
7. Laboratory reports of soil analysis in Zaarouria sector at kilometric point (pk102+000),
L.T.P.E Souk Ahras. (in French).
8. Cristiano MELO and Sunil SHARMA, Seismic coefficients for pseudostatic slope analysis.
13th World Conference on Earthquake Engineering, Vancouver, B.C., Canada. August 1-6,
2004. Paper No. 369.
 Solving the stability problem of vertical slope
according to the effective stress field

Do Thang

Thuyloi University, 175 Tay Son, Dong Da, Hanoi, Vietnam

dothang@tlu.edu.vn

Abstract. Soil mechanical behavior does not agree with the elastic theory, but
complies with Mohr-&RXORPE \LHOG FULWHULRQ DQG 7HU]DJKL¶V Hơective stress
principle. The deformation of soil as well as its shear strength depends on the
effective stress. In this paper, a new direct method has been developed to de-
termine the effective stress field in soil based on the shear potential. The meth-
od allows solving the stability problem of vertical slope.

Keywords: effective stress field, shear potential, stability problem, vertical

slope.

1 Effective stress field in soil

7HU]DJKL¶VHơective stress principle is an important principle in soil mechanics. The

effective stress is equal to total stress minus pore water pressure [1]. The deformation
of soil as well as its shear strength depend on the effective stress. According to Ar-
nold Verruijt [2], the stresses in the particles are generated partly by the concentrated
forces acting in the contact points between the particles, and partly by the pressure in
the water, that almost completely surrounds the particles. The pressure in the water
results in an equal pressure in all the grains, so the deformations of the particle skele-
ton are almost completely determined by the concentrated forces in the contact points,
because the structure can deform only by sliding and rolling in these contact points.
Clearly, the deformation potential of the soil is due to the sliding and the rolling of the
particles. Therefore, the effective stress field in soil is based on the shear potential.
To distinguish effective stress field in soil based on the shear potential with elastic
theory, we first study the elastic stress field in the soil.
If the soil is considered elastic material, the elastic stress field in the soil can be de-
termined by displacement field and strain. In the plane strain problem, using stress is
unknown, the stress field can be determined by the extremum problem is [3]:

© Springer Nature Singapore Pte Ltd. 2020 835

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_133
836 D. Thang

2 2
1 ªV x  V y W xy2  W yx2 º ½
Z ³V E « 2
«  Q .V x .V y  (1  Q ) » dV o min °
¬ 2 ¼» °
°
wV x wW yx °
 0 ¾ (1)
wx wy °
wV y wW xy °
 0 °
wy wx °
¿

where:
Z is the elastic deformation potential in the plane strain problem;
Vx, Vy, Wxy, Wyx is the stress state at a point in the soil;
E, Q is the elastic modulus and Poisson's ratio of soil.
In Eqs. (1), we do not consider the self weight. V is the area of the consideration do-
main. By variational calculus, Eqs. (1) become the system of three equations of the
elastic theory as [3]:

­
°

°’ 2 V  V
x y  0
°
° wV x wW yx
®  0 (2)
° wx wy
° wV y wW xy
°  0
°̄ wy wx

where: ’2 is a symbol of the Laplace operator,

w2 w2
’2 2
 2
wy wx

Clearly, Eqs. (1) are the problem of determining elastic stress field in soil. When
using the stress is unknown, the stress field can be determined according to the ex-
treme potential or Eqs. (2).
The shear potential is equal to total potential energy (elastic deformation potential Z)
minus volume deformation potential [4]. In the plane strain problem, the shear poten-
tial is expressed by Eq. 3:

1 ª 2
   2W xy2 º
2
Zs Z  Z0
«
¬
V x  V m   V y  V m ¼»
(3)
4G

Vx V y
where: V m
2
Substituting this value into Eq. (3), we have:
Solving the stability problem of vertical slope … 837

2 2 2
1 ª§ Vx V y · § V x V y · 2
º 1 ª§ V x  V y · 2
º
Zs «¨ V x  ¸  ¨V y  ¸  2W xy » «¨ ¸  W xy » (4)
4G «© 2 ¹ © 2 ¹ »¼ 2G «© 2 ¹ »¼
¬ ¬

Thus, the plane strain problem to determine effective stress field in soil according to
the shear potential is:

1 ª§ V x  V y ·
2
º ½
« W 2
» °
Zs ³ 2G «© 2 ¹ xy » dV o min
¨ ¸ 
°
V
¬ ¼ °
wV x W
w yx °
°
 0 ¾ (5)
wx wy °
wV y wW xy °
 J 0 °
wy wx °
°
¿

where: J is the volumetric weight of the soil.

To demonstrate the problem (5) is determinacy, using the variational calculus similar
to problem (1), Eqs. (5) become the system of three equations [3]:

­
°

°’ 2 V  V
x y 0
°
° wV x wW yx
®  0 (6)
° wx wy
° wV y wW xy
°  J 0
°̄ wy wx

where: ’2 is a symbol of the Laplace operator,

w2 w2
’2 2
 2
wy wx

Eqs. (6) have three equations for finding unknown as: Vx, Vy and Wxy. Thus, the
problem of determining effective stress field in soil (5) is solved.
Comparing the problem (2) determining the elastic stress field in the soil with the
problem (6) determining the effective stress field in the soil according to the shear
potential, we see the difference is sign "±" of the first equation.
Now, we have the effective stress field in the soil as the deterministic static field,
enough equation to solve. So, problems in soil mechanics are deterministic, that can
be used to solve different stress state problems.
To clarify the difference of the stress field in soil based on the shear potential with
elastic theory, the author solves the problem of determining the stress field due to a
uniform strip load on a homogeneous half space with calculating data: The soil get:
838 D. Thang

elastic modulus E = 30000kPa, Poisson coefficient Q = 0.3. The uniform load of mag-
nitude p=30kPa on a strip of width 2a=10m on a homogeneous half space.
The problem is realized by a code Matlab [3].
The results of calculating the vertical stress Vy by the finite difference method are
shown in Fig. 1 (according to the elastic theory) & Fig. 2 (according to the shear po-
tential).

Fig. 1. ıy (kPa) according to the elastic theory Fig. 2. ıy (kPa) according to the shear potential

The Fig. 1 and Fig. 2 show the distribution of vertical stresses with respect to hori-
zontal direction whose depth based on the elastic material is broader and deeper than
the case based on the shear potential.
The soil is only deformed at a certain point depth when applying external load.
Thus, according to the current standards, the settlement zone depth influenced by the
stress of the external loads, is usually equals 0.1 to 0.2 times the one by the stress of
the self weight. Clearly, the effective stress field according to the shear potential is
more suited to the nature of the soil than in the case considered by the elastic material.

2 Analysis of stability of vertical slope

The problem of vertical slope is shown in Fig. 3.

O x n1 n0

'y
y H 1 J1
c1,M
'x
(b) m1


0 J0
c0,M


m2

(a)


Fig. 3. Diagram to analysis the stability problem of vertical slope

Solving the stability problem of vertical slope … 839

Solving this problem by analytical method is very difficult, especially when con-
sidering the self weight. Therefore, the authors solve the problem by finite difference
method.
The soil mass is diviced into the differential grid as shown in Fig. 3a. At each
QRGHWKHUHDUHWKUHHXQNQRZQVıxıyIJxy. Splitting a rectangular cell from the dif-
ferential grid (Fig. 3EWKHKRUL]RQWDOGLPHQVLRQLVǻ[DQGWKHYHUWLFDOGLPHQVLRQLV
ǻ\ )L[HG ǻ[ JLYLQJ ǻ\ LQFUHDVHG WKH KHLJKW RI YHUWLFDO Vlope H = (m1-ǻ\ ZLOO
increase. When the height of the slope reached the value that the soil mass began to
form destructive mechanism is called the critical height. Hence, the height of the
slope H is unknown of the problem. This is a new way. Because of the common way
is reduce the shear strength of the soil until the soil is failure by dividing the stability
factor K or reducing the elastic modulus E to slope horizontal displacement increases
to a limit value, ie the critical height is not directly determined.
By the limit analysis method, the stress state is determined by the lower limit theo-
rem and the upper limit theorem. However, the author only need to use the lower limit
theorem combined with the condition that all points are able to yield [3].
The objective function of vertical slope stability problems due to self-weight is writ-
ten as follows:
2
ª 2 º
1 « § V x V y · 2
Vx V y » dV 
Z1 ³ ¨ ¸  W xy  sin M  c.cos M
V
2G « © 2 ¹ 2 »
¬ ¼ (7)
2
1 ª§ V x V y · 2
º
³ «¨ ¸  W xy » dV  H o min
2G «© 2 ¹ »¼
V
¬

The objective function (7) must satisfy two equilibrium equations and the con-
straints, as follows:
- The soil is not able to withstand pulling;
- Mohr- Coulomb yield criterion;
- Boundary conditions on horizontal, vertical slope and infinity.
Calculated data: The physical properties of soil are as follows: Soil with the cohe-
sion c1 = c0 = c and for change; the angle of internal friction M1 = M0 = M and for
change; the volumetric weight of the soil J1 = J0 = J=17kN/m3. Difference grid size
'x = 'y = 0.1m.
The problem is realized by a code Matlab [3].
The results of calculating the limited height by the finite difference method are shown
in Fig. 4.
The critical height of vertical slope always as:
2.3c
H cr tg (450  M / 2)
J
From Fig. 4, we find that with respect to the critical height, we see that the yield
points begin (zero values) to appear from the foot of vertical slope and develops to-
wards interior.
840 D. Thang

Fig. 4. Chart of contours yield ability

In case of perfectly cohesive soil (the internal friction angle is zero), the critical
2c
height of the vertical slope is within the lower limit Hcrlower and upper limit
J
4c
Hcrupper [5].
J

3 Conclusions

In this paper, the problem of soil stability has been solved using the effective stress
field of the soil based on the shear potential.
A new direct method of determining the critical height Hcr of the vertical slope is
totally different from the indirect method which is used to determine the limited
height through the stability factor K or the limit displacement.
The obtained results proved the reliability of the new method because the critical
height of the vertical slope in the case of perfectly cohesive soil is within the lower
limit and upper limit of Limit analysis theory.

References
1. K. Terzaghi: Theoretical soil mechanics, Wiley, New York (1943).
2. A. Verruijt: Soil mechanics, Delft University of Technology, Delft (2010).
3. '7KDQJ'RFWRUWKHVLV³5HVHDUFKRIHDUWKHPEDQNPHQWVWDELOLW\RQQDWXUDOJURXQG´8Qi-
versity of Transport and Communications, Hanoi (2014).
4. N.I. Bezukhop: Fundametantals of elasticity, plasticity and creep theory, Vysshaya Shkola,
Moscow (1968).
5. W. F. Chen: Limit analysis and soil plasticity, J.Ross Publishing edition is an unabridged
(2008).
 Finite Element Simulation of Water Content-Influenced
Progressive Failure of Sensitive Clays

Nhat-Phi Doan1, Seung-Wook Woo1, Yao-Long Hou1 and Sung-Sik Park2

1
School of Architectural, Civil, Environmental, and Energy Engineering, Kyungpook National
University, Daegu, 41566, South Korea
phidoan.ksxd@gmail.com
2 Department of Civil Engineering, Kyungpook National University, Daegu, 41566, South

Korea
sungpark@knu.ac.kr

Abstract. Large-scaled landslides often occur in sensitive clay deposits in

horsts and grabens, typical of spreads. Most large deformation numerical analy-
sis based on the strain-softening behavior of undrained shear strength. In prac-
tice, water infiltration into soils may result a variation of non-uniform crack and
shear band emergence at a low soil resistance. In this study, the mechanism of
water content effect on the stress-strain relationship was implemented in Mohr-
Coulomb (M-C) model using the user material subroutine of the commercial
software Abaqus/Explicit. The coupled Eulerian-Lagrangian method (CEL) was
used to model such large deformation analysis. The developed model was veri-
fied by a simulation of Saint-Jude landslide, which occurred in 2010 in Quebec,
Canada. There was a good agreement among numerical results with water con-
tent effect comparing to an analytical function. A larger mobile soil mass
movement with a significant reduction of soil resistance was simulated success-
fully using the developed model.

Keywords: progressive failure, sensitive clays and water content effect.

1 Introduction

Spreads are a type of large landslide occurring in sensitive clays [1], involving horsts
and grabens is the dislocation and translation of the soil mass. In eastern Canada and
Scandinavia, spread represents more than 37% [2]. The undrained shear strength is
one of important parameters for the risk assessment of slow- and fast-moving land-
slides. For that condition, water infiltration into soils may result in degradation of the
shear strength, which can lead to non-uniform crack creation and shear-band emer-
gence. Numerous studies have been performed to simulate the post-peak behavior of
the undrained shear strength of soft clays [3-5] because this is a very important phe-
nomenon for understanding landslide mobilization. There is an agreement between
theoretical and field data; however, the water content effect is not considered in the
previous models.

© Springer Nature Singapore Pte Ltd. 2020 841

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_134
842 N.-P. Doan et al.

In this study, the water content effect on the undrained shear strength of sensitive
clays was examined. The finite element (FE) analysis is performed to simulate the
2010 Saint-Jude landslide that occurred in Quebec, Canada. In numerical simulation,
a well-known analytical function was used to approximate the peak and post-peak
behavior of clayey soils according to the method of Einav and Randolph [6]. Then, a
modified M-C model with and without the water-content effect was developed. The
numerical simulation obtained from the proposed model were compared with field
observations and possible mechanism involved in this landslide is explained.

2 Governing equation

The strain-softening behavior of the soft clay is defined using an exponential relation-
ship between the undrained shear strength and the plastic shear displacement. Dey et
al. [3] modified the function of Eniav and Randolph [6] as the following equation:
(1)
where su is the mobilized undrained shear strength at displacement į; St is the sensitiv-
ity of the soil (sup/suR); į  įt – įe  įpc), with įe and įt being the elastic and total
shear displacement, respectively; į95 is the value of į at which the undrained shear
strength of the soil is reduced by 95% of (sup – suR); and suR is the residual undrained
shear strength.
The relationship between the water content w and the remoulded undrained shear
strength sur of fine-grained inorganic soil is given by Koumoto and Houlsby [7]:
(2)
where the coefficient a is determined as the water content corresponding to sur =1
kPa, and the coefficient b is given by the gradient of the linear function relating the
logarithm of the water content (%) to the logarithm of the undrained shear strength
(kPa).

3 Finite element modelling

3.1 Modified M-C yield criterion

Abbo and Sloan [8] proposed a smooth approximation of the hyperbolic yield surface
to eliminate the singularities of the M-C yield surface:
(3)
(4)
where a and .ș is the meridional rounded parameter in the approximation hyper-
bolic curve. A detailed discussion of the yield criteria of these parameters was pre-
sented by Abbo and Sloan [8]. As findings from previous study [8], the best approx-
imation to original M-C yield criteria is a = 0.05 and șt = 25º.
The yield functions are governed by the additional state variable w (%) to include
the water-content effect on the constitutive model. Hence, the modified consistency
condition of the modified M-C model becomes:
Finite Element Simulation of Water Content-Influenced … 843

(5)
Combining Eq. (5) into the constitutive equations, the plastic multiplier has the fol-
lowing form. The calculation was implemented using the Fortran subroutine in the
Abaqus software package. Finally, the plastic shear strain was calculated using Eq.
(6).

(6)

3.2 Numerical modelling method

This study investigates the water infiltration on the progressive failure and post-
failure of a sensitive clay landslide which occurred in Saint-Jude, Quebec, Canada on
May 10, 2010. Fig. 1 presents the profile of slope was obtained from the field investi-
gation [9] and the simulation of Wang et al. [10]. Soil parameters used for the model
in numerical simulations are presented in Table 1. The CEL method was to overcome
the element distortion problem of large deformation in the commercial software
Abaqus/Explicit. The Eulerian mesh is fixed in space, but the material can move
through its cells. The algorithm tracks the flow of material based on a scalar called the
Equivalent Volume Fraction (EVF), which is assigned to each Eulerian cell and varies
be-tween 0 and 1. EVF = 0 corresponds to a void cell, and EVF = 1 corresponds to a
cell entirely filled with material. Fig. 1 shows the void value of CEL method and the
boundary conditions. The FE analysis includes three steps. In the first step, the soil
elements are brought to the in-situ stress condition by the application of gravitational.
In the second step, the initial landslide is triggered by the activation of eroded block.
In the third step, the analysis is continued without any change in boundary or loading
conditions.

Fig. 1. Geometry and shear strength profile used in FE modelling of the 2010 Saint-Jude land-
slide (after Wang et al. [10]).
844 N.-P. Doan et al.

Table 1. Input parameters for numerical simulations.

sensitive stiff lay-

Soil properties crust*
clays er*
Young’s modulus, E (MPa) 10 10 10
3RLVVRQ¶VUDWLRȞ 0.495 0.495 0.495
Friction angle, ࢥ (deg) 1 1 1
'LODWDQF\DQJOHȥGHJ 1 1 1
Peak undrained shear strength, sup (kPa) 25-65 45 65-120
Residual undrained shear strength, suR (kPa) sup/6 - -
Large displacement undrained shear strength, suld (kPa)
sup/50 - -
Increase of water content (%)
** - -
Sensitivity (St = sup/suR) 6 - -
įpc (%) 0.004 - -
į95 (%) 0.025 - -
įld (%) 2.0 - -
Unit weight (kN/m3) 16.0 18.6 16.8
* crust and stiff layer is modeled as elastic-perfectly plastic without softening.
** water content is calculated by Eq. (2) using the mobilized undrained shear strength.

4 Results and discussion

Fig. 2. Shear band formation of modified M-C model with water content effect.

Fig. 2 presents the numerical results with respect to the shear band development with
water content effect using the modified M-C model, respectively. In general, the anal-
ysis witness the same type of failure shape (spreads) with horsts and grabens. In Fig.
2, it can be seen that the failure occurs initially near the toe of the slope. A shear band
f1 initiates from the top of the eroded block, propagates downward to the stiff layer
Finite Element Simulation of Water Content-Influenced … 845

and then develops horizontally through the sensitive clay layer. As the reduction of
shear strength, curved shear bands f2, f3 and f4 are created as branches of the shear
band f1 from the intersection point P1, P2 and P3. The failure planes propagate further
through the sensitive clay layer, such as shear band f5, which combine previous shear
bands to execute the horst and grabens failure shape. Especially, shear band f3 is
spread out to cover all the region from shear band f5 to f4. The additional shear band f6
is created to connect shear band f4.

Fig. 3. Numerical results of undrained shear strength with shear strain: (a) at location P1, (b) at
location P2, (c) at location P3 and (d) undrained shear strength with liquidity index.

Fig. 3a, 3b and 3c present the determination of the mobilized undrained shear strength
with shear strain for three observed location P1, P2 and P3, respectively. Three ap-
proaches were used for the existing data: (i) analytical fit proposed by Einav and
Randolph [6], (ii) the M-C model with water content effect and (iii) the M-C without
water content effect. In general, the numerical results obtained from M-C model con-
sidering water content effect match well to analytical results obtained from Einav and
Randolph [6] function. Along the failure surfaces, the peak shear stress value were
approximately 25 kPa. Then the mobilized shear stress decreases to suR = 4.2 kPa at
shear strain Ȗ = 0.175 (%) for point P1, Ȗ = 0.008 (%) for point P2 and Ȗ = 0.007 (%)
for point P3. Herein, there is a transition state from plastic to liquid of the sensitive
clay behavior. The mobilized undrained shear strength decreases gradually to suld =
0.5 kPa and then it remains constant at shear strain Ȗ = 1.85 (%). Fig. 3d shows the
semi-logarithmic relationship between the strength and liquidity index. There is a
positive relationship between the index properties and the undrained shear strength.
Clearly, in the transition from plastic to liquid, there are relationships between the
undrained shear strength and the liquidity index of clays. As shown in Fig. 3d the
numerical results were found to be within the range proposed by Houston and Mitch-
ell [11].
846 N.-P. Doan et al.

5 Conclusion

The concept of post-peak behavior provides a good description of the progressive

shear band formation and failure occurrence, but the determination of the undrained
shear strength subjected to a flow zone is difficult. For this reason, a numerical model
considering the water content effect is developed, because water infiltration is essen-
tial in landslide motion. The developed model with water content effect is compared
with an analytical function, which provides a good prediction results of the post-peak
behavior of undrained shear strength than the model without water content effect. The
slope failure mechanisms in spread type could be explained clearly in this present
study.

Acknowledgements. This work was supported by the National Research Foundation of Korea
(NRF) grant funded by the Korea government (MSIT) (No. NRF-2018R1A5A1025137).

References
1. Cruden, D.M., and Varnes D.J.: Landslides types and processes. In Landslides investiga-
tion and mitigation. Special Report 247. Transportation Research Board, National Re-
search Council. Washington, D.C. 37-75 (1996).
2. Demers, D., Robitaille, D., Locat, P., and Potvin, J.: Inventory of large landslides in sensi-
tive clay in the province of Quebec, Canada: preliminary analysis. Landslides in Sensitive
Clays - From Geosciences to Risk Management, Springer, Netherland (2013).
3. Dey, R., Hawlader, B., Phillips, R., and Soga, K.: Large deformation finite-element model-
ing of progressive failure leading to spread in sensitive clay slopes. Géotechnique 65(8),
657–668. (2015).
4. Dey, R., Hawlader, B., Phillips, R., and Soga, K.: Modeling of large-deformation behav-
iour of marine sensitive clays and its application to submarine slope stability analysis. Ca-
nadian Geotechnical Journal 53(7), 1138–1155 (2016a).
5. Dey, R., Hawlader, B., Phillips, R., and Soga, K.: Numerical modelling of submarine land-
slides with sensitive clay layers. Géotechnique 66(6), 454–468 (2016b).
6. Einav, I., and Randolph, M.F.: Combining upper bound and strain path methods for evalu-
ating penetration resistance. International Journal for Numerical Methods in Engineering,
63(14): 1991–2016 (2005).
7. Koumoto, T., and Houlsby, G. T.: Theory and practice of the fall cone test. Géotechnique,
51(8), 701-712 (2001).
8. Abbo, A., and Sloan, S.: A smooth hyperbolic approximation to the Mohr-Coulomb yield
criterion. Computers & Structures 54(3), 427–441 (1995).
9. Locat, A., Locat, P., Demers, D., Leroueil, S., Robitaille, D. and Lefebvre, G.: The Saint-
Jude landslide of 10 May 2010, Quebec, Canada: Investigation and characterization of the
landslide and its failure mechanism. Canadian geotechnical journal, 54(10), 1357-1374
(2017).
10. Wang, C., Hawlader, B. and Perret, D.: Finite element simulation of the 2010 Saint-Jude
landslide in Quebec. In: Proceedings of the 69th Canadian Geotechnical Conference, Que-
bec City, Quebec, Canada (2016).
11. Houston, W.N. and Mitchell, J.K.: Property interrelationships in sensitive clays. Journal of
the Soil Mechanics and Foundation Devision 95, 1037-1062 (1969).
 Non-microbial carbonate precipitation as an
improvement technique of sand

Sung-Sik Park1 [0000-1111-2222-3333], Trung-Tri Le2,* [1111-2222-3333-4444], Seong-Heon Kim2

and Zhenzhen Nong2
1
Associate prof., Dept. of Civil Engineering, Kyungpook National Univ., Daegu, 41566, Korea
2,*
Graduate student, Dept. of Civil Engineering, Kyungpook National University, Daegu,
41566, Korea
ltrungtri@gmail.com

Abstract. In this study, the formation of artificial Calcium Carbonate (CaCO3)

from the reaction between Calcium Hydroxide (CH) and Carbon Dioxide (CO2)
was taken into account to improve sand properties. Firstly, Jumunjin sand was
mixed with 2% of CH and water at a constant water-solid weight ratio of 0.1 be-
fore curing in a carbon dioxide chamber. The precipitated calcite content was
investigated by the variation of CO2 chamber pressure and number of repeatable
treatment. Then, unconfined compression tests of such samples were carried out
to evaluate the influence of calcite content on mechanical properties of sand. It
indicated that as increasing of repeatable treatment, the UCS of treated sand in-
creased due to the higher of calcite precipitation. The first and fifth cycle do not
lead to any significant variations of precipitated calcium carbonate content
(CCC) and unconfined compressive strength (UCS) with the change of carbon
dioxide chamber pressure from 100 kPa to 200 kPa, but the higher CCC and
UCS can be seen clearly at the tenth cycle. After 10 treated cycles, the UCS of
CP1.10 and CP2.10 were 5 times greater than CP1.1 and CP2.1, respectively.
The maximum UCS was 361 kPa of CP2.10 specimen. The stress-strain curves
obtained from UCS test at 200 kPa CO2 gas pressure exhibited more strain at
peak stress than that at 100 kPa and the samples subjected to 10 treated cycles
has more ductile behavior than other samples.

Keywords: calcium carbonate binder, mechanical properties, compressive

strength.

1 Introduction

Soil improvement using traditional cement-based binders have been widely investi-
gated and applied, including soil replacement and grouting. However, the by-product
of cement is energy-intensive and environmentally unfriendly so alternative tech-
niques have emerged within the last two decades such as calcite in-situ precipitation
(CPIS) [1-2] and microbial induced calcite precipitation (MICP) [3-6]. Both of the
artificial calcite precipitation methods improve the engineering properties of soil by

© Springer Nature Singapore Pte Ltd. 2020 847

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_135
848 S.-S. Park et al.

growing calcite crystals on the surface and connecting the particles to the particles at
contact points.
The calcium carbonate precipitation is a process based on the reaction between cal-
cium ions (Ca2+) and carbonate ions (CO32-) that were formed through the hydrolysis
of urea CO(NH2)2 (Eq. 1).
௎௥௘௔௦௘
CO(NH2)2 + 2 H2O ሱۛۛۛሮ 2 NH4+ + CO32- (1)

Ammonium NH4 is a non-toxic salt, it is ionized form of toxic ammonia NH3 and
does ability to instantly change to NH3 in change of pH and temperature. The genera-
tion of high concentrations of these by-product induces toxic effects on human health,
vegetation and atmospheric [7-10].
This study aims to use a different method to form calcite CaCO3 binders without
forming any environmentally un-friendly by-product, where calcite crystal precipitat-
ed by the reaction between calcium hydroxide Ca(OH) 2 and carbon dioxide CO2 (Eq.
2). Because the fact that the production of Ca(OH)2 has also impact on the environ-
ment, so this study considers that CH is a kind of source of supply Ca2+ ions.

Ca(OH)2 + CO2 o CaCO3 + H2O (2)

The effectiveness of this method, the strength development of CaCO3 binders un-
der different conditions such as CO2 gas pressure, exposure time, degree of initial
compaction and temperature was pointed out and discussed by P. De Silva et al.2006,
2009 [11-12].
The main objective of this study is to investigate the ability of this method on the
strength improvement in geotechnical stabilization. Two different CO2 chamber pres-
sures (100 and 200 kPa) and three different number of repeatable treatments (1, 5 and
10) were applied to understand its effect on precipitated carbonate content and uncon-
fined compressive strength.

2 Sample preparation

2.1 Materials
Jumunjin sand was used in the experiments. The average diameter of sand (D50) of
0.382 mm, the D10, D30 and D60 of 0.188, 0.3, 0.415 mm, respectively. The coefficient
of uniformity (Cu) of 2.21 and the coefficient of curvature (Cc) of 1.2. It is classified
as SP according to the Unified Soil Classification System.

2.2 Sample preparation and treatment

To investigate the effect of CO2 chamber pressure, number of repeated treatment on

the engineering properties of sand, several of these parameters were applied and given
in Table 1.
Non-microbial carbonate precipitation … 849

For each cycle of treatment, the sample preparation started with mixing the dry
sand, CH and water. Various number of treated cycle with CH to sand ratio of 0.02
and a constant ratio of water to solid weight of 0.1 were applied. Next, the mixtures
were carried out CO2 curing inside a closed chamber at 100 and 200 kPa pressure for
treated time of 2 hours and then the samples were put into an oven until dried before
doing next cycle.
At the last cycle, before exposing to CO2 in a chamber, the mixture was molded in
cylindrical mold which diameter and height of 5 and 10 cm, respectively.

3 Measurement of carbonate content

The precipitated calcium carbonate formed in sand sample was measured in order to
detect the influence of CO2 pressure, number of repeatable treatment on calcite pre-
cipitated content and engineering properties of treated sand. According to ATSM D
4373-14 [13], the measurement of CaCO3 content is based on the reaction between
calcium carbonate and hydrochloric acid (Eq. 2). Therefore, the CO2 gas pressure
produced from the reaction in a closed chamber will depend on the weight of calcium
carbonate reacting with HCl. A linear relationship between the gas pressure and the
weight of CaCO3 was proposed by series of measurement using fine different weights
of CaCO3 reagent of 0.2, 0.4, 0.6, 0.8 and 1 gram. This relationship was used as a
calibration curve in this study (Fig. 1).

CaCO3 + HCl o CaCl2 + H2O + CO2 (2)

About 5 grams of treated sand was dried in an oven at 110oC for 24 hours, crushed
by hammer, and then the samples were carried out to react with 20 ml of hydrochloric
acid (HCl) 1M in a closed reaction cylinder which has a CO2 pressure gauge. The
weight of carbonate contained in sand sample can be obtained through the relationship
below.

Fig. 1. The relationship between CO2 pressure and calcium carbonate content.

4 Results and Discussion

4.1 Calcium carbonate precipitated content

The precipitated calcium carbonate content in treated sand was summarized in Table
1. Although the change in carbon dioxide chamber pressure from 100 kPa to 200 kPa,
850 S.-S. Park et al.

there is no a significant difference in the CCC of the specimens CP1.1 (2.139%) and
CP2.1 (2.142%). The similar results are obtained for samples subjected to 5 repeata-
ble treatments, CP5.1 (11.369 %) and CP5.2 (11.189%). In contrast, at 10 th cycle the
CCC of CP1.10 and CP2.10 samples are 17.928 % and 22.875%, respectively. It is
clear that the CCC of CP2.10 is higher almost 5 % than that of CP1.10. It may be
explained by the increase of density and ability of carbon dioxide to penetrate through
the specimen.
The increase of density for each cycle treatment includes: (i) the precipitated calci-
um carbonate on the surface and particle skeletons; (ii) the residual calcium hydroxide
did not react with carbon dioxide due to the low solubility of CH in water. Additional-
ly, the formation of calcium carbonate is affected by amount of Ca 2+ and CO2 dis-
solved in water, which depends on the initial CO2 concentration and pressure in the
chamber [14-15]. Therefore, after the 5th cycle, the penetration and dissolution of car-
bon dioxide through the denser sand sample increase at the higher chamber pressure.

4.2 Unconfined compressive strength

The confining stresses in triaxial testing may destroy bonding or cementation between
weakly cemented soil particles [16]. Under such triaxial testing conditions, the ce-
mented soil is not as effective as observed under unconfined condition [17]. Hence,
serials of unconfined compression test to investigate effect of the treatment on com-
pressive behavior of treated sand with different parameters of mixture. Table 1 sum-
marizes the results of UCS stress and the axial strain at peak stress, given below.

Table 1. . Experimental results of unconfined compression tests.

Chamber Number of UCS Strain Precipitated
Test ID pressure cycle treat- kPa at peak stress calcite
(kPa) ment (%) (%)
CP1.1 1 50.70 0.64 2.139
CP1.5 100 5 188.93 1.23 11.369
CP1.10 10 275.22 1.45 17.928
CP2.1 1 67.18 0.8 2.142
CP2.5 200 5 174.31 1.55 11.189
CP2.10 10 360.89 1.89 22.875

Fig. 2 shows the UCS with axial strain of treated sand with change in number of
repeatable treatment and CO2 pressure for curing stage in the closed chamber. It indi-
cates that the UCS results for both case of the 100 kPa and 200 kPa chamber pressure
develop with the increase of the number of treatment due to the higher formation of
calcium carbonate in sand specimens. The effect of calcium carbonate content on
strength development can be seen clearly in Fig. 3. Calcium carbonate cemented and
filled pores between the sand particles, thus improving the strength of treated sand
[16]. The maximum of UCS of 360.89 kPa is obtained from CP2.10. Because of the
more precipitated calcium carbonate content, the UCS of CP2.10 (360.89 kPa) is
higher than CP1.10 (275.22 kPa), while samples subjected to 1 or 5 cycles are not a
Non-microbial carbonate precipitation … 851

significant difference at peak strength. After 10 treated cycles, the UCS of CP1.10 and
CP2.10 are approximately 5 times greater than CP1.1 and CP2.1, respectively. Partic-
ularly, the strain at the peak strength of all the samples at 200 kPa CO 2 gas pressure is
higher than other samples cured at 100 kPa. The stress-strain curves of CP1.10 and
CP2.10 depicts that after 10 cycles of treatment, specimens exhibit more ductile be-
havior than other samples. As a result of the high concentration of precipitated calci-
um carbonate and the forming of cemented particles may lead the structure of sand
has more fracture resistance.

Fig. 2. UCS results of treated sand with different CO2 pressure: (a) 100 kPa and (b) 200 kPa.

Fig. 3. Effect of repeatable treatment on development of carbonate content (a) and UCS (b).

5 Conclusions

This paper used calcium hydroxide and carbon dioxide as an environmentally friendly
method to form calcite binders. Jumunjin sand was treated with CO2 pressure of 100
kPa, 200 kPa and number of treated cycles of 1, 5 and 10. Then, the precipitated car-
bonate content was measured to investigate its effect on the strength and compressive
behavior through the unconfined compression test.
As the number of repeatable treatment increases, the UCS raised significantly be-
cause of the higher content of calcite precipitation. The maximum of UCS reached
360.89 kPa after 10 cycles at 200 kPa CO 2 chamber pressure. There was a small dif-
ference in CCC and UCS at 1st and 5th cycle when the chamber pressure changed from
100 kPa to 200 kPa. However, at the 10th cycle, the strength of treated samples were
852 S.-S. Park et al.

almost 5 times greater than that subjected at 1 treated cycle. The reason is the higher
calcite precipitated content was formed. Additionally, the samples at higher CO 2 pres-
sure provided more strain and more ductility than other samples.

References
1. Ismail, M.A., Joer, H.A., Randolph, M.F. and Merritt, A.: Sample preparation technique
for artificially cemented soil. ASTM Geotech. Testing J.23, No 2, 171-177 (2000).
2. Ismail, M.A., Joer, H.A, Randolph, M.F. and Merritt A.: Cementation of porous materials
using calcite. Geotechnique 52:313-324 (2002).
3. DeJong, J. T., Fritzges, M. B. and Nusslein, K.: Microbially induced cementation to con-
trol sand response to undrained shear. J. Geotech. Geoenviron. Eng., 1381-1392 (2006).
4. DeJong, J. T., Mortensen, B. M., Martinez, B. C. and Nelson, D. C.: Bio-mediated soil im-
provement. Ecol. Eng., 36(2), 197-210 (2010).
5. Chou, C., Seagren, E. A., Aydilek, A. H., Lai, M.: Biocalcification of sand through
ureolysis. J. Geotech. Geoenviron. Eng., 1179-1189 (2011).
6. Martinez, B. C. et al: Experimental optimization of microbially induced carbonate precipi-
tation for soil improvement. J. Geotech. Geoenviron. Eng., 587-598 (2013).
7. Tobler, Dominique, J., et al: Comparison of rates of ureolysis between Sporosarcina
pasteurii and an indigenous groundwater community under conditions required to precipi-
tate large volumes of calcite. Geochimica et Cosmochimica Acta, 75.11: 3290-3301 (2011)
8. Reddy,Sudhakara, M.: Biomineralization of calcium carbonates and their engineered ap-
plications: a review. Frontiers in microbiology,4: 314 (2013,)
9. Van Paassen, L. A., Daza, C. M., Staal, M., Sorokin, D. Y., van der Zon, W., and van
Loosdrecht, M. C.: Potential soil reinforcement by biological denitrification. Ecological
Engineering, 36(2), 168 (2010).
10. Haouzi, F. Z. and Courcelles, B.: Major applications of MICP sand treatment at multi-scale
levels: A review. In Conf. Proceed. GeoEdmonton (2018).
11. De Silva, P., Bucea, D.R. Moorhead, Sirivivatnanon, V.: Carbonates binders: reaction ki-
netics, strength and microstructure, Cem. Concr. Compos. 28, 613±620 (2006).
12. De Silva, P., Bucea, L., Sirivivatnanon, V.: Chemical, microstructural and strength devel-
opment of calcium and magnesium carbonate binders, Cement and Concrete Research 39,
460±465 (2009).
13. ASTM D4373-14: Standard Test Method for Rapid Determination of Carbonate Content of
Soils, ASTM International, West Conshohocken, PA (2014).
14. Moorehead, D.R.: Cementation by the carbonation of hydrated lime, Cem Concr Res,
16:700-8 (1986).
15. Van Balen, K., Gemert Van, D.: Modelling lime mortar carbonation. Mater Struct, 27:393-
398 (1994).
16. Saxena, S.K., Lastrico, R.M.: Static properties of lightly cemented sand, J Geotech Eng
Div, 104(12):1449±64 (1978).
17. Park, S.S.: Unconfined compressive strength and ductility of fiber-reinforced cemented
sand, Constr Build Mater, 25 (2), 1134-1138 (2010).
 Modeling on Cuttings Transport in Inclined and
Horizontal Well Drilling
Do Quang Khanh1,*, Truong Trong Tuan Dat1,2, Kieu Phuc1, Tran Thi Mai Huong1,
Hoang Trong Quang1
1
Faculty of Geology and Petroleum Engineering, HoChiMinh City University of Technology
(HCMUT), Vietnam National University - HoChiMinh City (VNU-HCM), Vietnam
2
PetroVietnam Drilling and Well Services Corporation (PVD), Vietnam
*Corresponding author: dqkhanh@hcmut.edu.vn

Abstract. In inclined and horizontal well drilling, cuttings transport is playing a very
important role in obtaining a safe and efficient drilling plan. The paper presents two
empirical models, namely Larsen's model and Rubiandini's model, which were considered
to evaluate the cuttings transport in both inclined and horizontal well drilling in offshore
Vietnam. A parametric study considering different drilling parameters such as mud weight,
rate of penetration (ROP), mud rheology, etc. indicated that the two empirical models
provided the same trends of the flow velocity and flow rate required for transportation of
mud cuttings. It is concluded from all simulations that the flow rates predicted by both
Larsen's model and Rubiandini's model are reasonably close to the actual flow rates in the
drilling operation. Moreover, it is also observed from the analyses of the horizontal drilling
case that the use of Rubiandini's model could generally result in larger flow rate required
for cuttings transport compared to Larsen's model.

Keywords: Cuttings transport, Empirical models, Inclined and horizontal well drilling,
Offshore Vietnam.

1 Introduction

Cuttings transport is one of the most common and complex subjects for drilling
operations. It is also required to be monitored and properly controlled during the entire
well drilling operations. Circulation of drilling fluid is an integral part of the drilling
operation. The drill bit crushes the rock formation into small pieces called cuttings. The
drilling fluid is pumped through pipes and then circulated back through the annulus
bringing cuttings to the surface facilities. The ability of a circulating drilling fluid system
to transport cuttings is known as the carrying capacity of the drilling fluid.
Cuttings transport predictions are essential for the planning of inclined and horizontal
drilling. However, there are still a little limited number of works relating to the cuttings
transport in deviated and horizontal wells. In this study, /DUVHQ¶V DQG 5XELDQGLQL¶V
models [1, 2] will be considered and compared to identify differences between them by
using a calculating program written from Matlab language. The obtained simulations
were based on the rheological data, drilling parameters, and cutting properties from a
horizontal well drilled at offshore Vietnam.

© Springer Nature Singapore Pte Ltd. 2020 853

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_136
854 D. Q. Khanh et al.

2 Modeling Of Cuttings Transport Process

2.1 /DUVHQ¶V0RGHO
In 1997, based on a large number of experimental studies Larsen et al. focused on cuttings
size, inclination angle and mud weight that significantly affect cuttings transport in
directional and horizontal wells [1]. The mud minimum rate Vmin is a sum of velocity of
the fallen cuttings Vcut and the slip velocity Vslip. Based steps to calculate the cuttings
WUDQVSRUWSURFHVVIURP/DUVHQ¶VPRGHODUHVKRZQLQWKHIROORZLQJIORZFKDUW)LJ

Fig. 1. Flowchart to calculate the cuttings transport from /DUVHQ¶Vmodel [3].

2.2 5XELDQGLQL¶V0RGHO
In 1999, Rubiandini presented a new equation to estimate the mud minimum rate for
cuttings transport in the inclined-until-horizontal well [2]. He believed that the cuttings
transport mechanisms are affected mainly by mud weight, inclination angle and rotary
speed RPM. Therefore, the correction factors of these parameters played a main role in
his model. Based steps to calculate the cuttings transport process from RXELDQGLQL¶V
model are shown in the following flowchart (Fig. 2).
Modeling on Cuttings Transport in Inclined … 855

Fig. 2. Flowchart to calculate the cuttings transport from Rubiandini¶VPRGHO>@

3 Case study for a horizontal well drilling at offshore Vietnam

3.1 Input data

In this study, predictive calculations for the case study well are carried out with drilling
parameters as:
Dpipe = 5.5 inch; Dhole = 8.5 inch;
Drilling fluid: ȡm=1.16 ± 1.26 g/cm3 ± 0.02;
PV = 25 cp; YP = 25 lbf/100ft2;
Dcutt = 0.1 mm = 0.004 inch;
ROP = 40 m/h or 131.234 ft/hr;
856 D. Q. Khanh et al.

RPM = 120 rpm.

3.2 Results and discussions on the required flow rate

Using the Matlab computer program, predictive curves are drawn from the experimental
GDWD VHW ,Q WKH VLPXODWLRQV WKH IROORZLQJ YDULDEOHV ZHUH XVHG IRU ERWK /DUVHQ¶V DQG
5XELDQGLQL¶VPRGHOVDQGonly one of these parameters is varied in each simulation. The
predictive curves of the required flow velocity or flow rate versus with variation of
parameters, which includes mud weight, rate of penetration ROP, mud rheology, etc. as
a varying parameter, IRUERWK/DUVHQ¶VDQG5XELDQGLQL¶VPRGHOVDUHVKRZQLQ)LJ
5.

Mud weight as a varying parameter

Fig. 3. Flow velocity vs. angle of inclination with mud weight as a variable IURP/DUVHQ¶VPRGHO
RQWKHOHIWDQGIURP5XELDQGLQL¶VPRGHORQWKHULJht).

ROP as a varying parameter

Fig. 4. Flow velocity vs. angle of inclination with ROP as a variable (IURP/DUVHQ¶VPRGHORQWKH
left and from 5XELDQGLQL¶VPRGHORQWKHULJKW).
Modeling on Cuttings Transport in Inclined … 857

Mud rheology as a varying parameter

Fig. 5. Flow velocity vs. angle of inclination with mud rhelogy as a variable IURP/DUVHQ¶VPRGHO
RQWKHOHIWDQGIURP5XELDQGLQL¶VPRGHORQWKHULJKW

Moreover, at inclination angle of 75 deg. the obtained results on the required flow rate
IURPXVLQJ/DUVHQ¶VDQG5XELDQGLQL¶VPRGHOVDUHFRPSDUHGLQWKHIROORZLQJWDEOHV The
error equation [2] was used in order to identify the distinctions or similarities between
LDUVHQெVDQG5XELDQGLQLெVPRGHOV

(1)

Table 1. Flow velocity value vs. inclination angle with mud weight as variable parameter.

Mud weight (ppg) 9.51 10.35 10.68

/DUVHQ¶V0RGHO 51 50 49
Flow rate (l/s)
5XELDQGLQL¶V0RGHO 69 68 67
Error of flow rate (%) 26.1% 26.5% 26.9%

Table 2. Flow velocity value vs. inclination angle with ROP as variable parameter.

ROP (ft/hr) 65.6 131.2 196.9

/DUVHQ¶V0RGHO 48 49 50
Flow rate (l/s)
5XELDQGLQL¶V0RGHO 65 68 70
Error of flow rate (%) 26.1% 28.0% 28.6%

Table 3. Flow velocity value vs. Inclination angle with mud rheology as variable parameter.

Mud Rheology (PV/YP) 15/25 25/25 35/25

/DUVHQ¶V0RGHO 47 50 52
Flow rate (l/s)
5XELDQGLQL¶V0RGHO 67 68 70
Error of flow rate (%) 29.9% 26.5% 25.7%
858 D. Q. Khanh et al.

%\ DQDO\]LQJ IORZ UDWH UHVXOWV IURP ERWK /DUVHQ¶V DQG 5XELDQGLQL¶V models, the
differences between these two models and how various drilling parameters affecting
cutting transport can be observed as following:
- Mud weight as a variable: both models showed the same trend, namely the flow
velocity decreases as mud weight increases. This means that high mud weight improves
cuttings transport. The difference between t h e s e two models increases slightly as mud
weight increases (mean error of 26.5%).
- ROP as a variable: both models also showed the same pattern, n a m e l y h i g h
R O P v a l u e s g e n e r a t e d h i g h f l o w v e l o c i t y . The difference between these
two models was rather significant (26.1%, 28.0% and 28.6%).
- Mud rheology as a variable: both models indicated the same trend, namely higher
mud rheology produced higher flow velocity. However, the difference between these
two models decreases as mud rheological parameter increases (29.9%, 26.5% and 25.7%).

4 Conclusions
Based on all simulation scenarios performed by using LarseQ¶V DQG 5XELDQGLQL¶V
models, it could be concluded that the flow rates for cuttings transport predicted by both
Larsen's model and Rubiandini's model are reasonably close to the actual flow rates in
the drilling operation. Moreover, it is also observed from the analyses of the horizontal
drilling case that the use of Rubiandini's model could generally result in larger flow rate
required for cuttings transport compared to Larsen's model.
In practice, it is recommended using both these models to predict the critical mud flow
rates, then comparing and monitoring actual data obtained from the cuttings transport
process to select the appropriate model at every given inclination angle in inclined and
horizontal well drilling.

References
1. Larsen T.I., Pilehvari A.A., Azar, J.J.: Development of a new cuttings-transport model for
high-angle wellbores including horizontal wells, Society of Petroleum Engineers. Document
ID: SPE-25872-PA, (1997).
2. Rudi Rubiandini R.S.: Equation for mud estimating minimum rate for cuttings transport in an
inclined-until-horizontal well, Society of Petroleum Engineers, Document ID: SPE-57541-
MS, (1999).
3. Ranjbar R.: &XWWLQJWUDQVSRUWLQLQFOLQHGDQGKRUL]RQWDOZHOOV0DVWHU¶V7KHVLV6WDYDQJHr, 102
p, (2010).
Study on Cement-Treated Soil with RoadCem Additive in
Construction of Rural Roads: A Case Study in Viet Nam
Ba Thao Vu1,2, Thi Thanh Huong Ngo2(*), Quoc Dung Nguyen1,
Anh Quan Ngo1, Lanh Si Ho3, 2
1Institutefor Hydraulic Construction, Vietnam Academy for Water Resources; Address:
No. 3 Alley 95 Chua Boc Street, Dong Da, Ha Noi, Viet Nam.
2
Faculty of Civil Engineering, University of Transport Technology; Address:54 Trieu
Khuc, Thanh Xuan, Ha Noi, Viet Nam. (*) Email: huongnt@utt.edu.vn
3
Hiroshima University, Department of Civil Engineering, 1-4-1, Kagamiyama, Higashi-
Hiroshima, Hiroshima 739-527, Japan.

Abstract. The use of soil in place for road construction could reduce construction
cost and environmental impact because this method can replace the traditional
materials that are depletion such as sand, gravel. This paper presents the labora-
tory results for selecting proportion and inspection testing results of experimental
road construction using cement-treated soils with RoadCem additives. RoadCem
additive improved the hydration of cement when cement reacts with soil particles,
thus increasing the strength of reinforced soils, especially tensile strength. The
results of inspection tests showed that the average of compressive strength, ten-
sile strength, and modulus elastic in the saturated condition of reinforced soil sat-
isfied the design and standard requirements in Vietnamese standard.

Keywords: Rural road, RoadCem additive, Soil in place, Cement-treated soils.

1 Introduction

The majority of rural road in Viet Nam is constructed using cement concrete. Cement
concrete road has advantages in terms of simple technology that local citizens could
construct by themselves. However, the cement concrete road has disadvantages because
there are cracks between joints due to impropriate technical treatment at road founda-
tion and joints. Many road construction technologies using cement mixed soils in place
with additive have been applied. Some additives from overseas had been used for rural
pilot road construction in Viet Nam, but most of them are not successful or even suc-
cessful projects have not been replicated because of inadequate construction process
and inappropriate for soil with high organic content and salty soil.
The RoadCem (RC) additive (i.e. called ROVO) has been applied widely to con-
struct road using soil in place [1]. RC could enhance the compressive strength of ce-
ment-treated sandy soil in the salty environment from 13.0% to 57.9% [2]. M.S. Ouf
[3] conducted experiments on soil sample in Egypt that mixed with binders: RC, blast
furnace slag, lime, and cement. It was found that the compressive strength and elastic
modulus of cement-treated soils increased significantly while swelling decreased. In
addition, when RC is added in cement-treated soils, the tensile strength can be en-
hanced, leading to a reduction 50% in thickness and number of cracks compared with

© Springer Nature Singapore Pte Ltd. 2020 859

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_137
860 B. T. Vu et al.

cement-treated soils without RC [4]. When 1% of RC added to the soil of local road
foundation using 9% of cement, the compressive strength and tensile strength satisfied
the requirement of durability level II in TCVN 10379:2014 [5]. Besides, N.D. Sy and
C.T. Linh [6] has conducted analyzing advantages and technical-economic efficiency
of this method compared with the method using traditional materials in Russia and rec-
ommended applying this method in Vietnam. However, they have only introduced the
results conducted from overseas; it has not been conducted and applied in domestic.
Thus, the aim of this paper is to investigate the effect of RC additive on the mechanical
properties of cement-treated soil based on the results of laboratory test and inspection
test for a pilot project for rural road in Viet Nam.

2 Principle of cement-treated soil using RoadCem additive

RoadCem product is a nanoparticle-based additive, consisting of about 20 different

components, including alkaline earth metals and synthetic zeolites, with the addition of
active complexes. Based on the results of research in many laboratories and practical
projects, when RC additive mixed with cement-treated soils, RC has many following
effects [1]. (1) The presence of RC additive in cement-treated soils can allow water to
penetrate deeper to the center of cement particle. This makes the reactions of C3S and
C2S minerals more radical, i.e. the volume of the non-reactive region decreases. The
lattice crystals formed will penetrate and develop in the pore system of soils. As a result,
the cement-treated soils will decrease permeability and increase the strength. (2) When
RC additive added to the mixture, the interaction and bonding between granular mate-
rial and cement paste are improved. For the case that only cement mixed with soils,
cement hydration process happened slowly because clay particle is covered by a poten-
tial water membrane around the surface. The amount of exceed water and cement hy-
dration products will create brittle links. However, when RC added to cement-treated
soils, RC reacts with exceed water to form fibrous crystals structure. These crystal
structures improve the links between soil particles and cement paste. Thus, the tensile
strength of cement-treated soils is improved [7].

3 Construction of pilot rural road

3.1 General introduction of pilot project

The interprovincial rural road that connects Xuan Dao village and Tuan Xuan village
in Luong Tai commune (Hung Yen province, Viet Nam) has a length of 1,450 m. The
pavement was made from soil, broken brick, coal slag, the width of the road is approx-
imately from 3.5 m to 4.0 m. Cross-section and dimension of the road are shown in Fig.
1. The main design parameters including: Vehicle load < 10 tons; traffic volume: 20
vehicles/day; Pavement width: B = 5.0 m, Width of road foundation: b = 3.0 m; thick-
ness of road foundation made from cement-treated soils with RC: h = 20 cm; pavement
was made from crushed stone-bitumen with thickness of 1.5 cm.
Study on Cement-Treated Soil with RoadCem … 861

Fig. 1. Cross-section of road

3.2 Experiment for selecting mix proportion in laboratory

The strength of cement-treated soils was determined using laboratory test (Fig. 2), the
proper ratios between cement/dried soil and RC/dried soil were selected to be 160 kg/m3
and 1.4 kg/m3 for one cubic meter of dried soil, respectively. The average compressive
strength (Rn28) was 2.0 MPa, the tensile strength (Rku28) was 0.8 MPa, and elastic mod-
ulus (E28) was 350 MPa. According to 22 TCN 211-06 standards [8], the splitting
strength was calculated from the tensile-bending strength based on the following for-
mula: Rku=Kn.Rech, the empirical relationship factor Kn = 1.6 ± 2.0 is for material rein-
forced with a non-organic binder and K = 2.0 for material reinforced with organic
binder. In this research, we chose Kn = 1.8 to calculate and convert the splitting strength
Rech28bh = 0.44 MPa. Based on the CBR index of road foundation at the construction
site, load and number of vehicle and strength of reinforced soil with chosen amount of
binder; the thickness of reinforced was calculated to be 20 cm.

a) Compressive strength, Rn b) Tensile strength, Rech

Fig. 2. The strength results of laboratory test of cement-treated soils overtime with different C
(cement)-R (RC) ratios: C-R=150-1.5; 160-1.4; 180-1.8; 200-2.0. Unit: kg/m3 of dried soil.

It is noted that RC plays an important role to improve the tensile strength of cement-
treated soil due fiber link was created with high performance in tension resistance.
Based on the results of many studies, for example, the study of N.H. Tri et al. [9] on
the test determines the strength of reinforced soil using cement, fly ash, and lime to
construct traffic road, the ratio between R ku28/Rn28 was approximately 0.1; while, the
result of this study showed that with different ratios of binder, the ratio of R ku28/Rn28
was approximately 0.4.
862 B. T. Vu et al.

3.3 Preparation and construction procedure

Materials and Construction machine. The material includes RC additive, Cement
PCB 40, Crushed stone with particle size of 2/5 mm or 5/8 mm with ratio of 13 kg/m2;
hot bitumen, ratio 2.4 kg/m2; tap water: require 80 l/m2; bamboo pile for positioning
route; bamboo pipe for dividing construction segment; plastic container for measuring
RC additive; Diesel oil was filtered clearly. In this study, Clegg hammer is used for
measuring foundation harness instead of CBR test; John Deer tractor and Written
WS2500 milling head; leveling machine; vibrating roller compactor 8 tons; air com-
pressors for cleaning the road surface.
Construction procedure. The construction procedure includes mainly 11 steps as
follows: Step 1: Leveling roadbed using bulldozer, excavator. Step 2: Plugging land-
mark and dividing the construction segment following design records. Step 3: Spread-
ing RC at the specified rate in the construction area. Step 4: Milling time 1: Soil + RC,
milling depth equal to 2/3 design depth. When mixing soils with RC, water needs to be
sprayed to get a wet mixture. Step 5: Flatted surface by manual. Step 6: Rolling and
flatting one turn (soil + RC). Step 7: Spreading cement at the specified rate in the con-
struction area. Step 8: Milling and mixing second time (soil + RC + cement) at the
design depth. Step 9: Rolling, compacting and flatting surface. Step 10: Spraying water
and maintaining surface 1 time/4 hours during 7 days after constructing. Step 11: Fin-
ishing works of road surface using 2 layers of macadam asphalt.
3.4 Results and discussions
After mixing uniformly the mixture at the site, the specimens at the site were cast and
maintained to test at the age of 7 days in a saturated condition. The project has been
operated at the age of 28 days, the samples were drilled based on standard TCVN
10379-2014 [10]. For every 1,000 m long, drilled 6 samples (3 samples for the com-
pression test and 3 samples for the splitting tensile test) that are not on the same section
but are distributed evenly over the 1000 m long in order to examine strength as well as
examine thickness and dried unit weight of the sample. The diameter and length of
compression of splitting tensile samples were 101 mm. The samples were pre-treated
and two surfaces were flatted then submerged into water for 96 hours before conducting
the test. The compression and splitting tensile tests were conducted in accordance with
ASTM D1633 standard [11, 12].
The inspection test results are shown in Table 1. Some representative stress-strain
curves in saturated and unsaturated condition of the unconfined compression test are
presented in Fig. 3. The compressive strength, tensile strength, and modulus were
higher than the designated values and the values determined in laboratory before con-
struction, these values also satisfied the requirement of TCVN 10379-2014 standard
[10]. The ratio with 160 kg of cement and 1.4 kg of RC in 1 m3 of soil is an appropriate
ratio in order to reinforce the pavement that meets the requirement of strength level II
as specified in TCVN 10379-2014; compressive strength in saturated condition at the
age of 7 days, Rn7bh =1.8 MPa •>03D@ and at the age of 28 days, Rn28bh =4.63 MPa
•>03D@; tensile strength in saturated condition at the age of 28 days Rech28bh =1.18
Study on Cement-Treated Soil with RoadCem … 863

MPa •>03D@and modulus, ( 03D•>03D@7KHWHVWUHVXOWVRIGLIIHUHQW

mix proportions in this study can be referred for other similar projects.

Fig. 3. Relationship between stress-strain of the unconfined compression test (BH: saturated,
KBH: unsaturated)

Table 1. Strength of reinforced soil after construction

Rn7bh Rn28bh Rn28bh E28bh

Curing condition
(MPa) (MPa) (MPa) (MPa)

Saturation 1.8 4.63 1.18 525

Unsaturation 2.5 6.95 1.64 630
1.0/2.0 0.0/0.8 200/350/
TCVN 10379-2014 0.5/1.0/2.0(*)
/3.0(*) /1.2(*) 400(*)
All strength indicators at the tests were higher than the de-
Comment
sign strengths and satisfy the grade II of the strength.

Note: - (*) three strength grade of I, II, III. (2)

- Rn7bh, Rn28bh, Rech28bh, E28bh are compressive strength at 7 days and 28 days, tensile
strength, modulus at 28 days, in saturated condition of reinforced soil, respectively.
The strength of cement-treated soils with RC additive in construction site was higher
than that in the laboratory (see values of strength in Fig. 2) before construction, this can
be attributed to the soils of old foundation that contained broken bricks with diameter
from 3 to 8 cm; at construction site, milling machine can crush broken brick into small
pieces, however in laboratory, the large pieces of brick were removed before mixing.
At the construction site, the small broken bricks with diameter from 1 to 2 cm played a
role as coarse aggregate that could enhance the strength of reinforced soil.

4 Conclusions

Based on the results of laboratory and pilot road project using cement-treated soils with
RC additive, some conclusions can be drawn as follows:
864 B. T. Vu et al.

- Road pavement construction technology using RC additive is suitable for rural

road pavement as this technology can utilize soil in place and provide strengths
that satisfied standards. Thus, it is not necessary to adopt cement concrete pave-
ment using coarse aggregates such as gravel or crushed stone.
- RC additive could enhance hydration ability of cement when mixed with soils that
produced fiber link, which could improve tensile resistance of reinforced soils.
- The average of compressive strength, tensile strength, and modulus elastic in the
saturated condition of reinforced soil satisfied the designed and requirements of
rural road design standard TCVN 10379-2014.
Acknowledgment
7KLVUHVHDUFKZDVVXSSRUWHGE\WKH,QGHSHQGHQW1DWLRQDO3URMHFW³5HVHDUFKRQWKHXVHRIQDWXUDO
pozzolan in the construction and maintains of rural roads and irrigation works in Dak Nong prov-
ince´ No. DTDL.CN-55/16, the Ministry of Science and Technology assigned Vietnam Acad-
emy for Water Resources to implement. Inspection experiments were conducted by Highway
Laboratory Center and the Faculty of Engineering at University of Transport Technology.
References
1. PowerCem Technologies: Technical report design, 12-13 (2010).
2. William A.T.: Laboratory Evaluation of the Effect of Cement Concentration, Water
Salinity and the Roadcem Additive on Kalahari Soil Strength. Master thesis (2006).
3. M.S. Ouf.: Effect of using pozzolanic materials on the properties of Egyptian soils.
Life Science Journal, 9(1), 554-560 (2012).
4. P. Wu: Cement Stabilized Materials with Use of RoadCem Additive. Delft University
of Technology. Master thesis (2015).
5. V.B. Thao, et. al.: Construction ability of rural road by cement-treated soil in place
with RoadCem additive and natural pozzolan. Journal of Transport, 58(special), 85-91
(In Vietnamese) (2016).
6. N.S. Duc, C.T. Linh.: Application of reinforced soils with RoadCCem in road con-
struction. Journal of Transport, 11, 47-49 (In Vietnamese) (2012).
7. PowerCem Technologies. Manual for working with RoadCem (2009).
8. 22TCN 211-06: Flexible pavement ± Design requirements and instructions. Ministry
of Transport, (In Vietnamese) (2006).
9. N.H. Tri et. al.: Research on appropriate technology for rural road construction. State-
level independent project, No: DTDL.2012-T/15 (In Vietnamese) (2015).
10. TCVN 10379:2014: Soils stabilized with inorganic adhesive substances, chemical
agent or reinforced composite for road construction ± Construction and quality control
(In Vietnamese) (2014).
11. ASTM D1633 - Standard Test Methods for Compressive Strength of Molded Soil-Ce-
ment Cylinders (2000).
12. TCVN 8862:2011 - Standard test method for splitting tensile strength of aggregate ma-
terial bonded by adhesive binders (In Vietnamese) (2011).
Application of Steel Pipe Piles with Wings installed in Soil
Cement Column for Building Structure in Vietnam

Pham Hoang Kien1,* and Daisuke Ito2

1
University of Transport and Communications, No.3 Cau Giay street, Hanoi, Vietnam
2
Asahi Kasei Construction Materials Corporation, Tokyo 101-8101, Japan
*Corresponding author: phkien@utc.edu.vn

Abstract. The steel pipe piles with wings installed in soil cement (referred to as
ATT) developed in Japan and Vietnam are composite foundation piles made of
steel pipes with wings with foundation improvement pillars (referred to as col-
umns) in soil cement and spiral wings attached to the shaft intermittently. This
kind of composite foundation pile has large vertical and horizontal bearing ca-
pacities and is a construction method of foundation piles with consideration of
environments with little surplus soil, using steel pipes with small diameters. In
this studyDIWHULQWURGXFLQJ$77¶Voverview and construction process, applica-
tion of this composite foundation piles for a building structure was introduced.
The in situ loading test results was also presented. Furthermore, from results of
loading tests done in the application project, the loading capacity and considera-
tion of $77¶VDSSOLFDELOLW\LQ9LHWQDPwas verified.

Keywords: composite pile, steel pipe, soil cement column, loading test.

1 Introduction
The steel pipe piles with wings installed in soil cement column have been developed
in Japan and Vietnam by Asahi Kasei Construction Materials and Tenox Corporation.
From now, this kind of composite foundation pile will be referred to as ATT
(AsahiKasei-Tenox Technology). Due to combination effect of soil cement column
and steel pipe with spiral wings, ATT composite pile can provide a greater bearing
capacity compared to the single body of column and single body of steel pipe with
wings. The existence of steel pipe with wings also helps distributing effectively the
load acting on pile head to the column and prevents the occurrence of stress
concentration. Consequently, compression failure which occurs at upper part of
column single body can be avoided. This kind of composite foundation pile is also an
environmentally friendly construction method that has low vibration and noise, and
produces little surplus soil. Up to now, there have been more than 5,000 cases of
construction results.
In Japan and Vietnam, tKHUHDUHPDQ\FDVHVRIZHDNJURXQGZKHUHSLOH¶VVXSSRUW
layer is very deep. In these cases, the piles become very long and problems have risen
such as the cost for substructure becomes very high in comparison to the cost for su-
perstructure. ATT is a foundation pile construction method developed to solve these

© Springer Nature Singapore Pte Ltd. 2020 865

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_138
866 P. H. Kien et al.

kinds of problems. This is because ATT is a pile foundation developed as a friction

pile with good cost performance, which does not become fixed to the support layer
and displays bearing capacity by pile shaft friction. In this study, after introducing
$77¶V overview and construction process, application of this composite foundation
pile for a building structure was introduced. In situ loading test result was also pre-
sented. Furthermore, from results of loading tests done in the application project, the
ORDGLQJFDSDFLW\DQGFRQVLGHUDWLRQRI$77¶VDSSOLFDELOLW\LQ9LHWQDPZDVYHULILHG.

2 Overview of ATT

Schematic sketch of the structure of ATT is shown in Fig.1. Steel pipe pile with in-
termittent spiral wings is installed in soil cement column. The end and the head of the
steel pipe is blocked by steel plates and inside of the steel pipe is hollow. There is a
spiral wing and drilling blade at the end of the steel pipe. The standard specifications
RI$77¶VVWHHOSLSHDQGFROXPQDUHshown in Table 1.

(d) Steel pipe

(b) Schematic sketch (c) Toe shape (a) Photo of ATT composite pile
Fig. 1. Overview structure of ATT
Table 1. Standard specifications
Column Dc Steel pipe D0 (mm)
Wings D (mm)
(mm) 114 139 165 190 216 267 318 355
250 㻌䕿㻌 㻌䕿㻌 㻌 㻌 㻌 㻌 㻌 㻌
500
300 㻌䕿㻌 㻌䕿㻌 㻌䕿㻌 㻌 㻌 㻌 㻌 㻌
350 㻌䕿㻌 㻌䕿㻌 㻌䕿㻌 㻌䕿㻌 㻌 㻌 㻌 㻌
600
400 㻌 㻌䕿㻌 㻌䕿㻌 㻌䕿㻌 㻌䕿㻌 㻌 㻌 㻌
450 㻌 㻌 㻌䕿㻌 㻌䕿㻌 㻌䕿㻌 㻌䕿㻌 㻌 㻌
700
500 㻌 㻌 㻌 㻌䕿㻌 㻌䕿㻌 㻌䕿㻌 㻌䕿㻌 㻌
600 900 㻌 㻌 㻌 㻌 㻌䕿㻌 㻌䕿㻌 㻌䕿㻌 㻌䕿㻌
700 1000 㻌 㻌 㻌 㻌 㻌 㻌䕿㻌 㻌䕿㻌 㻌䕿㻌
Application of Steel Pipe Piles with Wings … 867

3 ATT construction process

ATT standard construction process is shown in Fig. 2. The standard process can be
divided into two main construction processes: soil cement column construction and
installation of steel pipe with spiral wings. These two both processes are normally
performed by only one construction machine. The standard construction process is
detailedly described as follows: firstly, 1) Position the center of the soil-cement mixer
to the pile core; 2) While discharging cement milk from the discharge hole at the end
of the mixer, drill the ground and mix. Upon reaching the predetermined depth, finish
mixing by the cement milk and re-temper the pile toe. After re-tempering, raise the
mixer while rotating it backwards and mix again; 3) After constructing the soil ce-
ment column, detach the mixer from the rotating shaft and attach the steel pipe rota-
tion jig to the rotating shaft and build in the steel pipe; 4) Position the center of the
end of the steel pipe with wings to the pile core and after checking that it is vertical,
bury it into the column center while rotating it; 5) After burying it to the predeter-
mined depth, cut off the pile head to finish burying of the steel pipe with wings.
The above described process allows pile foundation construction with generating
little surplus soil because the column is constructed by integrating ground at the site.
Concerning the amount of surplus soil generation, an investigation was performed for
38 ATTs (drilling depth 4.1m-25.4m) at 9 construction sites. The average rates of
surplus soil generation (defined as ratio of surplus soil amount to column volume)
were 4.3-18.6% at each site, making the overall average rate of only 11.1%. Further-
more, surrounding environment is also impacted less because ground water does not
become contaminated and no secondary pollution is generated. With low vibration
and noise, little impact on environment, ATT is considered as an environmentally
friendly construction method.

Rotating
shaft
ᅇ㌿㍈ Rotating
ᅇ㌿࢟ࣕࢵࣉ cap

Winged⩚᰿௜ࡁ㗰⟶
steel pipe

Mixing equipment
ᨩᢾΰྜ⿦⨨
Soil cement column
ᨵⰋయ

1) Pile core alignment 2) Mixing 3) Positioning of steel pipe pile 4) Rotational embedding 5) Embedding complete
with wings

Fig. 2. Standard construction process

4 Case study for ATT application in Vietnam

ATT was first applied in Vietnam for foundation pile of an 18-story building. 290 of
ATT plies (pile length L=6m; soil cement column diameter Dc=1,000 mm; spiral
868 P. H. Kien et al.

wing diameter D=700 mm; steel pipe diameter D0=219 mm) with design bearing ca-
pacity of 137.7 ton/pile were constructed. ATT was selected due to better cost per-
formance than other pile types. The application results are reported in this section.
4.1 Soil investigation
Construction site was located in 15 ward, Tan Binh district, Ho Chi Minh city, near
Tan Son Nhat international airport (Fig. 3). Soil investigation results are shown in
Fig. 4. The ground on which ATT piles are placed is clay with SPT N-value of 10-20.

Site

Tan Son Nhat

International Airport

Fig. 3. Construction site location Fig. 4. SPT at construction site

4.2 Loading test

As it is the first application of ATT in Vietnam, the loading test was performed to
confirm bearing capacity of the pile (Fig. 5). The loading test results are shown in Fig.
6. The loading test was stopped when the pile head load reached the value of 3,074
kN (the red P-įFXUYHLQ)LJ. This value is larger than two times of design bearing
capacity meaning that the value of design bearing capacity of 1,350 kN was con-
firmed. However, the P-įFXUYHFDQEHHVWLPDWHGDVWKHEODFNOLQHLQ)LJZLWKHVWL
mated ultimate bearing capacity of 5,860 kN and estimated design bearing capacity of
2,930 kN. The estimated design bearing capacity is 2.17 times larger than the value of
theoretically calculated design bearing capacity of 1,350 kN.
Application of Steel Pipe Piles with Wings … 869

Fig. 5. Loading test

Fig. 6. Loading test results

4.3 Construction of ATT foundation plie

Photo of construction site and design drawings are shown in Fig. 7 and Fig. 8. Large
size of excavator base machine was used so that construction time was shortened. The
steel pipe piles with wings were fabricated at a nearby factory and transported to con-
struction site by trucks. Nghi Son cement (PCB 40) was used for soil cement column.

Fig. 7. Photo of construction site

870 P. H. Kien et al.

Fig. 8. Design drawings

5 Conclusion

ATT is the foundation pile technology developed in Japan and transferred to Vietnam.
In this study, the first application of ATT in Vietnam was introduced. The loading test
was also presented. The test results show that ATT has sufficient performance and
bearing capacity for application in Vietnam. Vietnam is a country with a lot of soft
grounds and ATT might be a good alternative because it is a pile foundation devel-
oped as a friction pile with good cost performance.

References
1. Japanese Geotechnical Society: Pile vertical load test method and explanation First revised
edition, Japanese Geotechnical Society, pp 23-47 (2002).
2. Murakami, H., Ito, D. & Mizoguchi, E.: An application example of ATT Column construc-
tion method, The Foundation Engineering & Equipment, Vol. 36(10), pp 72-75 (2008).
3. Tamai, T., Ito, D. & Mizoguchi, E.: Study on the in-situ pile loading test and bearing ca-
pacity characteristics of steel pipe piles with wings installed in soil cement column, JGS
Journal, Vol. 4(4), pp 273-287 (2009).
 Modelling the hydromechanical behaviour of a granular
expansive clayey soil upon hydration using discrete
element method

Benjamin Darde1;2, Jean-Noël Roux1, Patrick Dangla1, Jean-Michel Pereira1, Anh

Minh Tang1, Jean Talandier2 and Minh Ngoc Vu2
1
Laboratoire Navier, Université Paris-Est, UMR 8205 (Ecole des Ponts ParisTech ± Ifsttar ±
CNRS), Marne-la-Vallée, France
2
French National Radioactive Waste Management Agency (Andra), Châtenay-Malabry, France

Abstract. Bentonite-based pellet materials are considered as a sealing material

for the isolation of galleries in the French radioactive waste disposal concept,
owing to operational convenience. The influence of the granular nature of the
material is studied through Discrete Element Method (DEM) simulations. Each
pellet is modelled individually and represented by a sphere of same mass and
density as real pellets. Swelling pressure tests of pellet mixtures, carried out at
laboratory scale, are simulated using a model describing the hydromechanical
behaviour of a single pellet upon suction decrease. The mixture behaviour is
satisfactorily reproduced upon hydration from 89 MPa (initial state) to 7 MPa
of suction. The model is then used to study the behaviour of large granular as-
semblies of bentonite pellets on the same hydration path. Results highlight that
the mixture assembling process, the pellet strength and stiffness, and the mix-
ture density have an influence on the swelling pressure development upon hy-
dration. Numerical results obtained through DEM simulations will be of interest
for future Finite Element Method simulations of the full hydration path using
double structure models where pellets correspond to the microstructural level.

Keywords: Pellet material, Swelling pressure, Discrete Element Method.

1 Introduction

Bentonite-based materials are considered as a sealing material for the isolation of

galleries in the French radioactive waste disposal concept because of their low perme-
ability, radionuclides retention capacity and ability to swell upon hydration. Within
this context, bentonite pellet mixtures have been studied owing to operational conven-
ience (Volckaert et al., 1996; Molinero Guerra et al., 2017). Pellets are installed in a
dry state as a granular assembly. The mixture homogenises upon hydration from the
host rock pore water. At full saturation, homogenised mixtures are thought to behave
as classic compacted bentonite materials. However, before homogenisation, the gran-
ular nature of the material controls the mixture macroscopic behaviour. Interaction

© Springer Nature Singapore Pte Ltd. 2020 871

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_139
872 B. Darde et al.

between pellets and its consequences on the mixture macroscopic behaviour has to be
studied to carry out predictive simulations of the evolution of engineered barriers.
In this work, the influence of the granular nature of the material is studied through
Discrete Element Method (DEM) simulations. In DEM simulations, each pellet is
modelled individually and represented by a sphere of same mass and density as real
pellets. Swelling pressure tests of pellet mixtures, carried out at laboratory scale, are
simulated using a model describing the hydromechanical behaviour of a single pellet
upon suction decrease. Then, the model is used to study the behaviour of large granu-
lar assemblies of bentonite pellets on the same hydration path.

2 Material and Methods

2.1 Expansive pellet

Pellets used in the present study are MX80 bentonite subspherical granules. The pellet
diameter is 7 mm. The initial dry density is 1.90 g/cm3 and initial suction is 89 MPa.

2.2 Swelling pressure test

Two swelling pressure tests (SP01 and SP02) are carried out on a pellet assembly in
cylindrical isochoric cells (Fig. 1). The cells height and diameter are 30 mm and
60 mm, respectively. A pressure sensor on the top wall is used to measure the appar-
ent swelling pressure upon hydration. The material dry density is 1.05 g/cm 3 (solid
fraction is ~0.552). Samples are prepared by placing pellets one by one in the cells, so
the cell can be closed with minimum initial pressure.
Hydration is performed through the vapour equilibrium technique (Tang and Cui,
2005). Imposed suction values are: 82; 59; 40; 38; 25; 13; 9 and 4 MPa. On imposing
a suction value, equilibrium is considered reached when pressure does not evolve for
several days. The equilibrium swelling pressure-suction relationship is then plotted.

Fig. 1. Isochoric cells used to perform swelling pressure tests.

2.3 Discrete element method simulations

In DEM simulations, each pellet is represented as a sphere of same mass and volume
as real pellets. The model proposed by Darde et al. (2018) for a single pellet is used to
describe the hydromechanical behaviour of a particle. Interactions at contacts between
particles and with the walls are described by the Hertz law and the Coulomb condition
Modelling the hydromechanical behaviour of a granular … 873

(Agnolin and Roux, 2007). Particle swelling is modelled as a radius increase. Pellet
strength is taken into account by considering perfect plasticity at contacts. Three types
of DEM simulations are carried out: DEM01 aims at simulating SP01 and SP02;
DEM02 aims at studying the influence of the sample preparation in swelling pressure
tests; DEM A, B, C and D are carried out on large granular assemblies (4000 parti-
cles) with periodic boundary conditions to avoid wall influence.
In DEM01 and DEM02, 209 pellets (as in experiments) are placed one by one un-
der gravity in a rigid cylinder. Cell closure is simulated by incrementally reducing the
cylinder height to the target height. The friction coefficient is set to 0 (DEM01)/0.3
(DEM02) during preparation and closure, and 0.3 during hydration. The swelling
pressure is computed as the sum of normal forces on the pressure sensor area on the
top wall, as in experiments. 100 DEM01 and DEM02 simulations are performed.
DEM A-D sample preparation and initial states have been described in Agnolin and
Roux (2007). Friction during hydration is also taken equal to 0.3. The initial solid
fraction of these samples varies from ~0.57 to ~0.64. The swelling pressure is com-
puted as the mean stress in the granular assembly. 4 simulations of each A-D type are
performed.

3 Results

3.1 Experimental and numerical results for swelling pressure tests

SP01 and SP02 test results are presented in Fig. 2 along with DEM results. After clo-
sure step, initial SP01 and SP02 pressure were 0.01 and 0.05 MPa, respectively. Upon
suction decrease from 89 MPa to 25 MPa, SP01 swelling pressure increased to 0.13
MPa. Swelling pressure increased slightly to 0.15 MPa as suction decreased to 9 MPa,
then slightly decreased to 0.14 MPa as suction decreased to 4 MPa. SP02 exhibited
WKHVDPH³LQFUHDVH-plateau-GHFUHDVH´WUHQGVZHOOLQJSUHVVXUHLQFUHDVHGWR03D
as suction decreased to 40 MPa, slightly increased to 0.18 MPa as suction decreased
to 25 MPa, then slightly decreased to 0.13 MPa as suction decreased to 9 MPa.

Fig. 2. SP01 and SP02 test results along with mean value of swelling pressure and plastic con-
tacts upon suction decrease in DEM01 and DEM02 simulations.

In DEM01 simulations, the peak swelling pressure is slightly overestimated but re-
sults also exhibited the same trend. Swelling pressure increased to 0.30 MPa as suc-
874 B. Darde et al.

tion decreased to 70 MPa, then decreased to reach a plateau at 0.11 MPa as suction
decreased to 7 MPa. The peak coincides with the development of plasticity at contacts
in the numerical samples. The post-peak decrease of swelling pressure is character-
ised by a continuous increase of the proportion of plastic contacts. As suction de-
creased to 7 MPa, more than 75 % of contacts had plastified.

3.2 Influence of the preparation at laboratory scale

Following preparation, the initial pressure was 0.50 MPa in DEM02 as a consequence
of friction mobilisation. Upon hydration, the development of swelling pressure in
DEM02 almost only decreased to reach same values as DEM01. In DEM02, contact
plasticity arose during preparation and kept increasing, whereas the first phase of
pellet swelling was mostly elastic in DEM01 (Fig. 2).

3.3 Large assemblies on partial hydration

DEM A-D simulations presented the same trend as DEM01: elastic increase, peak,
decrease of swelling pressure; along with a significant increase of plastic contact pro-
portion around the peak value. The peak swelling pressure was found to be higher for
denser samples. The post-peak decrease of swelling pressure was more significant in
denser samples. As suction reached 7 MPa, the swelling pressure was found to be
slightly higher in denser samples (Fig. 3). The proportion of plastic contacts, at peak
and at 7 MPa of suction, was found to be more significant in denser samples (Fig. 3).

Fig. 3. 3HDNDQG³03D´VZHOOLQJSUHVVXUH and proportion of plastic contacts for DEM A-D

and DEM01 simulations (mean values) as functions of the sample initial solid fractions.

4 Discussion

4.1 Hydration of expansive clay granules under constant-volume

conditions
The ability of DEM to access microscopic features of the samples allowed to charac-
terise the material behaviour. In constant volume conditions, hydration occurs in two
phases: First, an elastic increase of contact forces resulting in an increase of swelling
pressure, then, contact forces reach the elastic limit and a plateau/decrease of swelling
Modelling the hydromechanical behaviour of a granular … 875

pressure occurs. Using DEM highlighted the importance of the sample preparation at
laboratory scale. Initial pressure during the cell closure can result in contact plasticity
and strongly modify the macroscopic response.

4.2 Influence of pellet stiffness and strength

In the investigated suction range, the granular nature of the material was found to
control the material behaviour upon hydration. During hydration first phase, the
swelling pressure increase is mainly controlled by the pellet stiffness and volumetric
strain, the latter being related to pellet stiffness (Darde el al., 2018). Post-peak evolu-
tion behaviour is characterised by the increasing influence of contact plasticity. The
second phase is therefore controlled by pellet stiffness and strength decrease upon
suction decrease (Darde et al., 2018), as suggested by Hoffmann et al. (2007) and
Alonso et al. (2010).

4.3 Influence of the cell size and sensor diameter at laboratory scale
SP01 and SP02 samples are loose granular assemblies (initial solid fraction ~0.552).
A closure pressure should not be required for these samples. Thus, laboratory results
suggest that the material can behave as a denser material because of wall influence. It
is also suggested by comparing DEM A-D with DEM01: post-peak contact plasticity
increases significantly faster in DEM01 than in DEMA-D samples (Fig. 3).
The stress in DEM A-D (no wall) is found to be isotropic whereas the pressure is
generally higher on the lateral wall than on the top and bottom walls in DEM01 and
DEM02, which also highlights the influence of walls on the material behaviour.
Swelling pressure calculated using different sensor area (DEM01 samples) for the
same simulation results are presented in Fig. 4. Variability is found to significantly
increase for sensor diameter lower than 4 times the initial pellet diameter.
1,60
Coefficient of variation:

1,20
dimensionless

0,80

0,40

0,00
0 1 2 3 4 5 6
Diameter ratio: dimensionless

Fig. 4. Coefficient of variation of the calculated swelling pressure in DEM01 simulations as a

function of the sensor diameter to initial pellet diameter ratio.

5 Conclusion

The hydromechanical behaviour of expansive clay pellet material upon partial vapour
hydration was studied performing swelling pressure tests and using DEM. The granu-
876 B. Darde et al.

lar nature of the material was found to control the macroscopic response of the mate-
rial in the investigated suction range.
Upon suction decrease in constant volume conditions, the swelling pressure devel-
ops in two phases. First, an elastic increase subsequent to increase of normal contact
forces. This phase is mainly controlled by the particle stiffness. Second, a plat-
eau/decrease as a result of contact plasticity. This phase is mainly controlled by the
decrease of particle strength and stiffness upon suction decrease.
DEM can be regarded as a complementary tool to laboratory tests. It helped
providing an insight into the material behaviour, while avoiding experimental studies
inevitable constraints such as test duration, repeatability, cell size, sensors diameters.
Numerical results obtained through DEM simulations will be of interest for future
Finite Element Method simulations of the full hydration path using double structure
models where pellets correspond to the microstructural level.

References
Agnolin, I., & Roux, J. N. (2007). Internal states of model isotropic granular packings. I.
Assembling process, geometry, and contact networks. Physical Review E - Statistical,
Nonlinear, and Soft Matter Physics, 76(6), 1±27.
https://doi.org/10.1103/PhysRevE.76.061302
Alonso, E. E., Hoffmann, C., & Romero, E. (2010). Pellet mixtures in isolation barriers.
Journal of Rock Mechanics and Geotechnical Engineering, 2(1), 12±31.
https://doi.org/10.3724/SP.J.1235.2010.00012
Darde, B., Tang, A. M., Pereira, J.-M., Roux, J.-N., Dangla, P., Talandier, J., & Vu, M. N.
(2018). Hydro-mechanical behaviour of high-density bentonite pellet on partial hydration.
Géotechnique Letters, 8(4), 330±335. https://doi.org/10.1680/jgele.18.00114
Hoffmann, C., Alonso, E. E., & Romero, E. (2007). Hydro-mechanical behaviour of
bentonite pellet mixtures. Physics and Chemistry of the Earth, 32(8±14), 832±849.
https://doi.org/10.1016/j.pce.2006.04.037
Molinero Guerra, A., Mokni, N., Delage, P., Cui, Y. J., Tang, A. M., Aimedieu, P., Bornert,
M. (2017). In-depth characterisation of a mixture composed of pow-der/pellets MX80
bentonite. Applied Clay Science, 135, 538±546. https://doi.org/10.1016/j.clay.2016.10.030
Tang, A.-M., & Cui, Y.-J. (2005). Controlling suction by the vapour equilibrium technique
at different temperatures and its application in determining the water retention properties
of MX80 clay. Canadian Geotechnical Journal, 42(1), 287±296.
https://doi.org/10.1139/t04-082
Volckaert, G., Bernier, F., Alonso, E. E., Gens, A., Samper, J., Villar, M. V., Martin-Martin,
P. L., Cuevas, J., Campos, R., Thomas, H., Imbert, C. & Zingarelli, V. (1996). Thermal-
hydraulic-mechanical and geochemical behaviour of the clay barrier in radioactive waste
repositories (model development and validation), EUR 16744 EN. Luxembourg: Publica-
tions of the European Communities.
Spatially-varying non-stationary seismic bedrock motions

Van Nguyen Dinh1,a and Biswajit Basu2,b

1 MaREI Centre, University College Cork, Beaufort Building, Ringaskiddy, Co. Cork, Ireland
2
School of Engineering, Trinity College Dublin, Dublin 2, Ireland
a nguyen.dinh@ucc.ie (Correspondence) b basub@tcd.ie

Abstract. Seismic bedrock motions required for full soil-structure seismic anal-
ysis have not always been available and are simulated in this paper. The stochas-
tic bedrock Fourier spectrum used accounts for source spectrum, attenuation, ge-
ometrical spreading and source-to-bedrock amplification effects. Earthquake
magnitude, geometrical properties of the source and the bed, and geological pro-
file including the source and path rock are the input. The simulation is further
enhanced by authors’ new developments as follows. The frequency-dependent
source-to-bedrock amplification factor evaluated by a numerical scheme im-
proves the bedrock Fourier amplitude spectrum. The parametric modulating
functions obtained for a specific earthquake magnitude and source properties fa-
cilitate the spatiality and non-stationarity. The parametric forms of lagged coher-
ency compatible to specific bedrock sites also ensure the spatiality content. The
case study consists of three consecutive examples of the same geometry, where
the variation of source-to-bedrock amplification factors at several bedrock depths
is investigated. The formulated bedrock Fourier spectrum is then validated. The
spatially-varying non-stationary site-compatible bedrock motions are finally sim-
ulated where the differences between the averaged Fourier spectra of a realization
of 100 simulated accelerograms at two bedrock sites are observed.

Keywords: Seismic Bedrock Motions, Spatial Variation, Non-stationary, Para-

metric Coherency, Parametric Modulating Function.

1 Introduction

In seismic analysis of soil-structure interactions of both superstructures and under-

ground structures, the soil profile should be considered down to an extensively-thick
consolidated hard layer termed as the bedrock. As earthquake-induced surface ground
motions are obviously inadequate for such analyses, the motions at the bottom of the
soil profile (at the top level of the bedrock) i.e. the seismic bedrock motions, are needed
by a designer. However, seismic bedrock motions are hardly available from standard
recording equipment; their numerical simulation is therefore necessary.
As on the surface, on the bedrock level too, the amplitude and phase of the seismic
motions at one site are different from those at other sites even if the sites are not too
distant from each other. The differences denoted as the spatial variability are mainly
caused by travelling of seismic waves from the source to the bedrock sites through the
earth’s crust, where the primary waves of different frequencies travel with different

© Springer Nature Singapore Pte Ltd. 2020 877

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_140
878 V. N. Dinh and B. Basu

speeds and are mixed with secondary waves originating from reflection and refraction
through different soil layers. The spatial-variability of ground motions have pro-
nounced effects on response of various structures such as short bridges [1], two-span
bridges [2], long multi-span structures [3] [4], and complex structures [5].
Numerical simulation of spatially-varying seismic bedrock motions is presented in
this paper. It involves formulating the bedrock Fourier spectrum, evaluating the param-
eters of the amplitude modulating function and lagged coherency, and finally simulat-
ing the bedrock motions. It includes the originalities recently developed by the authors
[6]. Frequency-dependent source-to-bedrock amplification factor evaluated by the nu-
merical scheme [6] improves the bedrock Fourier amplitude spectrum. Parametric mod-
ulating functions obtained for a specific earthquake magnitude and source [6] facilitate
the spatiality and non-stationarity. Parametric forms of lagged coherency compatible to
specific bedrock sites are essential ingredients to ensure the spatiality.

2 Bedrock Fourier Spectrum Formulation

The stochastic seismic spectrum [7] is used as it can account for important factors such
as earthquake magnitude, source spectrum and depth, path-dependent loss of energy,
and source-to-site amplification. Its Fourier amplitude spectrum of bedrock motions is:
H rock Z  C ˜ E s Z , M  ˜ G R  ˜ P Z , R  ˜ AZ  ˜ DZ  (1)

where the source parameters are the scaling factor C and the source spectrum Es(Ȧ,M).
The path effects consist of geometrical spreading function G R  and path-dependent
loss of energy PZ , R  . The amplification factor A(Ȧ) is related to the source and the
path. The term D(Ȧ) is the diminution factor accounting for path-independent attenua-
tion of high-frequencies. The spectrum parameters are related to the moment magnitude
M as it is the best single measure of overall size of an earthquake and is not subject to
saturation. Details of the parameters in Eq. (1) can be found in [6] [8].
The amplification factor AZ  is approximated by the source-to-site impedance ra-
tio in a numerical scheme developed by the authors [6] using the quarter-wavelength
approximation method. In which, the deep rock is layered and the depth-dependent
shear velocity and density are evaluated. The shear velocity, density, and frequency
from bedrock level to current depth are then averaged. Amplification factors with une-
qually-spaced frequencies are finally respaced (and extrapolated if necessary) to facil-
itate multiplying with other equally-spaced spectra in the bedrock spectrum, Eq. (1).

3 Simulation of Spatially-varying Non-stationary Motions

Earthquake motions at n bedrock sites, f1(t), f2(t), …, fn(t), can be considered to be com-
ponents of the one-dimensional multi-variate (1D-mV) non-stationary zero-mean sto-
chastic vector process. The spectral representation based algorithm [9] is used in this
paper because it is a non FFT-based algorithm, thus the length of simulated time-series
is not constrained by the power-of-two. Windowing the stationary power spectrum by
Spatially-varying non-stationary seismic bedrock motions 879

multiplying with the modulating function and then transforming into the time-domain
also ensures that the long period components of the motion are not distorted. The diag-
onal and off-diagonal elements of the cross-spectra density (CSD) matrix are given by:
2
S 0jj Z , t  A j Z , t  S j Z , j = 1, 2, …, n (2)

S 0jk Z , t  A j Z , t Ak Z , t  S j Z S k Z J jk Z  , j, k = 1, 2, …, n; j z k (3)

where Sj( Z) is the stationary power spectral density at site j, and obtained from the
bedrock Fourier spectrum in Eq. (1) as S j Z  H rock Z  . The term Aj( Z, t) is the time-
2

and frequency-modulating function of fj(t) and Jjk( Z) is the complex coherency function
between fj(t) and fk(t). The CSD matrix S0(Z,t) is decomposed at every time instant t
under consideration as H Z , t H T* Z , t  . The non-stationary process fj(t) is simulated by
j N 1
> @
f j t  2 ¦ ¦ H jm Zl , t  'Z cos Zl t  T jm Zl , t   Iml , j = 1, 2, …, n
m 1l 0
(4)

where Zl l'Z , 'Z Z u N , Iml is the N sequences of independent random phase an-
gles, and Tjm is the phase angle. The term Zu is the upper cut-off frequency. The effects
of zero padding and the conservation of the power energy [8] [10] are accounted for.
A parametric form of the modulating function Aj( Z,t) has been developed by the
authors [6], in which a time-modulator AM(t) is used to represent the change in evolu-
tionary power spectrum, relative to the power spectrum as S Z , t  AM t  S 0 Z  . The
2

piece-wise continuous function [11] in Eq. (5), one of the most-cited amplitude modu-
lations, is chosen. Its parameters evaluated by using the assumption that the flat part of
the modulating function is equivalent to the strong motion duration [12] are in Eq. (6).
In which, T0 is the seismic motion duration contributed by both source and path.

­ A0 t t1  for 0 d t d t1
°
AM (t ) ® A0 for t1 d t d t 2 (5)
° A exp> ct  t @ for t d t
¯ 0 2 2

A0 0.9 T0 ; t1 0.15 A02 ; t2 t1  T0 ; c A0  1.5 A0 (6)

Parametric coherency functions for bedrock motions have not been reported so far
in the literature, and are estimated by the authors in [6] using the following framework:
a. Numerically model geological profile including bedrock and surface soil layers.
b. Excite the model by an acceleration time series at the assumed source.
c. Estimate non-parametric coherency of bedrock site pairs
d. Estimate parametric coherency models at the bedrock. The functional forms of
lagged coherency are fitted to the averaged smoothed non-parametric coherency.
In this paper, the nonparametric lagged coherency model introduced by Harichandran
and Vanmarcke [13], hereafter named the H-V 1986 model is used. This is one of the
most widely used empirical coherency models in the literature [12].
880 V. N. Dinh and B. Basu

4 Bedrock Amplification and Motion Spectra Case Study

This case study consists of three consecutive examples of the same geometry. First, the
source-to-bedrock amplification factors AZ  at some bedrock depths are investigated.
Properties of the source and bedrock are listed in Table 1. The deep rock is divided into
five groups: 'z0 = 0.001 km for two upper layers, 'z1 = 0.002 km for depth from 0.002
km to 0.2 km, 'z2 = 0.005 km for depth from 0.2 km to 0.5 km, 'z3 = 0.01 km for depth
from 0.5 km to 1.0 km, and 'z4 = 0.02 km for 1.0 km to the source depth he.
The AZ  values evaluated by using
the developed numerical scheme [6]
at different bedrock depths: 0.05 km,
0.1 km, 0.2 km, and 0.4 km relatively
to a source depth of 8 km are shown
in Fig. 1 where the bedrock-to-source
amplification increases with the fre-
quency and the shallower the bedrock
is, the higher is the rate of increase.
The amplification magnitude at shal-
lower depth is higher and decreases
with increasing depth.
Fig. 1. Bedrock-to-source amplification of source
at 8 km and various bedrock depths.

Table 1. Properties of source and bed rocks.

Property [unit] Source rock Bedrock

Elastic modulus [MN/m2] 70533.120 3000
Poison ratio 0.23 0.25
Density [kg-m/m3] 2800 2500
Damping ratio 0.02 0.05
Shear velocity [m/s] 3200 565.68

This second example validates the Fourier spectrum of seismic bedrock motions in Eq.
(1) by using a near-source spectrum for moment magnitude M = 7.5, where the
parameters for the scaling factor are Re = 0.55, Ve 1 2 | 0.71, Fs = 1.0, Z0 = 1 km,
Vs0 = 3.2 km/s and U = 2.8 g-m/cm3, for the value of the diminution factor N0 = 0.03.
Fig. 2 shows good agreement with the near-source spectrum using J97 model [7].
The last example aims at simulating spatially-varying non-stationary site-compati-
ble seismic bedrock motions by using the formulations presented in Sections 2 and 3.
The geometric data are: the depth of bedrock sites hbed = 0.1 km, the epi-centre distance
R0 = 20 km and the source depth he = 8.0 km. Properties of the source rock and bedrock
are the same as those in the first example (Table 1). The parameters for the scaling
factor C are in the example 2. An earthquake with moment magnitude M = 7.5 is con-
sidered and the Fourier spectra and time-histories of the accelerations at two bedrock
sites 1 (0, -100 m) and 2 (58.82, -100) have been simulated.
Spatially-varying non-stationary seismic bedrock motions 881

Fig. 2. Fourier amplitude spectra of near- Fig. 3. Stationary Fourier amplitude spectra
source accelerations at M = 7.5. of bedrock accelerations

The far-fault Fourier amplitude spectrum of acceleration obtained by using Eq. (1) is
shown in Fig. 3. The strong motion duration at the bedrock sites is T0 = 17.297 sec [6].
The parameters of the piece-wise continuous modulating function evaluated from T0
and by using Eq. (6) are A0 = 0.2281, t1 = 2.883 sec and c = 1.6A0. The parametric H-V
1986 coherency model estimated for the separation distance 58.82 m [6] is used.
Fig. 4 shows the averaged Fou-
rier amplitude spectra of accelera-
tions over 100 realisations at the two
bedrock sites by using the paramet-
ric H-V1986 coherency model esti-
mated in [6]. The considerable dif-
ferences in amplitude among these
averaged amplitude spectra can be
attributed to the wave passage (WP)
effects and arrival time perturba-
tions (ATP), which affect the lagged
coherency. The sharp variation at 1
Hz and 2 Hz can be attributed to the
frequency content of the input accel- Fig. 4. Ensembles of over 100 realisations of Fou-
eration used in step (b) of Section 3. rier amplitude of bedrock accelerations

5 Conclusions

Numerical simulation of spatially-varying site-compatible seismic bedrock motions has

been presented that involves formulating the bedrock Fourier spectrum, evaluating the
parameters of the amplitude modulating function and lagged coherency, and simulating
the bedrock motions. The originalities recently developed by the authors including fre-
quency-dependent source-to-bedrock amplification factors, parametric modulating
functions, and parametric forms of bedrock lagged coherency have been introduced in
the simulation. Three consecutive examples of the case study have shown the variation
of source-to-bedrock amplification factors with respect to bedrock depths, validated the
882 V. N. Dinh and B. Basu

bedrock Fourier spectrum, and simulated spatially-varying non-stationary site-compat-

ible bedrock motions. The spatial variability and differences between the averaged Fou-
rier spectra of the accelerogram ensembles at two bedrock sites have been observed.

Acknowledgements

This research was partially funded under the EU FP7 Marie Curie IAPP project NOTES
(Grant No. PIAP-GA- 2008-230663). The first author is funded by Science Foundation
Ireland (SFI) Research Centre: MaREI - Centre for Marine and Renewable Energy.

References
1. T. E. Price, M. O. Eberhard: Effects of spatially varying ground motions on short bridges,”
Journal of Structural Engineering (ASCE) 124(8), 948-955 (1998).
2. V. N. Dinh, B. Basu: Monitoring of flexible support effects using the responses of multi-
span bridges via wavelet transform. In Proceedings of The 6th European Workshop on Struc-
tural Health Monitoring, Dresden, Germany (2012).
3. A. Chakraborty, B. Basu: Nonstationary response analysis of long span bridges under spa-
tially varying differential support motions using continuous wavelet transform. Journal of
Engineering Mechanics (ASCE) 134(2), 155-62 (2008).
4. V. N. Dinh, B. Basu, R. B. J. Brinkgreve: Wavelet-based evolutionary response of multi-
span structures including wave-passage and site-response effects. Journal of Engineering
Mechanics 140(8) (2014).
5. N. J. Burdette, A. S. Elnashai, A. Lupoi, A. G. Sextos: Effect of asynchronous earthquake
motion on complex bridges. I: Methodology and input motion. Journal of Bridge Engineer-
ing 13(2), 158-165 (2008).
6. V. N. Dinh, B. Basu: Development of a simulation algorithm for spatially-varying non-sta-
tionary bedrock earthquake motions. Manuscript to be submitted to CIGOS 2019 Special
Issue, Computer-Aided Civil and Infrastructure Engineering (2019).
7. D. M. Boore: Simulation of ground motion using the stochastic method. Pure and Applied
Geophysics 160, 635–676 (2003).
8. V. N. Dinh, B. Basu, R. B. J. Brinkgreve: Simulation and application of spatially-varying
non-stationary subsurface motions. In Proceedings of The Fifth Asian-Pacific Symposium
on Structural Reliability and its Applications, Singapore (2012).
9. G. Deodatis: Non-stationary stochastic vector processes: Seismic ground motion applica-
tions. Probabilistic Engineering Mechanics 11, 149-168 (1996).
10. V. N. Dinh, B. Basu: Zero-pad effects on conditional simulation and application of spatially-
varying earthquake motions. In Proceedings of The 6th European Workshop on Structural
Health Monitoring, Dresden, Germany (2012).
11. M. Amin, A. H. S. Ang: Non-stationary stochastic model of earthquake motions,” Journal
of Engineering Mechanics (ASCE) 94(2), 559-583 (1968).
12. A. Zerva: Spatial Variation of Seismic Ground Motions: Modelling and Engineering Appli-
cations, Boca Raton: Taylor & Francis Group LLC (2009).
13. R. S. Harichandran, E. H. Vanmarcke: Stochastic variation of earthquake ground motion in
space and time. Journal of Engineering Mechanics Division (ASCE) 112(2), 154–174 (1986)
 An experimental investigation on bearing capacity
behavior of rectangular footing over reinforced soil slope

Raju Sarkar1, Amit Kumar Shrivastava2, and Ankur Mudgal2

1
Centre for Disaster Risk Reduction and Community Development Studies, Royal University of
Bhutan, Rinchending, Bhutan
2
Department of Civil Engineering, Delhi Technological University, Delhi, India
rajusarkar.cst@rub.edu.bt

Abstract. Various Geotechnical engineering works demand the construction of founda-

tion over cohesive soil slope. The construction of foundation over these soils may cause
higher settlement which can obliterate the overlying structure. On the other hand, the use
of reinforcement techniques in the soil stability is also increasing rapidly. Amongst
them, the use of geosynthetics to reinforce the foundation soil is one of the cost-effective
options. In the present study, a number of model footing tests were performed in labora-
tory to check the effect of geosynthetics in the development of bearing capacity on shal-
low foundation resting over cohesive soil slope. For this purpose, a bi-oriented geo-grid
was chosen as a reinforcement material. Parametric studies such as effective depth of re-
inforcement, number of reinforcement layers, were considered in the present study. A
number of laboratory tests were conducted, and their results were analyzed. From the
analysis it was observed that soil reinforced with geosynthetics overlain by rectangular
footing showed substantial increase in ultimate bearing capacity of the reinforced soil,
thus signifying the potential benefits of reinforcements in soil.

Keywords: Soil slope; geogrid reinforcement; bearing capacity.

1 Introduction

The construction of shallow foundation over clayey soil slope is very difficult due to its poor
load carrying capacity and excessive settlement. The conventional method to increase the
bearing capacity of soil is replacement of existing cohesive soil with strong granular fill. This
replacement of cohesive soil layer by granular fill requires more laborious works and uneco-
nomical from the construction point of view. Several laboratory model footing tests, numeri-
cal studies and large-scale tests have been conducted to analyse the behaviour of reinforced
soil shallow foundation. [1-6]. [2] studied the bearing capacity of geogrid and geotextile rein-
forced earth slab. They investigated that bearing capacity improvement is not significant when
reinforcement provided after the optimum number of reinforcement layers. [3] conducted the
laboratory model footing tests on strip foundation over clayey slope reinforced with
geosynthetics. The soil slope was varied from 35º to 50º. Geosynthetic was placed at a differ-
ent location to find out the optimums in layout. Based on the laboratory tests results they
suggested that the first depth of reinforcement should be located at 0.4B from the footing base

© Springer Nature Singapore Pte Ltd. 2020 883

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_141
884 R. Sarkar et al.

to achieve the maximum bearing capacity. The major emphasis of the research is towards the
estimating the improved bearing capacity of foundation being construction over slopes, with
the application of geogrid under different working conditions. For this purpose, small scale
model footing tests have been conducted to analyze the performance of geogrid in bolstering
the bearing capacity of shallow foundation constructed over cohesive soil slopes and reducing
the percentage settlement induced due to weak soil strength.

2 Material used

2.1 Soil and geosynthetic

In the present study, A silty clay was used to prepare the foundation bed. Soil sample has been
collected from Gwalior, India. The geotechnical properties of soil are presented in Table 1.
Based on the unified soil classification system (USCS), the soil is classified as low compress-
ible soil. On the other hand, A bi-oriented geogrid was chosen as a reinforcement material.
The Mechanical properties of geosynthetics in both directions i.e. machine direction (MD)
and cross direction (CD) are presented in Table 2.

Table 1. Geotechnical properties of soil

Properties Values
Specific Gravity 2.62
Liquid limit (%) 34.0
Plasticity index 15
Plastic limit (%) 19
Maximum dry unit weight (kN/m3) 17.1
Optimum Moisture Content (%) 15

Table 2. Mechanical properties of geosynthetics (6RXUFH6XSSOLHU¶VGDWD

Parameters Polypropylene Geogrid

Tensile Strength (kN/m2) 115 MD 115 CD
Stiffness at 0.5% Strain (kN/m2) 550 MD 350 CD
Aperture Size (mm) 30×30

3 Methodology

3.1 Slope preparation

The test bed was prepared in a steel box of size length = 750 mm, width = 450 mm and height
= 600 mm. The soil was compacted at its maximum dry density, so a predetermined water
content was added into the soil. The slope was constructed in three lifts in case of unrein-
forced soil slope whereas for reinforced soil slope the thickness of each lift was dependent on
An experimental investigation on bearing capacity … 885

the number of reinforcement layers. After the compaction of each layer, the top surface of the
layer was levelled horizontally with the help of spirit level so that uniformity of layer could be
maintained.

3.2 Testing procedure

The aim of the study was to check the influence of slope angle, edge distance (D) and the
optimum depth of first reinforcement (u) to achieve the maximum possible benefit from the
reinforcements. The tests for u/B were conducted for corresponding values at 0.2, 0.4, 0.6, 0.8
and 1 respectively to find out the optimum of first reinforcement depth. The tests for u/B and
N were conducting by fixing the slope angles at 45º and at D/B = 1. The effect of slope geom-
etry and edge distance were analyzed for geogrid reinforced soil, by varying the slope angles
by 35º, 40º and 45º respectively and fixing the value of D/B = 1.

3.3 Experimental testing setup

A number of small-scale model footing tests were carried out on the geogrid reinforced soil
bed in the laboratory. A square footing of dimension of 75 mm × 75 mm was used as the
model footing. A geometric diagram of slope model is shown in Fig.1. The foundation of
clayey slope was used with a dimension of 750 mm × 450 mm ×200 mm and a slope of height
300 mm was prepared over the foundation at different slope angles. The sides of the steel box
were made of steel sheet and an acrylic sheet was used at front side of the box for the visuali-
zation of failure pattern of soil slope. The tank was tested in loading frame consisted of a plat-
en attached to the cross head of the loading frame. A load cell was used to measure the load
values and two dial gauges were used to measure the footing settlement in the experimental
study.

Fig. 1. Geometric view of reinforced soil slope

886 R. Sarkar et al.

4 Results and discussion

4.1 Effect of topmost reinforcement depth (u)

To find the value of optimum depth of top layer of geogrid, a series of number of tests were
conducted. Bearing capacity ratio (BCR) was determined for each settlement ratio which is
defined as the ratio of the bearing pressure of a reinforced soil to that of an unreinforced soil,
when evaluated at the same settlement ratio (Sr = 4%, 8%, 12%, 16% and ultimate bearing
capacity UBC). Fig. 2 (a-b) show the BCR and pressure settlement graphs of soil slope rein-
forced with geogrid. It is clearly seen from the graph that maximum value of bearing pressure
obtained at u/B= 0.40 and beyond this depth a decrease in bearing capacity was found as rein-
forcement depth increases. The probable reason behind this is that any further increment in
reinforcement depth would deteriorate the possible benefits of reinforcement.

(a) (b)

Fig. 2. (a) Bearing capacity ratio versus u/B for Geogrid (b) Pressure Settlement Curve for Geogrid at
different u/B ratios

4.2 Effect of number of reinforcement layers (N)

To check the effect of number of layers of geogrid, the first layer of reinforcement was fixed
at u/B then number of reinforcement layers were varied with a spacing of 0.4B (optimum
depth of top layer) up to those layers where significant improvement in bearing capacity im-
provement become insignificant. Fig. 3. Shows the relation between the number of reinforce-
ment layers and BCR for different settlement ratios. From the Fig 3. It can be seen that be-
yond the number of reinforcement layers equal to 4 the improvement in bearing capacity is
almost constant. The reason for this can be attributed that at N=4, reinforce depth located
beyond the pressure bulb of footing.
An experimental investigation on bearing capacity … 887

Fig. 3. Bearing capacity ratio versus Number of reinforcement layers (N) for Geogrid

4.3 Influence of slope angOHȕRQXQUHLQIRUFHGVORSH

In the laboratory, different soil slope models were made and then tested to check the effect of
slope angles on the bearing capacity of soil slope. Fig 4. (a-c) Show the BCR vs u/B curves
for different slope angles and s/B ratios at D/B=1. From the graph it is observed that with
increasing the slope angles the bearing capacity decreases and the maximum improvement in
EHDULQJFDSDFLW\ZDVIRXQGZKHQDWȕ º. The probable reason can be explained that as slope
angle decreases the stability of slope increases thus more improvement in bearing capacity can
be expected.

(a) (b)
888 R. Sarkar et al.

(c)

Fig. 4. Bearing capacity ratio curves for (Dȕ º (b) 40º (c) 45º

5 Conclusion
Based on the experimental work following conclusions were drawn from the study.
(1) The optimum depth of top layer reinforcement for geogrid was obtained at u/B = 0.4 and
maximum number of reinforcement layers were observed at N = 4.
(2) 0D[LPXPEHDULQJFDSDFLW\ZDVDFKLHYHGDWȕ ºZLWKUHVSHFWWRȕ º and 45º.
(3) Geogrid performance is very significant at lower settlement (e.g. s/B=4%) which is very
useful reinforcement in shallow foundation work.

References
1. Binquet, J., & Lee, K. L. (1975(b)). Bearing capacity tests on reinforced earth slabs. J.Geotech.
Geoenviron. Eng., 101(ASCE# 11792 Proceeding).
2. Guido, V. A., Chang, D. K., & Sweeney, M. A. (1986). Comparison of geogrid and geotextile
reinforced earth slabs. Canadian Geotechnical Journal, 23(4), 435-440.
3. Shin, E. C., & Das, B. M. (1998). Ultimate bearing capacity of strip foundation on geogrid-
reinforced clay slope. KSCE Journal of Civil Engineering, 2(4), 481-488.
4. Yoo, C., (2001) Laboratory investigation of bearing capacity behavior of strip footing on
geogrid-reinforced sand slope, Geotextiles and Geomembranes, (19) 279±298.
5. Chen, Q., Abu-Farsakh, M., Sharma, R., & Zhang, X. (2007). Laboratory investigation of be-
havior of foundations on geosynthetic-reinforced clayey soil. Transportation Research Record:
Journal of the Transportation Research Board, (2004), 28-38.
7. Latha G. M., & Somwanshi A., (2009). Bearing capacity of square footings on geosynthetic re-
inforced sand, (27), 281±294.
8. 6DZZDI0³([SHULPHQWDODQGQXPHULFDOVWXG\RIHFFHQWULFDOO\ORDGHGVWULSIRRWLQJV
UHVWLQJRQUHLQIRUFHGVDQG´-*HRWHFK*HRHQYLURQ(QJ-1518
 Meshfree SPH modelling of shrinkage induced cracking
in clayey soils

Hieu T. Tran1, Nhu H.T. Nguyen1, Giang D. Nguyen2 and Ha H. Bui1*

1
Department of Civil Engineering, Monash University, VIC 3800, Australia
2
School of Civil, Environmental and Mining Engineering, The University of Adelaide, SA
5005, Australia
ha.bui@monash.edu

Abstract. In this study, a computational approach that combines the mesh-free

Smoothed Particle Hydrodynamics (SPH) and a simple anisotropic damage
constitutive model is proposed to model shrinkage induced cracking in clayey
soils. To assess its performance, a numerical soil shrinkage test is conducted
and simulation results are compared with experimental data. Numerical results
show that the proposed approach can capture complicated cracking patterns
with multiple fracture networks in clayey soils, which demonstrates the poten-
tial of SPH for simulating desiccation cracking in these materials.

Keywords: SPH, desiccation cracking, clayey soil.

1 Introduction

Desiccation cracking in clayey soils is one of the major concerns in geotechnical and
geo-environment engineering. The presence of desiccation cracks in soils weakens
their strength and hence is the source of many problems such as damages in soil
support structures and shallow-buried structures (e.g. pavement and pipelines damag-
es), slope failure and landslides. Desiccation cracks occur due to volumetric shrinkage
of soil mass induced by moisture evaporation [1]. The cracking process is complicat-
ed, which involves multiple cracks evolution in multiphysical domain. This makes it a
challenging issue to be modelled and characterised by numerical methods. So far,
several methods have been proposed to model desiccation cracks in soils. One of the
most common methods is based on the finite element method (FEM) (e.g. [2]). The
use of FEM, however, poses a fundamental difficulty in modelling random crack
initiation and propagation for a mesh system representing a soil domain. In this sense,
the discrete approach appears to be more suitable as it allows cracks to naturally de-
velop following a contact strength criterion [3-6]. This approach, however, requires
calibration of model parameters, which would limit its predictive capability for soil
desiccation cracking as many unknown variables will need to be calibrated in such a
multi-physical process.
This paper examines the potential of the Smooth Particle Hydrodynamics (SPH)
method to simulate shrinkage induced cracks in clayey soils, which is a primary me-

© Springer Nature Singapore Pte Ltd. 2020 889

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_142
890 H. T. Tran et al.

chanical process in soil desiccation cracking. SPH is considered a suitable method to

simulate soil cracking as it allows cracks to naturally develop through particle separa-
tion while being available to combine with well-established continuum models for
soils. SPH has been used in a wide range of geotechnical applications such as granular
flows [7], soil-structure interaction [8-9], localised failure [10-13], seepage flow [14]
and soil cracking [15-17]. To facilitate the modelling of soil desiccation cracking, an
anisotropic damage model is proposed and implemented into SPH. A simulation of
shrinkage induced cracks in clayey soils is then conducted and numerical results are
compared with experimental data to assess the performance of this approach.

2 Modelling approach

In the proposed approach, SPH method is adopted as the numerical framework to

simulate soil cracking. Detailed formulations of the SPH method have been well doc-
umented in [18]. Herein, only key idea and formulations of this method are explained.
In SPH, a material domain is discretised into a set of particles, each of which possess
its own properties (e.g. density, stress and strain) and moving in accordance with the
momentum conservation law, which is expressed in the following form [18]:

dvi
N §ı ıj ·
dt ¦m ¨¨ Ui
i
2

U 2j
 Cij ¸ ’Wij  g
¸
(1)
j 1 © i ¹
where i and j denote the particle under consideration and its neighbouring particle,
respectively, v is the velocity vector, m is the mass of the particle, U is the density, V
are the stress tensor, Wij is the kernel function taken to be the cubic spline function
[18], Cij the stabilisation term employed to remove the numerical instability problems,
and g the gravitational force vector. To close the above governing equation, a consti-
tutive model which defines the relationship between stress and strain tensors is re-
quired. In this work, a simple constitutive model for soil cracking is proposed,
adopting the concept of kinematically enhanced model for unidirectional materials
and cohesive zone model [19]. The stress-strain relation is defined for principal frac-
ture planes and consists of a linear elastic part representing the undamaged stage of
material followed by a nonlinear softening part representing the material response
when damage develops. The transition from elastic to damage stage occurs when the
strain reaches a critical value H at . The damage variable associated with these two
states is defined as follows:

­° 0 if H ad d H at
da ® (2)
JH d
°̄1  e a if H ad ! H at

where da is the damage variable in a principal direction (i.e. a : 1, 2 or 3) of a par-

ticle, H ad the strain of particle in direction, andJ the fracture energy parameter.
Meshfree SPH modelling of shrinkage induced cracking … 891

Accordingly, the stress-strain rate relationships for three principal directions are
then obtained based on the generalised +RRNH¶VODZDVIROORZV
்
ߪሶଵ ሺͳ െ ݀ଵ ሻ ͵‫ ܭ‬൅ ʹ‫ ܭ ܩ‬െ ʹ‫ܩ‬Ȁ͵ ‫ ܭ‬െ ʹ‫ܩ‬Ȁ͵ ߳ሶଵ
൥ߪሶଷ ൩ ൌ ቎ሺͳ െ ݀ଷ ሻ቏ ൭൥‫ ܭ‬െ ʹ‫ܩ‬Ȁ͵ ͵‫ ܭ‬൅ ʹ‫ ܭ ܩ‬െ ʹ‫ܩ‬Ȁ͵൩ ൥߳ሶଶ ൩൱ (3)
ߪሶଶ ሺͳ െ ݀ଶ ሻ ‫ ܭ‬െ ʹ‫ܩ‬Ȁ͵ ‫ ܭ‬െ ʹ‫ܩ‬Ȁ͵ ͵‫ ܭ‬൅ ʹ‫߳ ܩ‬ሶଷ
where K and G are the bulk and shear modulus, respectively. Once the stress-strain
response in principal directions is known, the material response on the global coordi-
nate system is achieved by rotating the response to the global coordinate system.

3 Performance of the constitutive model

An element test is first simulated to illustrate the model response at the constitutive
level. Fig. 1 shows the setup of the numerical model. The following material parame-
ters are chosen for this simulation: Young¶V modulus E = 4 MPa, PoisVRQ¶VUDWLRv =
0.15 and critical strain in three principal directions ߝ௔௧ ൌ ͲǤͲͲͲʹ. The fracture energy
parameter J is varied from 50 to 80 and 100 to examinate its effect on the stress-
strain relationship. The specimen is then subject to strain rate dHxx = 3.0e-6 in x-
direction, dHyy = -1.0e-6 in y-direction and dHzz = -2.0e-6 in z-direction. Fig. 1 typically
shows the stress-strain behaviour of the element in the x-direction. In response to the
tensile strain applied in the x-direction, the tensile stress is shown to linearly develop
in this direction until the strain reaches its critical value of 0.0002. Subsequently,
damage evolves and strain softening occurs. The model is shown to properly respond
to the development of damage with no oscillation. Moreover, the fracture energy of
the element defined as the area below the stress-strain curve is shown to change cor-
responding to different values of J. In fact, J represents the fracture property of soils
which provides flexibility for this model to capture the cracking behaviour of differ-
ent clayey soils.

2000
Gamma 50
1800
Gamma 80
1600 Gamma 100

1400
Stress, ߪ (Pa)

1200

1000

800

600

400

200

0
0 0.0002 0.0004 0.0006 0.0008 0.001 0.0012
Strain ߳
Fig. 1. Numerical setup of the element test and the stress-strain results in x-direction.
892 H. T. Tran et al.

4 Soil cracking simulation

In this section, the above anisotropic damage model is used to simulate soil cracking
problems. The desiccation cracking experiment of metallurgical waste conducted by
Rodríguez [20] is chosen for the simulation. Fig.2 shows the geometry of the speci-
men. The specimen has dimensions of D = 64mm in diameter and H = 4mm in thick-
ness. A total of 99,848 SPH particles are used to represent the soil domain while
41,007 particles are used to model the boundary. Boundary conditions are fully re-
strained at the bottom surface of the sample and free in all other surfaces. The bottom
boundary is modelled by three layers of boundary particles with a fully fixed bounda-
ry condition [17]. Physical properties of these boundary particles are calculated by
extrapolating properties of nearby soil particles. More details of this simulation setup
and the extrapolation technique can be found in [17]. The material properties are as
IROORZV<RXQJ¶VPRGXOXVE = 4.0 03D3RLVVRQ¶Vratio v = 0.2, tensile strength Vt =
4.0 kPa and fracture energy GI = 0.4 N/m.
To simulate the shrinkage process, this work follows the approach used in [17],
which was based on the experimental evidence reported by Rodríguez [20]. In par-
ticular, a constant deformation/strain rate is applied to the sample until reaching a
maximum of 5% contraction. The stress of particles is then obtained from strain based
on the constitutive law. This is achieved in the numerical simulation by gradually
applying shrinkage stresses to soil particles. To consider the geometrical effect, the
volumetric shrinkage strain is applied to all SPH particles with the shrinkage magni-
tude gradually decreases from the top surface (2%) to the bottom surface (1%) of the
soil sample. This selection of shrinkage strain is reasonable in the sense that moisture
evaporation on the top surface is always larger than that at the bottom surface.
Fig. 3 shows a comparison between the SPH simulation and experiment for the fi-
nal cracking pattern of the soil sample. Cracks are captured in the simulation through
particle separation when da reaches the critical value. It can be seen that the compli-
cated cracking networks of clayey soil observed in the experiment can be qualitatively
captured by the proposed method. Differences in crack network may come from the
assumption of homogeneous material in the simulation which ignores the inhomoge-
neity of soil sample in experiments. Though the cracking networks predicted by the
simulation is not exactly the same as that in the experiment, the proposed SPH
framework demonstrates its capability in describing the complex fracture networks
occurred in soils without any ad-hoc treatment on SPH particles or using predefined
cracks. This is attributed to the attractive feature of the proposed framework that al-
lows the fracture to be developed in any direction depending on the loading condition.

Soil D

Fixed boundary
Fig. 2. Geometry and boundary condition of soil cracking test in accordance with [20].
Meshfree SPH modelling of shrinkage induced cracking … 893

(a) (b)

Fig. 3. Cracking patterns of soil sample obtained from (a) the SPH simulation and (b) the ex-
periment by Rodríguez, et al. [20].

5 Conclusions

In this paper, the computational framework that combined the advantages of the
mesh-free SPH method and a kinematically enhancement unidirectional constitutive
model capble of capturing anistropic damage of materials is proposed to simulate
shrinkage induced cracking in soils. Numerical applications show that the proposed
framework is capable of reproducing the complex fracture behaviour of clayey soils
subjected to shrinkage without any numerical treatment. This demonstrates the poten-
tial of SPH to model soil cracking problems. However, more quantitative comparisons
of shrinkage cracking between the experiment and simulation would be needed to
accurately assess the performance of the proposed approach. Besides, further devel-
opments are required to describe the fully coupled thermo-hydro-mechanical behav-
iour of soils and these works are in progress.

References

1. Tran K M, Bui H H, Kodikara J, Sánchez M: Soil curling process and its influencing factors
Canadian Geotechnical Journal (2019) https://doi.org/10.1139/cgj-2018-0489
2. Vo T D, Pouya A, Hemmati S, Tang A M: Numerical modelling of desiccation cracking of
clayey soil using a cohesive fracture method. Comp. Geotech. 85: 15-27 (2017)
3. Nguyen N H T, Bui H H, Nguyen G D, Kodikara J, Arooran S, Jitsangiam P: A
thermodynamics-based cohesive model for discrete element modelling of fracture in
cemented materials International Journal of Solids and Structures 117 159-176 (2017)
4. Tran, K M, Bui, H H, Kodikara J, Sánchez M: Numerical simulation of soil curling during
desiccation process In The 7th International Conference on Unsaturated Soils, Hong Kong
(2018)
894 H. T. Tran et al.

5. Nguyen T T, Bui H H, Ngo T D, Nguyen G D,: Discrete Element Modelling of the

Mechanical Behaviour of a Highly Porous Foamed Concrete. Poromechanics VI. pp. 1380-
1387 (2017)
6. Nguyen N H T, Bui H H, Nguyen G D, Arooran S, Kodikara J: Numerical Study of Particle
Size Distribution Effect on the Failure of Asphalt Mixtures Using Discrete Element
Method. Poromechanics VI. pp. 1371-1379 (2017)
7. Nguyen C T, Nguyen C T, Bui H H, Nguyen G D, Fukagawa R: A new SPH-based
approach to simulation of granular flows using viscous damping and stress regularisation
Landslides 14(1): 1-13 (2016)
8. Bui H H, Fukagawa R: An improved SPH method for saturated soils and its application to
investigate the mechanisms of embankment failure: Case of hydrostatic pore-water pressure
Int J Numer Anal Methods Geomech. 37(1) 31-50 (2013)
9. Bui H H, Kodikara J, Bouazza A, Haque A, Ranjith P G: A novel computational approach
for large deformation and post-failure analyses of segmental retaining wall systems Int. Int
J Numer Anal Methods Geomech. 38(13) 1321-40 (2014)
10. Wang Y, Bui H H, Nguyen G D, PG Ranjith: A Mesh-Free Continuum Based
Computational Approach to Modelling Rock Fracture. Poromechanics VI. pp. 2041-2048
(2017)
11. Zhao S, Bui H H, Lemiale V, Nguyen G D: SPH Simulation of Strain Localisation in
Geomaterials Using a Visco-Plastic Constitutive Model. Poromechanics VI. pp. 1876-1883
(2017)
12. Wang Y, Bui H H, Nguyen G D, Ranjith P G: A new SPH-based continuum framework
with an embedded fracture process zone for modelling rock fracture. Int J Solids Struct. 159
40-57 (2019)
13. Zhao S, Bui H H, Lemiale V, Nguyen G D, Darve F: A generic approach to modelling
flexible confined boundary conditions in SPH and its application. Int J Numer Anal
Methods Geomech. 43: 1005-1031 (2019)
14. Bui H H, Nguyen G D: A coupled fluid-solid SPH approach to modelling flow through
deformable porous media. IntҿQDWLRQDO Journal of Solids and Structures 125 244-64 (2017)
15. Bui H H, Nguyen G D, Kodikara J, Sanchez M: Soil cracking modelling using the mesh-
free SPH method 12th Australia New Zealand Conference on Geomechanics (2015)
16. Tran H T, Bui H H, Nguyen G D, Kodikara J, Sanchez M: A Continuum Based Approach
to Modelling Tensile Cracks in Soils. Poromechanics VI. pp. 337-344 (2017)
17. Tran H T, Wang Y, Nguyen G D, Kodikara J, Sanchez M, Bui H H: Modelling 3D
desiccation cracking in clayey soils using a size-dependent SPH computational approach.
Computers and Geotechnics, accepted, (2019)
18. Monaghan J J, Lattanzio J C: A refined particle method for astrophysical problems
Astronomy and Astrophysics. 149: 135-43 (1985)
19. Nguyen N H T, Bui H H, Nguyen G D, Kodikara J: A cohesive damage-plasticity model for
DEM and its application for numerical investigation of soft rock fracture properties
International Journal of Plasticity. 98: 175-96 (2017)
20. Rodríguez R: Hydrogeotechnical characterization of a metallurgical waste Canadian
Geotechnical Journal. 43(10): 1042-60 (2006)
Numerical predictions of post-flow behaviour of granular
materials using an improved SPH model

H.H. Bui1* and Giang D. Nguyen2

1
Department of Civil Engineering, Monash University, Australia
2
Schoool of Civil, Enviromental & Mining Egnineering, The University of Adelaide, Australia
ha.bui@monash.edu

Abstract. This paper presents the recent development of the smoothed particle
hydrodynamics (SPH) method for accurate predictions of granular flows. Gran-
ular materials are described within the classical plasticity theory framework,
while the large deformation and flow behaviour of the materials are simulated
by the mesh-free SPH. To improve the accuracy of SPH for the post-flow pre-
diction at which stress fluctuation under large shear deformation is usually ob-
served, a stress regularisation technique recently proposed by the authors is
adopted. Through several numerical validations with experiments under 2D and
3D conditions, the proposed SPH model shows significant improvements in the
accuracy and stability of SPH not only for simulations of granular flows, but al-
so for general applications for solid materials.

Keywords: SPH, granular materials, granular flows, plasticity, post-failure.

1 Introduction
The flow of granular materials is commonly observed in engineering applications
such as the transport of minerals, powder, or cereals and during geophysical events
such as landslides and debris flow. Therefore, understanding the mechanisms of gran-
ular flows will help optimise industrial processes and to mitigate risks caused by natu-
ral disasters. In this regard, development of a robust and efficient numerical approach
capable of accurately capturing the post-flow behaviour of granular materials would
significantly benefit engineering designs and industrial solutions.
Numerous research have been undertaken to simulate granular materials. Among
different approaches, the Discrete Element Method (DEM) [1] is perhaps the most
extensively used numerical method for predictions of granular materials [2, 3]. Alt-
hough DEM appears to be an ideal method for the simulation of granular flows, the
computational resource required in large scale simulations usually involves several
millions of particles and thus is demanding. As a much computationally cheaper al-
ternative, continuum based numerical approaches are usually adopted for large scale
simulations. Among these approaches, SPH has recently been successfully applied to
simulate granular flows by the authors [4-8] and showed to be a powerful alternative
to deal with the large deformation and post-failure of geomaterials [9-14, 23-25].
However, the standard SPH model suffers from the so-called short-length-scale-noise

© Springer Nature Singapore Pte Ltd. 2020 895

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_143
896 H. H. Bui and G. D. Nguyen

[6], which results in high stress fluctuation for particles undergoing large shear de-
formation. To address this issue, this paper presents a new approach, which makes use
of the stress regularisation technique [8]. Through both 2D and 3D numerical valida-
tions with experiments, the the proposed technique is shown to effectively remove the
inherent short-length-scale-noise problem associated with the standard SPH method.
The excellent agreement with experiments demonstrates the success of the proposed
approach, and serves as a basis for further improvements of SPH-based solutions for a
wide range of applications.

2 Governing equation of granular flow in SPH

In continuum mechanics, the motion of granular materials can be described
through the following momentum equation:
ߩ‫ܝ‬ሷ ൌ ߘ ‫ ڄ‬ો ൅ ߩ܏ (1)
where u is the displacement; ߩ is the density; ો is the effective stress tensor taking
negative for compression and g is the acceleration due to gravity.
Based on the SPH framework presented in [5, 15], the above equation can be ap-
proximated using the following kernel summation approximation:
ોೌ ାો್
‫ܝ‬ሷ ௔ ൌ σே
௕ୀଵ ݉௕ ൬ మ ାఘమ ൅ ۱௔௕ ൰ ‫ܹߘ ڄ‬௔௕ ൅ ܏ ௔ (2)
ఘೌ ್

where a indicates the particle under consideration, N WKH WRWDO QXPEHU RI ³QHLJh-
ERULQJ SDUWLFOHV´ RI a, ߩ the bulk density; ݉ the mass; ۱ the stabilisation term em-
ployed to remove the numerical instability issues associated with SPH [5], W the ker-
nel function taken to be the Cubic Spline function [16].
The stress-tensor appeared in equation (2) can be calculated using any material
constitutive models. It has been demonstrated by the authors that granular materials
could be well modelled using an elasto-plastic constitutive model with the Drucker-
Prager yield criterion [5, 6, 8, 17, 18]. This model is adopted in this paper and its
descriptions together with implementation features in SPH can be found in [5, 15].
Here, we only provide the final form of stress-strain relation, which is written as fol-
lows:
ீೌ
ોሶ ௔ ൌ ࣓௔ ‫ ڄ‬ો௔ െ ો௔ ‫࣓ ڄ‬௔ ൅ ʹ‫ܩ‬௔ ‫܍‬ሶ ௔ ൅ ‫ܭ‬௔ ઽሶ ௔ ۷ െ ߣሶ௔ ൤͵‫ܭ‬௔ ߙట௔ ۷ ൅ ‫ܛ‬௔ ൨ (3)
ඥ௃మೌ 
where H and Z indicates the total strain and spin tensors, e the deviatoric shear
strain tensor, I the Dirac delta function, K and G the elastic bulk and shear moduli, J2
is the second invariant of stress tensor, s the deviatoric shear stress, and O the plastic
multiplier, which can be defined from a consistency condition that prevents stresses
from exceeding the yielding limit:
ಸೌ
൫ଷఈക ௄ఌሶ ೡ ൯ ା ‫ڄ ೌܛ‬ઽሶ ೌ
ೌ ඥ಻మೌ 
ߣሶ௔ ൌ  (4)
൫ଽఈക ఈഗ ௄൯ ାீೌ
ೌ

with DI and D\ being constants calculated by fitting the Drucker-Prager with the
Mohr-Coloumb yield criterions [5, 15, 19, 20] and Hv the volumetric strain.
Numerical predictions of post-flow behaviour of granular … 897

3 An improved SPH model for simulations of granular flows

While the kinematics of SPH simulation of granular flows is generally realistic [4-8,
21], the stress field of SPH particles undergoing large shear deformation can exhibit
large stress oscillations. This problem is known as the short-term-length-scale [16],
and identified as one of the key challenges of the standard SPH method that needs to
be addressed in order to improve the accuracy of SPH simulations. Following the
same approach recently proposed by the authors [8], a stress regularisation technique
is adopted to smooth the stresses of SPH particles as follows:
‫ۃ‬ો௔ ‫ ۄ‬ൌ σே ෩
௕ୀଵ ો௕ ܸ௕ ܹ௔௕ (5)
෩௔௕ is the corrected kernel defined in [8] and ܸ௕ the particle volume.
where ܹ

4 Numerical demonstrations
Two-dimensional (2D) simulations of granular flows:
The proposed numerical approach is first applied to simulate a 2D granular column
collapse and results compared with the experiment conducted by Bui et al. [15]. Fig.
1a shows the initial configuration of the numerical sample and its materials listed in
Table 1. To highlight the effectiveness of the proposed regularisation technique, simu-
lations were conducted using a higher artificial viscosity, Dc = 1.0, as compared to
that of Dc = 0.1 [15]. Two case studies were considered, one with the regularisation of
stresses and the other without adopting this technique. In the SPH simulations, a total
number of 5000 particles was used to represent the granular volume with an initial
smoothing length of 0.24cm. Boundary conditions are assumed fully-fixed at the bot-
tom and free-slip on the vertical wall, which are modeled in SPH using fixed virtual
and ghost boundary particles, respectively [5, 15].
(a) (b) Granular column
20cm 0.5m

10cm a h/r
Granular volume
Solid boundary 2r = 0.2
Solid wall
0.5m

Fig. 1. Initial configuration of the 2D and 3D numerical models

Table 1. Material parameters for numerical simulations

Parameters 2D Simulation 3D Simulation Units
Elastic modulus 0.84 5.98 MPa
3RLVVRQ¶VUDWLR 0.3 0.3 -
Density 2,650 2,600 kg/m3
Cohesion 0 0 kPa
Friction 19.8 30 deg
Dilatation 0 0 deg
Artificial viscosity Dc = 1.0; Ec = 0 Dc = 1.0; Ec = 0 -
898 H. H. Bui and G. D. Nguyen

Without regularisation (Pa) Experimental result

0 865 1730 2595
Soil particles that experienced large
deformation show disturbed stress

With regularisation (Pa) Simulation result

0 865 1730 2595
5cm
The stress noise was
completely removed

Fig. 2. Comparison of numerical simulations and experiment

Fig. 2 shows the comparison between two SPH models for the vertical stress pro-
file (left images) and between the improved SPH model and the experiment (right
images) for the final run-out distance. In the standard SPH model, the stress noise
appears at the location where the material experiences large shear deformation,
though the high artificial viscosity has been adopted in the simulation. As discussed
above, while the kinematics of SPH simulation is realistic, the stress noise issue could
potentially introduce numerical errors to the elasto-plastic computation of granular
material during the post-failure process. However, the stress field in the undisturbed
material zone and the zone with small shear deformation remains smooth. Contrarily,
by adopting the stress regularisation technique, the stress noise is completely removed
over the entire computational domain. The final run-out distances of two models are
almost similar indicating that the proposed technique could effectively improve the
accuracy of the stress prediction in SPH simulations, and thus improving the overall
SPH performance for elasto-plastic computations. Comparing to the experiment, both
numerical solutions agree fairly well with the experiment for the predicted final run-
out distance. The minor difference between SPH simulations and the experiment is
attributed to the use of high artificial viscosity. By reducing the amount of artificial
viscosity to Dc = 0.1, excellent agreement with the experiment could be obtained [5,
15]. The above result also suggests that the kinematics of SPH simulation of granular
materials is not affected the short-term-length scale issue. Thanks to the kernel ap-
proximation, which helps naturally remove this issue in the standard SPH.

Three-dimensional (3D) simulations of granular flows:

The improved SPH model is also applied to simulate 3D granular column collapse
experiments. Fig. 1b shows the initial setup for the 3D simulation, which follows the
same setting conditions in [22]. It is a circular granular column having the initial di-
ameter of 10cm. The initial height of the granular column can be varied, allowing the
change of the initial aspect ratio a (ratio between the initial height and diameter of
granular column). Lube et al. [22] conducted a series of these experiments to investi-
gate the relationship between the final run-out distant (rv) and the initial aspect ratio a
of granular samples. Based on the experimental data they suggested the following
empirical equation for the final run-out distance of granular materials: rf = r (1 +
1.24a). This equation is used to validate the SPH numerical simulations.
Numerical predictions of post-flow behaviour of granular … 899

a 0.55 rf r (1  1.24a) 0.4

Final runout radius

0.3
2rv
0.2

Lude's emperical
empirical equation
0.1
SPH prediction

0
0 0.5 1 1.5 2
r = 16.8cm
v Aspect ratio a = h/r
(a) (c)

r =16.2cm
v

(b) (d)

Fig. 3. Comparison between SPH and experimental data conducted by Lube et al. [22]

Two case studies with the initial aspect ratio of a = 0.55 and a = 0.9 are conducted
using the improved SPH model with an initial smoothing length of 0.36cm. The
boundary condition is assumed fully-fixed at the bottom [5, 15]. Fig. 3(a-c) show the
comparison between SPH and the experiment by Lube et al. [22]. Excellent agree-
ment between SPH and the experiment is obtained for the final run-out distance (Fig.
3c). The undisturbed zone (flat surface in Fig. 3a) observed on the top surface of the
material sample in the experiment is also captured by the SPH model (Fig. 3b). Re-
garding the prediction of 3D stress field (Fig. 3d), the contour plot of the one-quarter
cross-section of the granular material sample shows very smooth vertical stress pro-
file. This indicates that the proposed regularisation technique is also applicable to the
3D conditions.

5 Summary
This paper presented an improved SPH solution for simulations of granular flows and
validated against the 2D and 3D granular column collapse experiments. The results
demonstrated that the proposed technique could significantly improve the stress pre-
diction of granular materials. Excellent agreements with the experiments at the la-
boratory scale suggest that the proposed technique could be applied to investigate the
fundamental mechanism of granular materials. For large scale applications, further
validations and investigations are still required and this work is in progress.

6 References
1. Cundall PA, Strack ODL. A discrete numerical model for granular assemblies. Géotechnique. 29
1:47-65 (1979).
2. Zhang X, Krabbenhoft K, Pedroso DM. Particle finite element analysis of large deformation and
granular flow problems. Comput Geotech. 54 0:133-142 (2013).
900 H. H. Bui and G. D. Nguyen

3. Ooi JY, Magnanimo V, Sun J, Luding S. Particle Modelling with the Discrete Element Method:
A success story of PARDEM. Powder Technol. 293:1-2 (2016).
4. Bui HH, Fukagawa R, Sako K, Wells JC. Numerical Simulation of Granular Materials Based on
Smoothed Particle Hydrodynamics (SPH). AIP Conference Proceedings. 1145, 575-578 (2009).
5. Bui HH, Kodikara JK, Bouazza A, Haque A, Ranjith PG. A novel computational approach for
large deformation and post-failure analyses of segmental retaining wall systems. Int J Numer
Anal Methods Geomech. 38 13:1321-1340 (2014).
6. Nguyen C, Bui HH, Fukagawa R. Two-dimensional numerical modelling of modular-block soil
retaining walls collapse using meshfree method. Int J Geomate. 5 1:647-652 (2013).
7. Nguyen CT, Bui HH, Fukagawa R. Failure Mechanism of True 2D Granular Flows. Journal of
Chemical Engineering of Japan. 48 6:395-402 (2015).
8. Nguyen CT, Bui HH, Nguyen GD, Fukagawa R. A new SPH-based approach to simulation of
granular flows using viscous damping and stress regularisation. Landslides.1-13 (2016).
9. Bui HH, Nguyen GD. A coupled fluid-solid SPH approach to modelling flow through
deformable porous media. Int J Solids Struct. 125:244-264 (2017).
10. Wang Y, Bui HH, Nguyen GD, Ranjith PG. A Mesh-Free Continuum Based Computational
Approach to Modelling Rock Fracture. Poromechanics VI. p. 2041-2048 (2017).
11. Bui HH, Sako K, Fukagawa R, Wells J. SPH-based numerical simulations for large deformation
of geomaterial considering soil-structure interaction. The 12th International Conference of
International Association for Computer Methods and Advances in Geomechanics
(IACMAG)570-578 (2008).
12. Wang Y, Bui HH, Nguyen GD, Ranjith PG. A new SPH-based continuum framework with an
embedded fracture process zone for modelling rock fracture. Int J Solids Struct. 159:40-57
(2019).
13. Zhao S, Bui HH, Lemiale V, Nguyen GD, Darve F. A generic approach to modelling flexible
confined boundary conditions in SPH and its application. Int J Numer Anal Methods Geomech.
43 5:1005-1031 (2019).
14. Tran HT, Wang Y, Nguyen GD, Bui HH. Modelling 3D desiccation cracking in clayey soils
using a size-dependent SPH computational approach. Comput Geotech. (2019).
15. Bui HH, Fukagawa R, Sako K, Ohno S. Lagrangian meshfree particles method (SPH) for large
deformation and failure flows of geomaterial using elastic±plastic soil constitutive model. Int J
Numer Anal Methods Geomech. 32 12:1537-1570 (2008).
16. Monaghan JJ. Smoothed Particle Hydrodynamics and Its Diverse Applications. Annual Review
of Fluid Mechanics. 44 1:323-346 (2012).
17. Bui HH, Fukagawa R, Wells JC. Slope stability analysis and discontinuous slope failure
simulation by elasto-plastic smoothed particle hydrodynamics (SPH). Géotechnique. 61 7:565-
574 (2011).
18. Nguyen GD, Nguyen CT, Bui HH, Nguyen VP. Constitutive modelling of compaction
localisation in porous sandstones. Int J Rock Mech Min Sci. 83:57-72 (2016).
19. Verghese SJ, Nguyen CT, Bui HH. Evaluation of plasticity-based soil constitutive models in
simulation of braced excavation. Int J Geomate. 5 2:672-677 (2013).
20. Bui HH, Fukagawa R. An improved SPH method for saturated soils and its application to
investigate the mechanisms of embankment failure: Case of hydrostatic pore-water pressure. Int J
Numer Anal Methods Geomech. 37 1:31-50 (2013).
21. Bui HH. Lagrangian mesh-free particle method (SPH) for large deformation and post-failure of
geomaterial using elasto-plastic constitutive models: Ritsumeikan University, (2007).
22. Lube G. Axisymmetric collapses of granular columns. J. Fluid Mech. 508:175-199 (2004).
23. Bui H H, Sako K, Fukagawa R: Numerical simulation of soil±water interaction using smoothed
particle hydrodynamics (SPH) method J. Terramechanics 44(5) 339-46 (2007).
24. Nguyen G D, Bui H H: The roles and effects of friction in cohesive zone modelling: A
thermodynamics-based formulation. Lecture Notes in Civil Engineering. 8:288-296 (2018).
25. Zhao S, Bui HH, Lemiale V, Nguyen G.D. SPH Simulation of strain localisation in geomaterials
using a visco-Plastic constitutive model. In Poromechanics VI: 1876-1883 (2017).
 Stochastic Site Response Analysis in Consideration
with Various Probability Distributions of
Geotechnical Properties

Thanh-Tuan Tran1,2,*, Phu-Cuong Nguyen3, Seung-Ryong Han4 and Dookie Kim1

1 Department of Civil Engineering, Kunsan National University, Republic of Korea
2
Faculty of Technology and Technique, Quy Nhon University, Vietnam
3
Faculty of Civil Engineering, Ho Chi Minh City Open University, Vietnam
4
Research Institute, Korea Electric Power Corporation Engineering & Construction,
Republic of Korea
*
tranthanhtuan@hotmail.com.vn

Abstract. Site response and soil profiles are a function of one another, and the
probable distribution of soil profiles have significant effects on the site response.
This study presents the effects of random variations of geotechnical properties on
the site response using the different probabilistic distributions. The uncertainties
of soil properties including shear wave velocity (ܸ௦ ), the density, and the nonlin-
ear soil properties are investigated and applied for the site-specific in the United
States. A computer program, PSHAKE, written in Python has been developed for
solving the variabilities soil properties via Monte Carlo Simulations. The com-
patibility and the enhancement of the proposed procedure are verified with the
variability in ܸ௦ using the Toro model. The outcomes obtained from PSHAKE are
verified based on the agreement with the reference solution given by the program
Strata. Additionally, the results in sensitivity analysis indicate that uncertain unit
weight has a smaller impact than the material degradation curve on probabilistic
model.

Keywords: Site Response Analysis; Probabilistic Distribution; Stochastic Anal-

ysis; Response Spectra; PSHAKE.

1 Introduction

Site response analysis is a process to simulate the seismic waves from the underlying
bedrock motion to surface ground motion through soil deposit. Output from ground
response analysis provides input to seismic analysis for structures especially in nuclear
industry [1]. In the practical earthquake engineering applications, there is no data avail-
able about the random variables, such as the layer thickness, shear wave velocity (ܸ௦ ),
density (ߛ), and material degradation. Thus, it is necessary to develop a simulation
technique for estimating probabilistic site response analysis. A probabilistic model is a
method that capable of solving a large number of simulations for predicting the site
behavior. In the probabilistic model, the selection of probability distribution functions

© Springer Nature Singapore Pte Ltd. 2020 901

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_144
 902 T.-T. Tran et al.

to represent the variation of geotechnical properties is a significant aspect for a statisti-

cal analysis [2-6].
The main objective of this research is to develop a probabilistic framework of site
response analysis that allows the uncertainties in geotechnical properties of soil deposit,
including ܸ௦ , density, and nonlinear soil properties. For these purposes, a computer pro-
gram PSHAKE [3] is coded in Python programing language based on the original
SHAKE91 framework.

2 Soil Property Variability

The ground response analysis is conducted for the Sylmar County Hospital (SCH) site
located in the San Fernando Valley of Southern California [8]. The total depth is about
90 m of alluvium above bedrock with ܸ௦ ranging from about 250 m/s at the surface and
rising to above 700 m/s at 60 m (see Fig. 1). The nonlinear soil properties are modelled
using Darendeli’s model which is a function of over-consolidation ratio (ܱ‫)ܴܥ‬, plas-
ticity index (ܲ‫)ܫ‬, and mean effective stress (ߪ଴ᇱ ). The uncertainty in nonlinear soil prop-
erty for each layer is generated with ߪ଴ᇱ equals to 0.36 atm, 2.2 atm, 5.6 atm and 7.7
atm. Both ܲ‫ ܫ‬and ܱ‫ ܴܥ‬for all layers in the model are defined as 10 and 1.0, respectively.
The soil is split into four layers and its parameters are shown in Fig. 2.

Fig. 1. Shear wave velocity Fig. 2. Soil layers

profile

Table 1. Summary of the process for site response analysis

Parameters Layer thickness Shear wave velocity Density Material degradation

Case 1 - Lognormal - -
Case 2 - Toro model - -
Case 3 - - Lognormal -
Case 4 - - - Darendeli model
Stochastic Site Response Analysis in Consideration … 903

In this study, four cases in Table 1 are applied to generate the variability in soil prop-
erties. A log-normal distribution has been adopted for ܸ௦ (case 1) and unit weight (case
3). Another case is applied for the variability of ܸ௦ using the Toro model (case 2). The
Darendeli model is used (case 4) for nonlinear soil property curves.

3 Computer Program

3.1 Methodology

A proposed procedure based on the SHAKE91 framework is developed for probabilis-

tic site response analysis via Monte Carlo Simulation (MCS), namely PSHAKE. How-
ever, instead of fixing the soil profile, an uncertainty of geotechnical parameters is con-
ducted to be stochastic processes, but with the different probability distributions. A
flowchart of the process is shown in Fig. 3. It starts with running the MCS module. This
module repeats the sub-process and conducts site response analysis for each simulation
and record the response. Finally, the statistical analyses (i.e. calculating mean values,
standard deviations and variance) are performed.

Start

Create soil
Site models
MCS profiles

Statistical analysis Perform analysis

End

Fig. 3. A flow chart of stochastic analysis

3.2 Verification of the Program

In order to verify the accuracy of the proposed process, the randomization of shear wave
velocity following the Toro model (case 2) is performed. The coefficient of variation
of 0.317 is taken to simulate the site profile. The results of PGA profile, maximum
shear strain profile, surface response spectra (ܵܽ௦ ), and amplification factor (‫ )ܨܣ‬are
presented and discussed. The outcomes generated by PSHAKE are compared to the
Strata program.
The comparison of peak ground acceleration (PGA) profile and maximum shear
strain profile obtained from PSHAKE and Strata are shown in Fig. 4. In this figure, the
gray and black circles represent responses for each simulation and the mean values from
PSHAKE, respectively. The red-dashed lines represent mean values calculated from
the Strata program. A comparison of the mean values along the profile is found with
the maximum difference of 5.86% and 9.70% for PGA profile and maximum shear
strain profile, respectively.
904 T.-T. Tran et al.

20

40
Depth (m)

60

80

this study
reference
100
0 0.5 1 -5 0 5
10 10 10
PGA (g) Max. shear strain (%)

Fig. 4. Mean PGA profiles and maximum shear strain profiles of PSHAKE and Strata

The ܵܽ௦ and ‫ ܨܣ‬calculated by PSHAKE and Strata are shown in Fig. 5. In this figure,
mean responses from PSHAKE (black lines) and Strata (red-dashed lines) are deter-
mined at surface. The red-dashed lines are mean values with plus or minus one standard
deviations obtained by PSHAKE. According to the observed results, it indicates that
the proposed program can generate acceptably accurate in probabilistic site response
analysis.

(a) Response spectra (b) Amplification Factor

3 5
this study
reference
4
Spectral Acceleration (g)

Amplification factor

2
3

2
1
1

0 0
-2 -1 0 1 -1 0 1
10 10 10 10 10 10 10
Period (s) Frequency (Hz)

Fig. 5. Mean surface response spectra and amplification factors of PSHAKE and Strata

4 Stochastic Site Response Analysis

The influence of geotechnical parameters of soil on the stochastic site response analysis
is evaluated based on the proposed procedure. The analysis is performed for a specific
site in the San Fernando Valley of Southern California. The spatial variability is con-
ducted by randomizing one parameter and keeping the others constant. Different cases
of random variations in soil profiles (see Table 1) are presented for this purpose. The
results of five hundred simulations of PGA at the surface are recorded.Fig. 6 shows the
comparison of the histogram of PGA obtained from four models. The variability of
geotechnical properties is presented by the mean and standard deviation values that are
displayed in Table 2. It is noted that the histogram of Case 3, which allows the uncertain
 Stochastic Site Response Analysis in Consideration … 905

unit weight, shows the lowest dispersion in the response with a standard deviation of
0.057. For randomization of nonlinear soil property, the largest standard deviation in
response is found with value of 0.139. Moreover, when comparing with Case 2, the
differences in PGA for the first case, third case and fourth case are 13.81%, 4.28% and
0.58%, respectively.

(a) Case 1 (b) Case 2

80 80

60 60

40 40

20 20

0 0
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9
PGA (g) PGA (g)

(c) Case 3 (d) Case 4

80 80

60 60

40 40

20 20

0 0
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9
PGA (g) PGA (g)

Fig. 6. Probabilistic distribution of PGA

Table 2. Statistical parameters of PGA obtained from different cases

Parameters Case 1 Case 2 Case 3 Case 4

Mean (g) 0.443 0.514 0.536 0.517
Std. (g) 0.130 0.120 0.057 0.139

5 Conclusions

A solution for performing the probabilistic procedure of ground response analysis is

proposed and implemented in the coded program, PSHAKE. This framework allows to
consider the uncertainties in soil characterizations, including the nonlinear soil proper-
ties, the density, and the ܸ௦ considering the different probabilistic distributions. To de-
termine the influence of the probabilistic distributions on the evaluations of the site
response, five hundred simulations of soil profiles were generated using the proposed
solution. The results show that the uncertainty in ܸ௦ with the log-normal distribution is
the most effective parameter compared with Toro model. Moreover, the density has a
less effect in the surface response that corresponds to the standard deviation of 0.057g.
906 T.-T. Tran et al.

Acknowledgment

This work was supported by the Korea Institute of Energy Technology Evaluation and
Planning (KETEP) and the Ministry of Trade, Industry & Energy (MOTIE) of the Re-
public of Korea (No. 20171510101960).

References
1. Tran, T.T., Cao, A.T., Nguyen, T.H.X., Kim, D.: Fragility assessment for electric cabinet in
nuclear power plant using response surface methodology. Nuclear Engineering and Tech-
nology 51(3), 894-903 (2019).
2. Wang, Y., Zhao, T., Cao, Z.: Site-specific probability distribution of geotechnical proper-
ties. Computers and Geotechnics 70, 159-168 (2015).
3. Tran, T.T., Han, S.R., Kim, D.: Effect of probabilistic variation in soil properties and profile
of site response. Soils and Foundations 58(6), 1339-1349 (2018).
4. Toro, G.R.: Probabilistic models of site velocity profiles for generic and site-specific
ground-motion amplification studies. Technical Rep, 779574 (1995).
5. Rota M., Lai, C.G., Strobbia, C.L.: Stochastic 1D site response analysis at a site in central
Italy. Soil Dynamics and Earthquake Engineering 31(4), 626-639 (2011).
6. Bazzurro, P., Cornell, C.A.: Ground-motion amplification in nonlinear soil sites with uncer-
tain properties. Bulletin of the Seismological Society of America 94(6), 2090-2109 (2004).
7. Darendeli, M.B.: Development of a new family of normalized modulus reduction and mate-
rial damping curves (Dissertation). The University of Texas at Austin (2001).
8. Gibbs, J.F., Tinsley, J.C., Joyner, W.B.: Seismic velocities and geological conditions at
twelve sites subjected to strong ground motion in the 1994 Northridge, California, earth-
quake (No. 96-740). US Geological Survey (1996).
 Analysis of strain measurements in large span soil-steel
bridge structure

Czesław Machelski1 Adam Czerepak2 and Piotr Tomala2

1
Department of Bridges and Railways, Wroclaw University of Science and Technology,
Wybrzeże Wyspiańskiego 27, 50-370 Wrocław, Poland
2
ViaCon, Przemysłowa 6, 64-130 Rydzyna, Poland

adam.czerepak@viacon.pl

Abstract. In this paper the effects of live loads on the displacement of the cor-
rugated steel structure buried in soil, i.e. soil-steel structure, will be analysed.
The loads are movable, but have characteristic of static passes. The results of
strain measurements with relatively dense layout located around the periphery
of the corrugated steel structure are used to calculate its deformations.
By using dense layout of strain gauges any direction of the structure dis-
placement can be calculated based on the changes of the strains in the analysed
circumferential strip (part) of the structure. Simple FE models are constructed
in order to accurately perform such calculation. Note that the existing analytical
calculations method dedicated to soil-steel structures do not provide any possi-
bility to calculate deformation of the steel shell. Therefore, this paper describes
sufficient methodology that allows to calculate steel shell deformation in soil-
steel structures bridge subjected to different live load models.

Keywords: soil-steel structure, moving loads, shell deformations, displacement

functions.

1 Introduction

The buried structures are made of corrugated steel plates surrounded by soil. All the
cross sections consists of circular arches. The typical geometric feature of such shells
is a circular plate (with radius of curvature R) in their top part (crown). The geomet-
rical characteristics of the shell’s shape are its span L and height H, which for the
considered bridge are given in Fig. 1. The geometry of the shell’s corrugated plate is
expressed by its technical designation: UC 500×237×9.65, standing for type UltraCor
a×f×t, where a – corrugation pitch, f – corrugation height and t – plate thickness.
The paper presents a displacement-computing algorithm, which is developed based
on the results of measurements of a bridge located in Ostróda (Poland). A wildlife
migration route and a local road pass under the bridge. There are four two-lane car-
riageways, separated by median strips, on the bridge. Hence the large width, amount-
ing to Bg = 61.82 m at the crest and to Bd = 95.70 m at the embankment toe. This

© Springer Nature Singapore Pte Ltd. 2020 907

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_145
908 C. Machelski et al.

bridge structure holds the world record for shell span length (the previous record be-
longed to a 24 m long span structure built in Canada).

Fig. 1. Cross section of the structure and location of the live load vehicle.

Testing of soil-steel structures under a load moving on the roadway usually turns to
quasi-static load [1]. The influence functions used for bridge models, e.g. [2], can be
useful in such analyses. Influence functions can also be determined on the basis of
measurements performed on soil-steel structures being in service, subjected to a load
in the form of a system of concentrated forces comes from the vehicle wheels [1].
Then the interaction of all these elements, including the pavement (which in classical
bridges is considered to be a secondary element not belonging to the structure), is also
taken into account.

2 Shell displacement

Corrugated plates in soil-steel structure models are modelled as orthotropic shells or

as beam elements [3]. The beam layout facilitates computations in many ways, e.g. it
becomes possible to formulate influence functions of the internal forces in the shell,
which help to determine the critical location of loads. Such approach will be de-
scribed below.
In order to determine the displacement of any shell point in radial direction one
uses the classic (Mohr’s) integral algorithm as in the equation for the beam elements
approach.

1 1
= ∫ + ∫ (1)

Calculation result (1) consists of two components coming from bending and com-
pression, respectively. The elements with subscript result from a vehicle load in the
form of a system of concentrated forces as shown in Fig. 1 Internal forces
Mp(s)and Np(s) are the functions related to the line of the shell circumferential section,
treated as the axis of a beam with an arch geometry, lying in plane xz. Functions
Mp(s) and Np(s) in (1) are considered to be the described below result of the meas-
urements of the structure model shown in Fig. 1, subjected to the system of forces ,
which means that the functions were obtained from 3D geometry.
Analysis of strain measurements in large span … 909

Integration is conducted over the whole structure perimeter, i.e. along an arch with
length , and so in planar system xz Since the geometrical parameters of this section
as a beam are constant, the stiffnesses are respectively and , whereby they can
be transposed to the front of the integral. The displacement tracking functions, denot-
ed with subscript , are internal forces Mr and Nr stemming from the unit load applied
in the considered displacement point and radial direction. According to the reduction
theory used in structural mechanics, in this case an equivalent model in the form of a
planar arch with the geometry of the shell circumferential section, i.e. a 2D model, is
used. The model is described further in this paper.
Therefore equation (1) can be now presented in the following form

= ∫κ ⋅ + ∫ε ⋅ (2)

Fig. 2 shows the geometrical relation (based on the planar cross section principle) for
unit strains in the cross section of the circumferential section. It follows from them
that the radius of curvature of beam ρ and its curvature κ are interrelated by the
strength dependence given in the formula

1 ε −ε
κ= =
ρ (3)
Equations (1) and (2) are exactly the same, but they are used for different load func-
tions. The effects of external loads in (1) are expressed by internal forces while in
(2,3) they are expressed by unit strains. In both the solutions of planar cross sections
principle as applied to beam systems is used.

Fig. 2. Geometry of circumferential sections and strain plot in corrugated steel plate.

Because of the complex shapes of the functions in (2), instead of the integral form of
the algorithm it is convenient to use a matrix algorithm, as in the equation

=
6
(κ ⋅B⋅m + ε ⋅B⋅n )
(4)
In this approach, when the arch line is divided into several segments, each having
length (as in Fig. 2), from the tracking functions in (2) one can form vectors mr and
nr used in (4). These are vectors formed from the values of the functions in the arch
910 C. Machelski et al.

points marked in Fig. 2. In the same way transposed vectors κT and εT are created
from the load function.
Assuming that functions ε(s) and κ(s) are continuous and when continuous tracking
functions r(s) and r(s) occur, one can use the band form of matrix B as below
⎡. . ⎤
⎢. 4 1 ⎥
⎢ ⎥
B=⎢ 1 4 1 ⎥
⎢ ⎥
⎢ 1 4 .⎥
⎢⎣ . . ⎥⎦
(5)
When the function is discrete in the point of application of the concentrated force,
as, e.g., in the case of r(s), one can adopt in this point the division of matrix B into
two blocks:
⎡. . ⎤
⎢. 4 1 ⎥
⎢ ⎥
⎢ 1 2 ⎥
B=⎢ ⎥
⎢ 2 1 ⎥
⎢ 1 4 .⎥
⎢ ⎥
⎣⎢ . . ⎦⎥
(6)
When displacement function (xp), represented by vector r in the matrix algorithm,
is to be determined, it is advantageous to form influence matrices Lk and Lε from
vectors κT and εT and to use the relation.

r=
6
(L κ ⋅ B ⋅ m + Lε ⋅ B ⋅ n )
(7)
The advantage of the solution included in (2) and (7) is limiting the size of the task to
the circumferential band separated from the structure. The values εg and εD contain
full information about the structure geometry (3D) and the boundary conditions. The
actual characteristics of the backfill material and road superstructure are also
determined. In the distribution (function) of unit deformations along the length of the
perimeter band, the actual principles of interaction of structural elements, eg soil slip
with respect to the shell in the contact layer, are taken into account. The functions εg
(s) and εD (s) are mapped to changes in the position of the load.

3 Shell deformations as a result of strain gauges measurements

The results of live load tests based on the concept of a moving load in the quasi-static
sense are presented in this paper. Two trucks driving parallelly to each other along the
road axis constituted the load. The trucks would drive stepwise, i.e. would stop at
distances xp relative to the shell’s crown point – hence driving step /4. Measuring
instrument readings were done at each of the stops.
Analysis of strain measurements in large span … 911

The load on each of the truck’s front axles was 1 = 60 kN and that acting on each
of the identical rear axles was 2 = 100 kN, hence the truck’s total weight = 32 tons
(all together 64 tons). In each of the measuring locations of the trucks during their
travel there (the primary pass) and back (the secondary pass) the position of the
wheels on the road pavement was identical, as shown in fig. 1. The trucks moved
without any turning around. The full loading cycle took for nearly three hours, which
gives the average time of about 12 min per one measurement cycle.
Fig. 3 and Fig. 4 show selected examples of calculations using formula (7) and the
results of strain gauges measurements in the soil-steel structure’s points given in fig.
2. On the vertical axis, the deflection (vertical displacement) of the structure was
assumed and on the horizontal axis the position of the vehicle relative to the structure
crown point, i.e. xp. In the legend to the graphs, the point numbers from Fig. 2 are
given.

Fig. 3. Functions of the vertical deformations of the points located on the left side of the struc-
ture.

Fig. 4. Functions of the deflections of the points located on the right side of the structure.

Using the general algorithm one can calculate displacements in any of selected shell
points, in the form of components belonging to the two groups: (radial) and (tan-
gent) and (horizontal) and (vertical). On the basis of the two groups of displace-
ment components one can determine resultant displacement v (Fig. 1) as the geomet-
ric sum.

2
= + 2= 2
+ 2 (8)
In the proposed algorithm it is possible to select a point (not necessarily consistent
with the configuration shown in Fig. 1) and a displacement direction. As a rule, the
value of resultant is not determined since its direction is variable.
In the soil-steel bridges one of the important element of the structure is the road
superstructure. The physical properties of material, which is made of, are signifi-
912 C. Machelski et al.

cantly different from the engineering backfill. The key rule in a bearing capacity
plays the backfill material. The results given in this paper were compared with in-
ductive measurements and laser geodesy results. The comparisons have been placed
in [4].
In the calculation linear elastic behaviour of the steel shell has been assumed as in
the testing range linear behaviour of the steel shell has been observed. The non-
linearity of the problem given in Fig. 3 and Fig. 4 results from the behaviour of
ground backfill.

4 Summary

The results of strain gauges measurements performed for a dense grid of points locat-
ed on the corrugated plate’s circumferential section were used to calculate displace-
ments characterizing the deformation of the shell buried in soil. Consequently, this
method comprises all the soil-steel structure’s elements: the corrugated plate, the
backfill, the pavement with its foundation and takes into account all the natural (real)
principles governing the interactions between these elements that have potential influ-
ence on the steel shell deformation. This means that the structure’s geometry and
physical properties and the load are faithfully (accurately) reproduced in 3D.
The algorithm for converting strains into displacements proposed in this paper al-
lows to obtain accurate results. The advantage of this measuring method (electric
resistance extensometer) is that there is no need for the solid scaffold used for dis-
placement measurements. The results of the measurements show that the changes in
strains are different between the primary pass and the secondary pass. This was ob-
served previously when measuring the displacements of the shell in soil-steel struc-
tures [2]. The characteristic feature of the experimental results (strains) and the calcu-
lated geometric effects (e.g. ρ) and displacements is a hysteresis loop [5].

References
1. Machelski C.: Dependence of deformation of soil-shell structure on the direction of load
passage. Bridge and Road 13, pp. 223-233 (2014).
2. Machelski C.: Kinematic method for the determination of influence function of internal
force in the steel shell of soil-steel structures. Studia Geotechnika et Mechanica. No. 3 pp.
27-40 (2010).
3. Machelski C., Janusz L.: Application of Results of Test in Developing 2D Model for Soil-
Steel Railway Bridges. Journal of the Transportation Research Board. Solid Mechanic, pp.
70-75 (2017).
4. Machelski C.: Soil-steel structure shell displacement functions based on tensometric meas-
urements. Studia Geotechnica et Mechanica 2/2018 p. 55-57
5. Sielver M. L. Seed H. B.: Changes in Sands during Cyclic Loading. ASCE Soil Mechanics
and Foundations Division Vol. 97, No SM9 (September 1971).
 Part V
Urban Planning, Transport and
Environment
 A Landscape Structure Proposal for a Sustainable
Development – Quy Nhon City

,IH9-CIG5'5HCG3


4"IM9B.<=(;IM9B3


4.<C05B
*<5B3


45B8*<5B.<=&9.<IM3


4

/B=J9FG=HMC:PJCF5
"# &5F;C'5FEIYG89'5F=5@J5 
 PJCF5 *CFHI;5@
rsm@uevora.pt

+IM(<CB/B=J9FG=HM[!9C;F5D<M9D5FHA9BH  BICB;0ICB; (;IM9B05BI +I=
(<CB7=HM =B<=B< 0=9H(5A
nguyenthihuyen@qnu.edu.vn

Abstract. 77CF8=B;HCG=5   H<9-CIH<95GHG=5GIF65B=N5

H=CBF5H9K=@@=B7F95G9:FCA=B  HC=B  #BH<9:579C:H<=G H<9

0=9H (5A ;CJ9FBA9BH 5DDFCJ98 =HG (5H=CB5@ !F99B !FCKH< -HF5H9;M  
K<=7< 7CAD@9A9BHG H<9 F5A9KCF? '5GH9F *@5B :CF /F65B 9J9@CDA9BH =B
0=9H(5AHC 5B8H<9  0=G=CB.<=G:F5A9KCF?7CADF=G9G565@5B798
5B8GHF5H9;=7;FCKH<H<FCI;<H<989J9@CDA9BHC:G97CB85FM5B8H9FH=5FM7=H=9G 
5G+IM(<CB=HM+( 5G<I6GK=H<=BDFCJ=B7=5@IF65BGMGH9AG.<9!F99B
=H=9G 7H=CB *@5B 9A6C8=9G G B9K !F99B =H=9G 5DDFC57< IF65B 89J9@

CDA9BH6MA95BGC:H<9@9BG9BJ=FCBA9BH 97CBCAM 5B89EI=HM5=A=B;HC

89J9@CD7=H=9G=B5GIGH5=B56@9K5MHC69;F99B 7CAD9H=H=J9 5B8=B7@IG=J9=B5B
9BJ=FCBA9BH5@  97CBCA=75B8GC7=5@ K5M1=H<=B H<9 !F99B=H=9G:F5A9KCF? 
K9DFCDCG95B9K89G=;B5DDFC57<HC+(D@5BB=B;.<=G5DDFC57<65G98CB
@5B8G75D9 GMGH9AG =G A5H9F=5@=N98 =B 5 7CBH=BICIG 5B8 DFC8I7H=J9 @5B8G75D9
GHFI7HIF9DFCACH=B;H<9C77IFF9B79C:97C@C;=7 97CBCA=75B8GC7=5@DFC79GG9G
.<=GAI@H=:IB7H=CB5@GHFI7HIF9=G:IB85A9BH5@:CFH<9GIGH5=B56@9;FCKH<C:7=H

=9G GI7< 5G +(  9B56@=B; 5 F95@H=A9 F9GDCBG9 HC 97C@C;=7 7IFF9BH :F5;=@=H=9G
:5798 6M IF65B GD579G  DF9G9FJ5H=CB C: 7I@HIF5@ J5@I9G  <9F=H5;9  5B8 =89BH=HM 
59GH<9H=77CB79FBG5B8@9=GIF9#H5@GC9B56@9GD5FHB9FG<=DG5ACB;H<9@C75@5B8
79BHF5@;CJ9FBA9BH H<97CAAIB=H=9G5B8H<9DF=J5H9G97HCF =B8IGHFM5B89BH9F

DF=G9G .<9G9 D5FHB9FG<=DG K=@@ 588F9GG H<9 DFC6@9AG  =89BH=:M 5B8 =AD@9A9BH
H<9589EI5H9GC@IH=CBGHCH<9G9=GGI9GK=H<CIH7CADFCA=G=B;GC7=5@9EI=HM

Keywords:&5B8G75D9-HFI7HIF9 /F65B9J9@CDA9BH -IGH5=B56=@=HM

1 Introduction

0=9H (5A =G CB HF57? HC A99H 5@ACGH 5@@ =HG '=@@9BB=IA 9J9@CDA9BH !C5@G ,5D=8
IF65B=N5H=CB=G8F=J9B 6M 5 GHFI7HIF5@ G<=:H :FCA5;F=7I@HIF9 HC =B8IGHFM5B8 G9FJ=79G 
K<=7<H5?9BHC;9H<9FBCK577CIBH:CFB95F@M C:97CBCA=7CIHDIHG=B79 34
#B 0=9H (5A =G ACGH BCH56@9 :CF H<989J9@CDA9BH C:H<9 GCIH<
79BHF5@ 7C5GH5@ 7=H=9G 
B5A9@M +IM (<CB  K<=7< F9@M CB HCIF=GA 5B8 HF589 9J9@CD=B; DFCJ=B7=5@ 75D=H5@

© Springer Nature Singapore Pte Ltd. 2020 915

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_146
916 R. S. Matos et al.

7=H=9G5G+IM(<CB 5GACHCFG=BH<98=::9F9BHF9;=CBGC:H<97CIBHFMK=@@=ADFCJ9H<9
@C75@ 97CBCAM  9EI5@=N9 F9;=CB5@ 89J9@CDA9BH  GHF9B;H<9BIF65B
FIF5@@=B?G 5B8 58

8F9GGCJ9FDCDI@5H=CB=BA5=BIF65B79BH9FG-C K975BG5MH<5HIF65B=N5H=CBK=@@69

:IB85A9BH5@=BH<989J9@CDA9BHC:GCIH<95GH0=9H(5A B5A9@M=B+(HF5BG:CFA=B;
=HG97CBCAM5B8DFCJ=8=B;GIDDCFHHC;F99B97CBCA=7;FCKH<#BH<=GG9BG9 D@5BB=B;
H<5H=B7@I89G@5B8G75D9GMGH9AG5B897C@C;M=GB99898HC;9HIF65B89J9@CDA9BH=B

GH958 C: =HG ;FCKH<.<=G 75B69 57<=9J98H<FCI;< B5H=CB5@GHF5H9;=9G  K<=7< DFCJ=89
CF=9BH5H=CBGHCIF65B89J9@CDA9BH @958=B;HC5B998:CF5BIF65B89J9@CDA9BHGHF5H9

;M34#BH<=GK5M Green Cities5DDFC57<G<CI@8697CBG=89F985GC@IH=CBHCH<97<5@

@9B;9GC:7IFF9BHIF65B=N5H=CB6M=BH9;F5H=B;97C@C;=77CB79FBG=BH<9IF65BD@5BB=B;
=ADFCJ=B;@=J956=@=HM5B8F9G=@=9B7934
.<9 89G=;B 5DDFC57< K9 5F9 DFCDCG=B; HC +( D@5BB=B; =G D9F:97H@M :F5A98 =B H<9
Green CitiesGHF5H9;M5B8=BG=5B9J9@CDA9BH5B?DC@=7=9G#H7CBG=GHGC:
H<9 DFCDCG5@ C: 5 @5B8G75D9 GHFI7HIF9 65G98 CB @5B8G75D9 GMGH9AG 5B8 CB @5B8G75D9
AI@H=:IB7H=CB5@=HM H<5H =B7@I89G H<9 :IB7H=CB C: DFCH97H=CB C: 97C@C;=75@@M G9BG=H=J9
5F95G H<9:IB7H=CBC:DFC8I7H=CB5B8H<9:IB7H=CBC:@9=GIF9.<=G@5B8G75D9GHFI7HIF9
K=@@ @=B? 5B8 5FH=7I@5H9H<97CBGC@=85H98 IF65B :56F=7 K=H< =HG D9F=D<9FM 5B8 K=H<H<9
7CIBHFMG=895B8K=@@=B7@I89CD9BGD579G 97C@C;=7G9BG=H=J95F95G 7I@HIF5@GD579G5B8
5;F=7I@HIF5@5F95G.<=G@5B8G75D9GHFI7HIF9K=@@697CBH=BICIG=B5B97C@C;=7 7I@HIF5@
5B8 F97F95H=CB5@ K5M5B8 DFC8I7H=J9 34 #H K=@@ GIDDCFH ;F99B89J9@CDA9BH7CBHF=6

IH=B; HC DCG=H=J9 <95@H< =AD57HG  =ADFCJ9 7@=A5H9 F9G=@=9B79 5B8 =ADFCJ9 IF65B D9F

:CFA5B79=BHC5AI@H=:IB7H=CB5@IG9=B7F95G=B;5HHF57H=J9B9GG5B8DFCJ=8=B;GIDDCFHHC
5;F99B97CBCAM

2 Methodology

.<=G89G=;B5DDFC57<=G5F9GI@HC:@CB;
H=A9F9G95F7<CB7CBH9ADCF5FM7=H=9G =BH9FGH=

H=5@GD579G @5B8G75D9AI@H=:IB7H=CB5@=HM CD9BGD579G 7I@HIF5@GD579G 97C@C;=7GHFI7

HIF9G 5B8IF65B5;F=7I@HIF9
CF H<=G D5D9F  H<9 F9G95F7<
:C7IG98  A5=B@M  =B 97C@C;=7 GHFI7HIF9G 34  CD9B GD579G
GMGH9AG5B8IF65B5;F=7I@HIF95GK9@@5GH<9=BH9F57H=CBG69HK99BH<97=HM H<97=HM5B8
=HG D9F=D<9FM 5B8 H<9 7CIBHFMG=89 :CF A=@@9BB=5 34  5B8 K<9F9 H<9 97C@C;=7 GMGH9AG
5B8 5;F=7I@HIF9 5@@CK H<9A HC GHFI7HIF9 H<9 @5B8G75D9 GIDDCFH=B; 5 AI@H=:IB7H=CB5@ 
7CAD@9A9BH5FM5B865@5B798IG934
.<=GGHI8M:C7IG9GCB+IM(<CB=HM+(5B8CB=HGF9@5H=CBG<=DK=H<=HGD9F=D<9FM
5B8FIF5@GD57934 34#BH<9:=FGHACA9BH =H=GGIAA5F=N98H<99JC@IH=CBC:H<97=HM
=BCF89FHCIB89FGH5B8H<9;FCK=B;IF65BDFC79GGC77IFF98=BH<9G=5B7CBH9LH A5=B

@M=B0=9H(5A
.<9H<9CF9H=75@GIDDCFH:CFH<=GD@5BB=B;DFCDCG5@F9GI@HG:FCA5F9G95F7<H<5HC77IFG
G=B79  345B8 C: =HG9JC@IH=CB5B8 F9:@97H=CBG CF=;=B5H98 :FCA H<9F9 .<=G @5B8

G75D9GHFI7HIF9DFCDCG5@=BH9B8GHC6957CBHF=6IHCFHC5GC7=5@DFC>97H 9=H<9F97CBCA

=75@@M  GC7=5@@M5B89BJ=FCBA9BH5@@M 65@5B798  5B8 H<5H =GD9F:97H@M :F5A98 =B 57CB

H9ADCF5FMGreen Cities5DDFC57<
A Landscape Structure Proposal … 917

3 Quy Nhon City – Urban Context

.<9CF=;=BGC:+IM(<CB+(75B69HF5798HC=HG=895@@C75H=CBG5@CB;H<9:9FH=@9
J5@@9MG 5B8 D@5=BG C: "5.<5B< 5B8 %CB F=J9FG 5B8 B95F6M H<97C5GH@=B9 K<=7< 9B5

6@98=HHC;9B9F5H95;F=7I@HIF9GIFD@IG 9B;5;9=B5DFC:=H56@9HF5895B889J9@CD5G5B
IF65B79BH9F+IM(<CBG9FJ985G5F9;=CB5@79BH9FC:HF589 5;F=7I@HIF5@A5F?9H A=@=

H5FM;5FF=GCB 5B8HF5=B5B86IGGH5H=CBGH<5H5F97CBB97H98HCDF=A5FM7=H=9G6MFC58G 
F5=@K5MG CFF=J9F=B9HF5BGDCFHG34
 1=H< =BJ9GHA9BHG =B=B:F5GHFI7HIF95B8 GC7=5@ :57=@=H=9G  +IM (<CB<5G 9L
D5B898
#B H<=G DFC79GG 97CBCA=7 7CB79BHF5H=CB C77IFF98 5G K9@@ 5G :C@@CK=B; ;FCKH< 5B8
DFC8I7H=J=HM =B7F95G9G  F98I7=B; DCJ9FHM  5B8 =B7F95G=B; @=J=B; GH5B85F8G "CK9J9F 
IF65B=N5H=CB<5G5@GC=B9J=H56@M6FCI;<H56CIHGD5H=5@5B8=B7CA98=GD5F=H=9G5B8IB

7CBHFC@@98 ;FCKH< K=H< =HG 58J9FG9 7CBG9EI9B79G GI7< 5G 7CB;9GH=CB  9BJ=FCBA9BH5@
DC@@IH=CB 5B8GC7=5@=AD@=75H=CBG34GH<97CF95F95C:H<97=HM697CA97CB;9GH98
5B8 @5B8 F9BHG =B7F95G9  H<9F9 =G 5 H9B89B7M :CF ;FCKH< HC G<=:H 5B8 9LD5B8 CIHK5F8
HCK5F8 D9F=
IF65B 5B8 CIH@M=B; 5F95G ;F=7I@HIF5@ @5B8 =B H<9 D9F=D<9FM  5G K9@@ 5G
97C@C;=7 G9BG=H=J9 5F95G  K9F9 7CBJ9FH98 HC IF65B @5B8 .<9 GDF5K@ K5G IBF9;I@5H98
5B8 89BG=H=9G =B7F95G985B8 5F9 GH=@@ =B7F95G=B; 5G 5 F9GDCBG9HC H<9 ;FCK=B; HCIF=GA
:5G<=CB 57=B;H<=G @C75@5IH<CF=H=9G5F9=@@
9EI=DD98HCDFCJ=89DI6@=7G9FJ=79G34
 #BH9;F5H98 K=H< H<9 !F99B =H=9G 5DDFC57< :CF GIGH5=B56@9 89J9@CDA9BH :CF G=5B
=H=9GH<=GIF65BD@5BB=B;DFCDCG5@C::9FG=HG9@:5G5B5@H9FB5H=J9HCGD97I@5H=J9IF65B
;FCKH<H<5H<5G699BK=89GDF95857FCGG0=9H(5A K<9F9+IM(<CB=GBC9L79DH=CB
#BH<9B5A9C:97CBCA=7;FCKH< K9<5J9699B5GG=GH=B;HCH<9A5GG=J9K5H9FDFCC:=B;
C:7C5GH5@GHF=DG H<9:9@@=B;C:KCC8985F95G H<9=AD@9A9BH5H=CBC:HCIF=GHF9GCFHG=B
97C@C;=75@@MG9BG=H=J95F95G.<=G;FCKH<75B69HF5BG:CFA98=BHCIF65B89J9@CDA9BH
K<9B 7CBG=89F=B; H<9 D<MG=75@ 5B8 7I@HIF5@ 7<5F57H9F=GH=7G C: H<9 H9FF=HCFM 5B8 @5B8

G75D9'I@H=:IB7H=CB5@IG9G=B7@I8=B;IF65B5;F=7I@HIF9 DFCH97H=CBC:97C@C;=7G9BG=

H=J95F95G5B8CD9BDI6@=7GD579GH<5HDFCACH9IF65B:56F=75FH=7I@5H=CB5B87CAAI

B=HM_G GC7=5@=N5H=CB AIGH C77IF 7CAD@9A9BH=B; H<9 97CBCA=7 89J9@CDA9BH [4], [6

&5B8G75D9D@5BB=B;=G:IB85A9BH5@=BH<9DFC79GGC:F9H<=B?=B;5B87F95H=B;GIGH5=B5

6@97=H=9G.<=G=GK<5HH<=GGHI8M5=AGHCC::9F

4 A Landscape Structure to Quy Nhon City

.<=GDFCDCG5@K5G89H9FA=B9865G98CBH<96=CD<MG=75@7<5F57H9F=GH=7GC:H<9H9FF=

HCFM 5B8 =B H<9 9L=GH=B; 5B8 7CBGHFI7H98 75FHC;F5D<M 7CBGH=HIH=B; 5 H9FF=HCF=5@ 65G9
H<5H75BG7=9BH=:=75@@M89H9FA=B9K<=7<5F9H<95F95GHC588HC5AI@H=:IB7H=CB5@@5B8

G75D9 GHFI7HIF9 H<5H K=@@5FH=7I@5H9H<9 7=HM =HG9@:  H<97=HM K=H<H<9 D9F=D<9FM 5B8 H<9
D9F=D<9FM K=H<H<9 FIF5@ GD579 [4] [6] [7] .<=G GHFI7HIF9 K=@@ 5GGIF9 5 7CBH=BI=HM 69

HK99BH<9CD9BGD579G7I@HIF5@ 97C@C;=7 DFC8I7H=J9 F97F95H=CB5@577CF8=B;K=H<=HG

7<5F57H9F=GH=7G 5G=HK5GF9:9FF9869:CF9 5B865G98CBH<9IB89FGH5B8=B;C:H<9@5B8

G75D9 GMGH9AG 5GGIA=B;H<9A5G 89:=B9FGC:H<9 GHFI7HIF9 H<5H =B7@I89GHKC7CADC

B9BHG5DFC8I7H=J9CB95B85B97C@C;=75B87I@HIF5@CB9H<5H5@@CKH<95=A987CBH=BI=

HM.<9DFC8I7H=J97CADCB9BH7CFF9GDCB8GHCH<95F95GC:IF65B5B8FIF5@GD579HC69
918 R. S. Matos et al.

A5B5;98HC699BJ=FCBA9BH5@@M5B897CBCA=75@@MDFC8I7H=J9[7] [8].<97CBH=BICIG

7CADCB9BH 7CFF9GDCB8G HC 5 B9HKCF? K=H< 97C@C;=75@  7I@HIF5@ 5B8 F97F95H=CB5@ J5

@9B79G  5FH=7I@5H=B; H<9 DFC8I7H=J9 7CADCB9BH  H<9 IF65B :56F=7 5B8 H<9 FIF5@ GD579
DFCACH=B;AI@H=:IB7H=CB5@IG9343434

4.1 Areas to Add to the Productive Component of the Landscape

.<9 89@=A=H5H=CB C: H<9 \DFC8I7H=J9 7CADCB9BH C: H<9 @5B8G75D9^ K5G 65G98 CB H<9
5J5=@56@9 =B:CFA5H=CB 56CIH H<9 5F95G K=H< 5;F=7I@HIF5@ DCH9BH=5@ =B IF65B 5B8 FIF5@
GD579B5A9@M-C=@'5D:FCA#BGH=HIH9C:-C=@<9A=GHFM  &5B8/G9'5D:FCA
(5HIF5@ ,9GCIF79G 5B8 BJ=FCBA9BH C: =B< =B< *FCJ=B79   5B8 CF9GH '5D
:FCA CF9GH#BJ9BHCFM5B8*@5BB=B;-I6
#BGH=HIH9 
975IG9H<9C6>97H=J9 K5G H<9 89@=A=H5H=CB C: DFC8I7H=J9 GD579 H<99L=GH=B;5;F=7I@

HIF9 5B8 :CF9GH GD579G =BG=89 5B8 GIFFCIB8=B; +( K9F9 7CBG=89F98 7CBGH=HIH=B;
GD579GK=H<H<9DCH9BH=5@HCH<989J9@CDA9BHC:5DC@M7I@HIF9IG9
:H9F H<=G G9@97H=CB  K9 89H9FA=B98 H<9 GC=@G K=H< DCH9BH=5@ :CF 5;F=7I@HIF5@ IG9 .<9
89H9FA=B5H=CBC:H<9G9GD579GK5GA589K=H<H<9G9@97H=CBC:GC=@GK=H<<=;<IG975

D57=HM 5B8 H<95F95G K=H< GC=@ IB=HG 7@5GG=:=98 5G@CK5@@IJ=5@5B87C@@IJ=5@ 5B8H<9=F

DCGH9F=CF5GGC7=5H=CB=B5G=B;@9D5H7< :FCAIF65BHCFIF5@GD579G
77CF8=B;HCH<9DFCDCG98A9H<C8 H<95F95GH<5H5F95DHHC<5J95B5;F5F=5B@5B8IG9 
HC;9H<9FK=H<H<99LD5BG=CB5F95GC:H<9IF65BGC=@H<5H5F9BCHC77ID=986M7CBGHFI7

H=CB K<9B7CB>I;5H98K=H<H<9GC=@GD5H7<9GK=H<5;F=7I@HIF5@DCH9BH=5@7CBGH=HIH9H<9
\DFC8I7H=J97CADCB9BHC:H<9@5B8G75D9^:=;5

4.2 Determining the Continuous Component of the Landscapeeading

.<9\7CBH=BICIG7CADCB9BH C:H<9@5B8G75D9^H<5H K9 89:9B85=AG5H :=B8=B;5 B9H

KCF? C: GD579G H<5H 9GH56@=G< H<9 7CBH=BI=HM 5B8 5FH=7I@5H=CB 69HK99B H<9 D5H7<9G C:
\DFC8I7H=J9@5B8G75D9^5@F958M89@=B95H98=B IF65B 5B8 FIF5@ GD579  H<9 IF65B :56F=7
5B8H<97=HM5B8H<97CIBHFMG=89
"5J=B; 5G 65G=G H<9 /F65B D@5BB=B;   0=G=CB  H<9 F9:9FF98 CF9GH '5D 5B8
&5B8 /G9 '5D    H<9 F95G K=H< I@HIF5@ 05@I9 65G98 CB )D9B -HF99H '5D
   @CC8 '5D :FCA IFCD95B -D579 ;9B7M    =;=H5@ @9J5H=CB 'C89@
:FCA @5G?5 -5H9@=H9 57=@=HM    &5B8 /G9 *@5BB=B; '5D :FCA 9D5FHA9BH C:
CBGHFI7H=CBC:=B<=B<*FCJ=B79  ,=J9F'5D:FCA=B<=B<9D5FHA9BH
C:#FF=;5H=CB  5B87C@C;=75@F95G'5D65G98CB)D9B-HF99H'5D  5B
97C@C;=7GHFI7HIF9K5G89:=B986975IG9C:=HG=ADCFH5B79=B5&5B8G75D9-HFI7HIF9#HG
89@=A=H5H=CB K5G65G98 CBH<997C@C;=75@@M G9BG=H=J95F95G 5B87I@HIF5@ GD579G C: H<9
@5B8G75D9 =B5F95GK=H<=BH9F9GHHCB5HIF97CBG9FJ5H=CB=BNCB9GC:<=GHCF=7 5F7<=H97

HCB=7 @5B8G75D95B8=AA5H9F=5@<9F=H5;95B8CH<9FNCB9G:IB85A9BH5@HC6=C8=J9FG=HM
5B8F9GCIF79GG5:9;I5F85B8HCH<9F9;I@5H=CBC:H<96=C@C;=7DFC79GG9GH<5H5@@CKH<9
GIGH5=B56@9 89J9@CDA9BH C:<IA5B57H=J=H=9G5G98 CBH<=G =B:CFA5H=CB K9 G9@97H98
H<9D5H7<9GH<5H7CFF9GDCB8HCH<9:IB85A9BH5@97C@C;=7GMGH9AG:=;6.<97CBH=

BI=HMC:H<9@5B8G75D9=GGH=@@J5@I98:CFH<9=BH9;F5H=CBC:5@@CD9BGD579G5B87I@HIF5@
5B8<9F=H5;9 5F95G H<5H 7CBGH=HIH95 B9HKCF? C: K5@?56@9 5B8 F97F95H=CB5@ D@579GH<5H
A Landscape Structure Proposal … 919


;C H<FCI;< H<9 IF65B :56F=7 5@@CK=B; H<9 =BH9F7CBB97H=CB 69HK99B H<9 7=HM 5B8 H<9
GIFFCIB8=B;FIF5@GD579

4.3 Productive and Continuous Areas to be Included in a

Multifunctional Landscape Structure.
.<9 CJ9F@5DD=B; C: 6CH< 7CADCB9BHG DFC8I7H=J9 5B8 7CBH=BICIG 89H9FA=B9G H<9
5F95GHC588HCH<9DF9H9B898AI@H=:IB7H=CB5@@5B8G75D9GHFI7HIF9:CF+(.<=GD@5B

B=B;9L9F7=G95B8H<9C6H5=B98GHFI7HIF9K=@@69H<9GIDDCFH:CFH<9:IHIF9CD9BGD579G
89G=;BH<5HK=@@5GGIF9H<9A5H9F=5@=N5H=CB5B8A5B5;9A9BHC:H<=GB9HKCF?C:7CBH=B

ICIGGD579G7CBB97H=B;H<9IF65B:56F=75B8H<9IF65B5B8FIF5@GD579G:=;7


Fig. 1.5 *FC8I7H=J97CADCB9BH6 CBH=BICIG7CADCB9BH7 *FC8I7H=J95B8CBH=BICIG5F95G

5 Conclusion

.<9=B7@IG=CBC:5@5B8G75D9GHFI7HIF9=B+(F9:@97HG5B9K89G=;B5DDFC57<65G98
CB H<9 6=CD<MG=75@ 7<5F57H9F=GH=7G C: H<9 H9FF=HCFM 5B8 7I@HIF5@ J5@I9G #B7@I8=B; 6=C

D<MG=75@7<5F57H9F=GH=7GA95BGHC7CBG=89F=HGB5HIF5@A5HH9F5B8J5@I9G;9C@C;M GC=@ 
ACFD<C@C;M B5H=J9J9;9H5H=CB.<=G7CADCB9BH5GGIF9GH<9G5:9;I5F8C:DFC8I7H=J9
DCH9BH=5@GC=@GK=H<5;F5F=5BDCH9BH=5@5B897C@C;=7J5@I9GF=J9FG GHF95AG @5;CCBG 
6957<9G 5B8 7C5GH@=B9G  G@CD9G  B5H=J9 J9;9H5H=CB  <958K5H9FG DCFCIG GI6GHF5H9 HC 5
A5GG=J9 K5H9FDFCC:=B; 5B8 C77ID5H=CB K=H< 6I=@8=B;G 5B8 IF65B =B:F5GHFI7HIF9G 34
#B7@I8=B;H<97I@HIF5@J5@I9GA95BGHC7CBG=89FH<97CAAIB=H=9G5DDFCDF=5H=CBC:H<9
H9FF=HCFMH<FCI;<=HGA5F?G5B8DF=BHGH<5HF9J95@H<9=F<=GHCFM H<9=FGD5797CBGHFI7H=CB
@5B8G75D9 5B8 =89BH=HM F9@=;=CIG GD579G  <5A A5F?G  @9D9F <CGD=H5@ 34 9MCB8
H<9G9B5HIF5@5B87I@HIF5@J5@I9G=HK5G:IB85A9BH5@HC5B7<CFH<=GGHFI7HIF9=B5@9;5@
:F5A9 B5A9@M9L=GH=B;D@5BG5B8GHF5H9;=9G/F65B*@5BB=B; 0=G=CB  *@5B

B=B; &5K   C: H<9 (5H=CB5@ GG9A6@M  B +" 97=G=CB B  +

.85H98   C:H<9*F=A9'=B=GH9FCBDDFCJ=B;H<9.5G?C:8>IGH=B;H<9
!9B9F5@ *@5BB=B; C: +IM (<CB =HM 5B8 =HG J=7=B=HMHC    K=H<5 J=G=CBHC   
920 R. S. Matos et al.

GHF9B;H<9B=B; H<9 DFCDCG5@ 5B8 5GGIF=B; =HG =AD@9A9BH5H=CB  @GC  H<9 DI6@=7 CD9B
GD579GD5F?G ;5F89BG 5J9BI9G GEI5F9G D5H<GK9F97CBG=89F986975IG9H<9M7CBGH=

HIH9 :F99 GD579G 5@@CK=B; GC=@  5=F 5B8 K5H9F 7M7@9G  97C@C;=7 :@IL9G 5B8 DFC79GG9G 
IF65B :56F=7 5FH=7I@5H=CB  DFCACH=B; 7CBH=BI=HM 69HK99B IF65B 5;F=7I@HIF9 GD579G 
97C@C;=75@ G9BG=H=J9 GD579G 5B8 7I@HIF5@ GD579G  7CBGH=HIH=B; GC7=5@ GD579G .<=G B9K
89G=;B DFC79GG 5@GC =B7@I89G F9B9K98 59GH<9H=7 J5@I9G 5B8 7CBH9ADCF5FM GMA6C@G
F9=B:CF7=B;7C@@97H=J9A9ACFM5B8=89BH=HMK<9F9B5HIF9 7I@HIF9 @9=GIF9 DFC8I7H=CB 
5B8DFCH97H=CB5F9DF9G9BH 7CAD@9A9BH5FM 5B87F95H95DI6@=7GD579GB9HKCF?K=H<5
GHFCB;=BJC@J9A9BHC:7CAAIB=H=9G
19 69@=9J9 H<5H GH 79BHIFM IF65B 89G=;B AIGH697@CG9F HC B5HIF5@ 97CGMGH9AG 89

G=;B 5B8 DFC79GG9G  K=H< =HG A9H56C@=GA  K<9F9 K5GH9 =G F97M7@98 =BHC F9GCIF79G :CF
:IHIF989J9@CDA9BH1969@=9J9H<5HH<=GDFCDCG5@89:=B=H9@M7CBHF=6IH9GHCIF65B5B8
@5B8G75D9 GIGH5=B56=@=HM5B8 AI@H=:IB7H=CB5@=HM 69=B; :F5A98=B 5@@ H<9 7CBH9ADCF5FM
GHF5H9;=9G:CFIF65B89J9@CDA9BHH<5H7CB79FBGHCDC@=H=7=5BG6IH5@GCHC7CAAIB=H=9G
5GGIF=B;H<97C9L=GH9B79C:97CBCA=7 GC7=5@5B89BJ=FCBA9BH5@7CB79FBG

References
 -H9=B69F;   "5?=A$/F65B9J9@CDA9BH=BH<9!F95H9F'9?CB;-I6F9;=CBG=5B9

J9@CDA9BH5B? '5B85@IMCB;=HM *<=@=DD=B9G 

 G=5B9J9@CDA9BH5B?0=B<29B!FB7=HM57H=CBD@5BG=5B9J9@CDA9BH5B? 
'5B85@IMCB;=HM *<=@=DD=B9G  
 -5B8<I - -=B;FI , 57<A5BB$ -5B?5F5B 0 FBCIL *!F999BGC@IH=CBG:CF&=J5

6@9=H=9GG=5B9J9@CDA9BH5B? 9J9@CDA9BH5B? '5B85@IMCB;=HM *<=@=DD=B9G 

 
 '5HCG  ,  ,9=BJ9BVTC85 'I@H=:IB7=CB5@=8589 85 *5=G5;9A 9A GD5VC /F65BC [ ,9

:@9LZ9G /B=J9FG=8589 89 PJCF5 G 9 =GG9FH5VTC 5DF9G9BH585 D5F5 C6H9BVTC 8C ;F5I 89
CIHCF9AFH9G9.W7B=75G85*5=G5;9A PJCF5  
 15@8<9=A  .<9@5B8G75D9IF65B=GAF9589F, *F=B79HCBF7<=H97HIF5@*F9GG (9K2CF? 
 
 05@9BH9  '5HCG , 5H=GH5  -=AZ9G * /F65B;F=7I@HIF95K5MHC5GIGH5=B56@9
7CAD@9A9BH5FM69HK99BH<97=HM5B8H<97CIBHFMG=89 #BH<#BH9FB5H=CB5@'I@H=8=G7=D@=

B5FM -7=9BH=:=7 !9C7CB:9F9B79-!'     !F99B 9G=;B 5B8 -IGH5=B56@9 F7<=H97

HIF9  

 0=@>C9B */&G 7CBH=BICIGDFC8I7H=J9IF65B@5B8G75D9G[89G=;B=B;IF65B5;F=7I@HIF9
:CFGIGH5=B56@97=H=9G.F7<=H97HIF5@*F9GG@G9J=9F )L:CF8  
 CB58=9I *5AD5;B9IF65B9 IB5BICJ5DFCDCGH58=D59G5;;=C89@@57=HHRCBN9@@=8

=HCF9 ,CA5  )F=;=B5@:F9B7<DI6@=75H=CB=B

 Travel behavior change patterns under adverse weather
conditions
- A case study from Ho Chi Minh City (HCMC), Vietnam

VU Anh Tuan1[0000-0001-8859-5392],
NGUYEN Thi Thanh Huong2[0000-0001-7099-6278]
1Vietnamese-German Transport Research Centre (VGTRC), Vietnamese-German University (VGU)

drtuan.va@vgtrc.vgu.edu.vn
2Section of Transport Planning and Management, University of Transport and Communications Campus in Ho Chi Minh City, Vietnam

ntthanhhuong@utc2.edu.vn

Abstract. In major cities, road flooding caused by heavy rain and/or high tidal
rise frequently happen and cause road traffic congestions and accidents. To help
design effective traffic management measures for HCMC, this paper focuses on
D VXUYH\ RI WKH SHRSOH¶V WUDYHO EHKDYLRU FKDQJHV XQGHU WKH DGYHUVH ZHDWKHU
conditions. A revealed adaptation interview survey was conducted on 400 road
users in 2018. Typical patterns of travel behavior changes and influential fac-
tors were analyzed based on the surveyed data and by using the Pearson Chi-
square Independence Test. While trip cancelation, delayed departure, waiting
for resuming of trip, route change, and destination change are significant, mode
change is very modest. A road flood causes the changes more strongly than a
heavy rain. Influential factors to such changes are trip characteristics, including
trip purpose, trip length and frequency, and personal characteristics. The results
would be input data for travel demand forecasting model in adverse weather
conditions, and then helpful in formulating traffic management strategies to
mitigate the negative traffic impacts of urban floods and heavy rains.

Keywords: adverse weather, road flooding, travel behavior, Ho Chi Minh City

1 Introduction

Like other major cities in Southeast Asia, HCMC has

been suffering from heavy road flooding caused by
heavy rainfalls and/or tidal rise. It has caused economic
loss due to interrupted freight transport, increased traffic
congestion and traffic accidents in the city [6]. HCMC
is the biggest city in Vietnam where the transport sys-
tem is heavily dependent on private vehicles, especially
motorcycle. Road congestions and accidents are serious
problems and become even more serious under urban Fig. 1 A road under flooding
flood conditions. Given the high motorization level and in Ho Chi Minh City
underdeveloped road infrastructure, the impacts of adverse weather conditions be-

© Springer Nature Singapore Pte Ltd. 2020 921

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_147
922 A. T. Vu et al.

come even more serious since motorcycles are more vulnerable to extreme weather
conditions than private cars and public transport.
It is urgent to formulate and implement a package of dynamic traffic management
measures to cope with the impacts of heavy rain and road flooding for the city. This
ZRUNUHTXLUHVDQXQGHUVWDQGLQJRISRVVLEOHFKDQJHVLQSHRSOH¶VWUDYHOEHKDYLRUXQGHU
the adverse weather conditions. People might adapt to adverse weather differently
depending on culture and transport system [1, 2]. International studies pointed out that
if administrators wish to help drivers in adjusting their on-road behavior in adverse
weather conditions, the methods should be local and technical by nature [5]. There
were many studies on this topic in North America and Europe. Some focused on one
or some types of bad weather (such as rain, flood, wind, extreme cold, overheating,
etc.) affecting one or some types of users (including public transport riders, pedestri-
ans, motorcyclists, and car users) [3]. However, the number of studies conducted in
developing countries is rather limited [1]. In Bangkok, Thailand, under heavy rain,
pedestrians attempted to walk throughout the flooded areas while many local people
were impossible to leave home to work, car users had to use motorcycle taxi to get to
the nearest train stations, and train stations become overcrowded [Bangkok post pa-
per, 2016]. In Malaysia, to avoid flooded routes, travelers often select alternative
roads that are longer [7]. Those papers pointed out the patterns of travel behavior
change under rains and flooding conditions in other countries.
In Vietnam, there was only one previous research that examined the patterns of
change in Hanoi City [4]. In 2017-2018, the Vietnamese-German Transport Research
Centre at the Vietnamese-German University conducted a research project entitled
³7UDIILF 0DQDJHPHQW 0HDVXUHV LQ 5HVSRQVH WR 8UEDQ )ORRGLQJ &DXVHG E\ +HDY\
5DLQDQG7LGDO5LVH´One of the study tasks was to conduct travel interview surveys
and analyses to capture possible changes RI SHRSOH¶V WUDYHO EHKDYLRU XQGHU WKH Dd-
verse weather conditions in HCMC. This paper aims to present the results of these
surveys. The surveyed data is also used to test if the weather condition determines the
likelihood of a travel behavior change and influential factors to change, such as trip
purpose, information availability, and individual characteristics. The findings would
be used to calibrate travel demand simulation model to forecast travel changes under
adverse weather conditions. The results would be helpful in design and testing of
possible measures to manage the traffic system to effectively respond to bad weather
conditions.
This paper is structured in 4 chapters. The first chapter introduces urban flooding
issues, previous studies and the objective of this paper. The second chapter describes
methodology and data collection. The third chapter presents the results of analysis on
patterns of change and influential factors. The fourth chapter concludes with solution
recommendations and further works.

2 Methodology and Data collection

Under the research project presented in the introduction, a formal process was applied
to design interview surveys. First, two focus group discussions were conducted with
Travel behavior change patterns … 923

administrators, experts and residents. Based on the results of focus group discussions,
weather conditions, typical changes in travel, trip characteristics and personal charac-
teristics were considered important. Second, based on the review of previous studies
and the results of focus group discussions, a questionnaire was designed for a re-
YHDOHGDGDSWLRQVXUYH\WRFDSWXUHWUDYHOHU¶VEHKDYLRUFKDQJHVLQUHVSRQVHWRDGYHUVH
weather conditions and to prepare input data for traffic model development in the
overall research project. Each respondent was asked to reveal if and how they would
eventually change their travel in four cases, namely (1) encountering a heavy rainfall
just before the trip, (2) encountering a heavy rainfall during the trip, (3) encountering
a road flood just before the trip, and (4) encountering a road flood during the trip. For
each trip purpose (e.g.., commuting, shopping, leisure, and other), each respondent is
given a number of choices, including (1) cancelation, (2) postponement, (3) destina-
tion change, (4) route change, (5) mode change to car, motorcycle, bicycle, e-bike,
and bus, and (6) no change. About 400 respondents were successfully interviewed,
including 160 motorcyclists and e-bike riders, 80 car drivers, 80 bus riders and 80
cyclists. To ensure an optimal correspondence between the surveyed samples and the
actual population, the sample were weighted. The observations were weighted to
match the sample distribution and the population distribution.
The analysis in this paper includes two parts. In the first part, change patterns were
comparatively analyzed by trip purpose, namely commuting trips, shopping trips, and
leisure trips. The second part was attempted to test hypotheses that the weather might
determine the likelihood of a change in travel behavior and changes in travel behavior
because of weather conditions might be associated with trip characteristics (such as
purpose) and personal characteristics. The Pearson chi-square independence tests
were deployed for the hypothesis test. The results of the descriptive analysis and the
tests are presented in the following chapters.

3 Analytical results of travel behavior change patterns under

adverse weather conditions

3.1 Patterns of behavioral change by trip purpose

The results are presented in Table 1. Change patterns were comparatively examined
for three trip purposes, including commuting (C), shopping (S), and leisure (L). For
commuting trips, while changes in departure time are common to both heavy rain and
flood, changes in route are common to flood only. Postponing the trip to a later mo-
ment is especially a common practice; more than three persons in ten would postpone
their trips in case of heavy rain and/or flood. Almost half of the respondents changed
the route in case of flood. Such changes can be explained by the fact that people try to
avoid traffic jams and to be safe while traveling by changing the route and the depar-
ture time. Since work or school location is unchangeable, very low percentages of the
respondents changed the destination. Because the available or accessible modes are
limited, few percentages of the respondents changed mode. However, changes in
destination would be more feasible if telecommuting alternatives, satellite offices, and
924 A. T. Vu et al.

e-learning are available. The most prevalent changes in commuting trips are changes
in the departure time and changes in the route chosen. For shopping trips, trip cancel-
lations are much more common, about 30%-60% respondents canceled their shopping
trips. Trip postponements are also very common. These behaviors are different from
commuting trips and the differences could be explained by fact that it is much easier
to postpone or cancel shopping activities than work or study activities.

Table 1. Percentages of travel behavior changes by adverse weather condition and trip purpose
Heavy rain Heavy rain Flood Flood
Weather Condition
before the dep. during the trip before the dep. during the trip
Trip purpose C S L C S L C S L C S L
Trip cancelation 4.2% 39.0% 45% 1.6% 11.8% 0.0% 7.1% 59.0% 59.8% 1.4% 18.2% 18.2%
Delayed dep. time 33.5% 30.1% 29% n.a n.a n.a 22.2% 10.5% 16.3% n.a n.a n.a
Waiting for res. n.a n.a n.a 45.1% 53.8% 36.1% n.a n.a n.a 39.9% 37.6% 37.6%
Destination change 1.1% 3.4% 4.0% 0.9% 3.7% 8.5% 0.1% 8.7% 4.7% 0.4% 9.1% 9.1%
Route change 7.6% 0.1% 2.1% 2.4% 2.4% 8.2% 44.2% 5.6% 10.2% 27.5% 13.6% 13.6%
Mode change 2.5% 0.5% 2.6% 1.4% 0.0% 0.7% 4.9% 0.0% 2.6% 0.1% 0.1% 0.1%
None 55.3% 26.9% 17.6% 48.6% 28.3% 46.5% 21.5% 16.2% 6.4% 30.7% 21.4% 21.4%
Total 100% 100% 100% 100% 100% 100% 100% 100% 100% 100% 100% 100%
1RWHV³&´ FRPPXWLQJWULS³6´ VKRSSLQJWULSDQG³/´ OHLVXUHWULS³QD´ QRWDSSOLHGGHS GHSDUWXUHUHV UHVXPLQJof the trip

Similar to the case of shopping trips, trip postponements and cancellations are the
most common for leisure trips. About 25%-60% of the respondents postponed their
leisure trips and 15%-35% canceled their leisure trips. Destination changes, routes
changes and mode changes are more frequent for leisure trips than shopping trips.
However, nobody postponed their leisure trips in case of heavy rain during the trip.
There seems to be behavioral homogeneity between leisure trips and shopping trips
because both activities are non-obligatory and more flexible than study and work
activities.

3.2 Influential factors to the behavioral changes

,Q WKH SUHYLRXV VHFWLRQ FKDQJHV LQ SHRSOH¶V WUDYHO EHKDYLRU LQ UHVSRQVH WR DGYHUVH
weather conditions are dependent on the context and the trip purpose. Furthermore, it
is important to understand how such changes might be influenced by factors related to
socio-demographic characteristics of the travelers, vehicle ownership status, and trip
characteristics. Based on the interview data, Pearson chi-square independence tests
were deployed to determine significantly influential factors.
The results of the analysis have confirmed the dependency of behavioral changes
on three groups of factors, including individual characterizes (e.g., gender, age, vehi-
cle ownership, personal income, multiple job status, and the main job), the context of
adverse weather condition (i.e., heavy rain/flood before or during the trip), and trip
features (e.g., trip purpose, mode used, trip length, and trip frequency). The null hy-
pothesis of independence is rejected for all behavioral changes with a p-value smaller
WKDQ&UDPHU¶VFRQWLQJHQF\FRHIILFLHQWUHIHUUHGWRDV&UDPHU¶V9ZDVFDOFu-
lated for determining the strength of the association between the factor under consid-
eration and the behavioral change. The results of Pearson chi-square independence
tests show that gender and age are highly influential to the behavioral changes under
Travel behavior change patterns … 925

KHDY\UDLQDQGIORRGFRQGLWLRQVIRUFRPPXWLQJDQGVKRSSLQJWULSVEHFDXVH&UDPHU¶V
V is about 0.4-0.5. Female and older people tend to make trip cancelation, time
change and mode change much more than male and younger people. Similarly, car
and motorcycle ownership may have stronger influence to the behavioral changes as
&UDPHU¶V9LVDERXW-0.4. The motorcycle and bicycle users tend to switch to car in
case of heavy rain. As shown in Table 2, trip length and trip frequency have the
VWURQJHVWLQIOXHQFHVDVWKH&UDPHU¶V9LVDERXW-0.7. The longer and more frequent
the trip the higher likelihood of change.

Table 2. Dependence of travel behavioral changes on trip length and trip frequency
Weather Trip distance Trip frequency Results are based on
Behavioral change nonempty rows and
condition Ȥ df Sig. Cra Ȥ df Sig. Cra
Trip cancelation 3070 10 .000* 0.691 3537 12 .000* 0.742 columns in each inner-
Heavy most suitable.
Time change 3104 10 .000* 0.695 3150 12 .000* 0.700
rain before *. The Chi-square statistic
Destination change 3124 10 .000* 0.697 3192 12 .000* 0.705
the depar- is significant at the .05
Route change 3144 10 .000* 0.700 3211 12 .000* 0.707
ture level.
Mode change 3126 10 .000* 0.698 3191 12 .000* 0.705
Heavy Trip cancelation 2956 10 .000* 0.678 3191 12 .000* 0.705 b. More than 20% of cells
rain Time change 3358 10 .000* 0.723 3240 12 .000* 0.710 in this suitable have
during the Destination change 3043 10 .000* 0.688 3123 12 .000* 0.697 expected cell counts less
trip Route change 3161 10 .000* 0.701 3169 12 .000* 0.702 than 5. Chi-square results
Mode change 3161 10 .000* 0.701 3107 12 .000*,b 0.695 may be invalid.
c. The minimum expected
Trip cancelation 3103 10 .000* 0.695 3743 12 .000* 0.763
Flood cell count in this suitable
Time change 3097 10 .000* 0.694 3170 12 .000* 0.702
before the is less than one. Chi-
Destination change 3094 10 .000* 0.694 3178 12 .000* 0.703
departure square results may be
Route change 3075 10 .000* 0.692 3450 12 .000* 0.733
Mode change 3073 10 .000* 0.692 3155 12 .000* 0.701 invalid.
Trip cancelation 3021 10 .000* 0.686 3264 12 .000* 0.713
Flood Time change 3222 10 .000* 0.708 3259 12 .000* 0.712
during the Destination change 3071 10 .000* 0.691 3138 12 .000* 0.699
trip Route change 3074 10 .000* 0.692 3153 12 .000* 0.701
Mode change 3049 10 .000*,b,c 0.689 3096 12 .000*,b,c 0.694

4 Conclusion

In this paper, the patterns of travel behavioral changes due to adverse weather condi-
tions have been examined and factors to the changes have been determined. Trip can-
celation and delayed departure are very common if the adverse weathers occur before
the trip. Waiting for resuming of the trip is very common in case of the adverse
weathers occurring during the trip. Route change is very common only to commuting
trips under the flood condition and becomes quite significant to shopping and leisure
trips. Destination change is quite significant to only shopping and leisure trips. Mode
change is very modest in all cases. Urban flood seems to have stronger influences to
the behavioral changes than heavy rain. Such patterns of behavioral change would be
input data for calibrating travel demand simulation model that can be used to forecast
demand changes in adverse weather.
Besides, specific issues coming out from this paper shall be useful to the formula-
tion of traffic management strategies to addressing adverse weather in HCMC. First,
SHRSOH¶VWUDYHOEHKDYLRUFKDQJHVDUHKLJKO\GHSHQGHQWRQWKH PRPHQWVRIUHFHLYLQJ
relevant information. If they are informed prior to departure time, they would re-
schedule their journey in terms of departure time, destination, and route. Pre-journey
926 A. T. Vu et al.

weather forecast, traffic information, and warning system may be an important solu-
WLRQWRLQIOXHQFHSHRSOH¶VWUDYHOGHFLVLRQ-making and thus would help better manage
WKHURDGWUDIILFQHWZRUN6HFRQGURDGIORRGLQJKDVFDXVHGODUJHFKDQJHVLQSHRSOH¶V
travel behavior, especially it strongly causes route changes. Therefore, alternative
routes need to be available in case of road flooding. Road planning and management
shall pay attention to this point. Third, mode change is very modest, except for com-
muting trips under road flooding prior to the trip. This implies that either the number
of alternative modes are very limited (i.e., the majority of people are using motorcycle
for daily travel and very few people are using car and bus) or people may resist to
mode change because of their habitual mode usage. This issue should be investigated
in another paper.
In conclusion, this paper contributes to the literature by empirically examining the
W\SLFDOFKDQJHVLQSHRSOH¶VWUDYHOEHKDYLRULQUHVSRQVHWRDGYHUVHZHDWKHUFRQGLWLRQV
in a Vietnamese city. The strong dependency of the behavioral adjustments on activity
type and individual characteristics provides policy makers with a deeper understand-
ing of how adverse weather conditions may affect the traffic and what might be im-
plemented to mitigate the negative impacts on the traffic system. However, to support
further the design of effective traffic management measures, additional studies need
to be carried out. Further works might include assessing the impacts of adverse
weather conditions on the operation of the road traffic network, changes in travel cost,
travel time and traffic safety. As such changes might influence road user decisions,
stated adaption behavior survey shall be conducted in a next step.

5 Acknowledgements

The authors would like to acknowledge the financial support of the Ministry of Edu-
FDWLRQ DQG 7UDLQLQJ 02(7 IRU WKH RYHUDOO UHVHDUFK SURMHFW ³7UDIILF 0DQDJHPHQW
0HDVXUHV LQ 5HVSRQVH WR 8UEDQ )ORRGLQJ &DXVHG E\ +HDY\ 5DLQ DQG 7LGDO 5LVH´
Thanks to this sponsorship, the research team was able to conduct the questionnaire
survey on revealed weather event adaptations by road users in HCMC in 2018.

6 References
1. Maria Pregnolato and al., 2017. The impact of flooding on road transport: A depth-
disruption function. http://dx.doi.org/10.1016/j.trd.2017.06.0201361
2. Mario Cools and al., 2010. Changes in travel behavior in response to weather conditions:
whether type of weather and trip purpose matter? https://trrjournalonline.trb.org/doi/ref/10.3141/2157-03.
3. Chengxi Liu and al. (2015). Investigating the impacts of weather variability on individu-
DO¶VGDLO\DFWLYLW\±travel patterns. Transportation Research: Policy and Practice 82, 47-64.
4. Anh Tuan VU and al. (2017). The framework of traffic management dealing with bad
weather conditions in developing countries. EASTS 2017, TP. HCM, Vietnam.
5. M. Kilpeläinen and al., 2007. Effects of weather and weather forecasts on driver behavior,
Transportation Research:Traffic Psychology and Behavior 10(4):288-299 · July 2007
6. Alfelor, Billot, El Faouzi and Pisano. 2013. ³75%$QQXDO0HHWLQJ3DSHU5HYLVHGIURP2ULJLQDO6XEPLWWDO´
7. Othman, M. H., & Abdul Hamid, A. . (2014). Impact of Flooding on Traffic Route Choic-
es. SHS Web of Conferences, 11, 01002.doi:10.1051/shsconf/20141101002
 Traffic Impact Assessment of Infrastructure
Development Projects for Sustainable Urban Growth

Trinh Dinh Toan 1[0000-0002-9113-1975] and Dao Van Dong 2[1111-2222-3333-4444]

1
Department for Transportation Engineering,
ThuyLoi University, 175 TaySon, DongDa, Hanoi
Trinhdinhtoan@tlu.edu.vn
2
University of Transport Technology
54 Trieu Khuc, Thanh Xuan, Hanoi.

Abstract. Infrastructure development projects in urban areas often create im-

pacts on the urban life. As the size of the projects increases, undesirable effects
such as traffic congestion, traffic accidents, and environmental problems may
become unacceptable to the community. In Vietnam, there witnesses rapid de-
velopment of infrastructures, especially high-rise buildings in the big cities.
These infrastructures generate high traffic demand and may cause serious traffic
congestion on the surrounding transport network, thus assessment of traffic im-
pacts (TIA) imposed by the projects is essential in controlling spontaneous de-
velopment for sustainable urban growth. For TIA to be actuated, there should
be a statutory document that institutionalizes the requirements on conducting
TIA in the project appraisal, and a technical framework that guides the struc-
ture, content, procedure and method for TIA execution. However, in Vietnam,
such essential documents are not yet available. This paper reviews TIA imple-
mentation in some countries, addresses issues in TIA execution practice, figures
out the key contents of a TIA report, and proposes solutions to actualize the
execution of TIA for development projects for sustainable urban development
in Vietnam.

Keywords: Traffic Impact Assessment, Development Project, Sustainable Ur-

ban Growth.

1 Introduction

Infrastructure development projects, depending on their nature and magnitude, often

impose impacts on the environment, especially under the urban context. As the size of
the project increases, undesirable effects such as traffic congestion, traffic accidents,
and environmental problems get worse and may become unacceptable to the commu-
nity. It has been realized in some cases that attempts to push physical growth impose
devastating social and environmental consequences, thus objectives in sustainable
urban development are spoiled in the long run (Fig.1).

© Springer Nature Singapore Pte Ltd. 2020 927

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_148
928 T. D. Toan and D. V. Dong

Fig. 1. Uncontrolled high-rise buildings and impact on urban growth.

Transport networks form the structure of a city, being one of principal infrastructures
that play essential functions for sustainable urban development. Most of infrastructure
developments generate additional traffic, therefore it is necessary to address the im-
pacts of traffic, especially private vehicles on the background traffic and on the envi-
ronment, in the land-use planning decision-making process.
Traffic Impact Assessment (TIA) is a study carried out to anticipate the effects of
generated traffic that a proposed development will have on the surrounding transport
network, and recommend necessary measures to mitigate the negative impacts [1].
TIA is an important document in helping authorities make decisions on land use and
urban planning. TIA can be used to evaluate whether a development proposal is ap-
propriate and what improvements in transport facilities should be made in the long
run to maintain sustainable development.
,Q WKH LQWHUQDWLRQDO FRQWH[W ³7UDIILF ,PSDFW $VVHVVPHQW´ may be used inter-
changeably with other terminologies VXFKDV³7UDIILF,PSDFW$QDO\VLV´³7UDIILF,m-
SDFW6WXG\´DQG³7UDQVSRUW,PSDFW$VVHVVPHQW´HWF:KLOHPDQ\RIWKHVHUHIHURQO\
to traffic, a standard TIA study should address the impacts of all modes of transport
related to the development project under consideration [2].
Primary objectives of a TIA study may include: (i) 7RLGHQWLI\WKHGHYHORSPHQW¶V
impacts on the surrounding transport network and recommend necessary measures to
mitigate its negative impacts; (ii) 7R HQKDQFH WKH GHYHORSPHQW¶V RYHUDOO DFWLYH DQG
mass transport connectivity, accessibility and convenience in relation to the wider
transport network; and (iii) To determine the transportation requirements of a new
development / redevelopment, and propose adequate and appropriate design features,
facilities, and infrastructural improvements to meet future transport demand [1].

2. TIA Implementation Practices

2.1 TIA Implementation in Some Countries

United Kingdom. There are a number of documents providing guidance on preparing
TIA reports that have been produced by a wide variety of organizations and authori-
Traffic Impact Assessment of Infrastructure … 929

ties in the United Kingdom: Planning Policy Guidance Note 13 - Transport indicates
requirements for sustainable transport to be considered for TIA; Transport Assess-
ment and Implementation - A guide (Scottish Executive 2005) provides controls plan-
ning legislation in Scotland; Guidance on Transport Assessment (2006) provides
guidance for the production of transport assessments and transport statements [2].
United States. The relevant guides and reports produced by the United States In-
stitute of Transportation Engineers (ITE) include Transportation Impact Analyses for
Site Development (ITE 2006) summarizes ITE proposed recommended practice. Oth-
er documents that provide supplement guidance for TIA implementation include Trip
Generation Handbook (ITE 2003), and Parking Generation, 3rd Ed. (ITE 2004) [2].
Singapore. Guidelines for preparation of traffic impact assessment reports for de-
velopers and transport professionals were firstly issued in 2003. The Guidelines cov-
ered all the information required to be included in TIAs. The Guidelines was updated
subsequently in 2007 and 2011. Recently the new version [1] introduced Walking &
Cycling Plan (WCP) to encourage developers to consider pedestrian / cyclist connec-
tivity and provide supporting facilities upfront in the development design stage.
Thailand. TIA has been initiated in Thailand since early 1990s by a number of
Acts legalized to control land development projects such as Building Control Act of
1992, Land Development Act of 1992 [3]. The first TIA introduced under the Acts,
however, focused on environmental problems rather than traffic engineering: the im-
pacts of traffic were evaluated through the law on environmental impact study (EIS).
For example, developers of projects having 2,000 square meters of gross floor areas
or occupying more than 300 parking units were required to conduct the EIS. Howev-
er, full TIAs were required in Bangkok for public transportation projects, high-rise
buildings or development proposals in traffic-sensitive areas.
The Philippines. Similar to Thailand, in the Philippines in 1990s there was legisla-
tion requiring the conduct of an EIS for environmentally critical projects or projects
located in environmentally critical areas. Depending on the type of projects, EIS stud-
ies also considered certain content on traffic impacts, however, they have never in-
cluded a standard TIA. Given the lack of a clear legal requirement on TIA, the study
of traffic impacts of a development had usually been superficial [3].
Since early 2000s, TIA implementation has been motivated via three strategies:
(1) through the Environmental Impact Assessment (EIA) framework; (2) through
national legislation; and (3) through local government unit ordinance. With this stra-
tegic arrangement, it has been expected that all development levels in the Philippines,
from national to municipal levels, can be handled. Nevertheless, strategic proposals
for TIA implementation have yet to see actualization given the lack of clear legal and
institutional requirements for conducting TIA in practice [3].

2.2 Critical Issues in TIA Implementation

In developed countries such as the United States, United Kingdom, Australia and
Singapore, the concept of TIA has been introduced and legalized soon, routinely and
successfully implemented. In developing countries in South-East Asia including Thai-
land, the Philippines, and Indonesia, however, even though the TIA concept has been
930 T. D. Toan and D. V. Dong

sooner or later introduced for a number of decades, TIA execution achievements have
yet been limited due to many obstacles: there is weak institutional enforcement for
translation of TIA legal requirements in to reality, there is wide variation in percep-
tion of content and scope of TIAs, there is little advice on how TIAs should be under-
taken, and there is lack of opportunities to integrate developments into transport net-
works, etc. It has been identified that the limitations in TIA execution in South-East
Asia are the results of a number of challenges, notably institutionalization and legisla-
tion, unavailability of standard references, and lack of traffic engineering knowledge
[3].
Legislative and institutional issues: The developing countries often face legislative
and institutional problems in conducting TIA. In Thailand for example, it is clearly
stipulated in the law that all projects having 2,000 square meters of gross floor areas
or occupying more than 300 parking units are required to conduct an EIS. However,
some developers avoid the EIS study by dividing the project into a number of smaller
projects for easy approval. Likewise, in the Philippines, most projects consider traffic
component under the EIS context, where the impacts of traffic are often overlooked,
and TIA studies are often superficially carried out. For example, in some cases, in-
stead of conducting a comprehensive TIA, only geometric design features are re-
viewed [3].
Lack of standard technical references: Although many TIA standards that have
been developed and used widely in advanced countries, they are not simply tailored
for use in developing countries, given significant differences in travel characteristics
and socio-economic conditions. As a result, TIA implementation in developing coun-
tries suffer critical technical problems such as inappropriate trip generation rate, poor
internal circulation, overlooked pedestrians and cycling, and unattractive public
transport facilities, etc.
Lack of domain knowledge: There is a few experts or experienced engineers having
sufficient traffic engineering knowledge for TIA execution in the government and
consultants. In private sectors the situation is even worse. Given the limited
knowledge in traffic engineering to prepare and review the TIA documents in the
local staffs, in many cases the TIA reports are simply passed on to the central offices,
leading to overloaded central interface, thus the progress delayed and the quality re-
duced accordingly.

2.3 TIA in Vietnam

The issues (weak institutional enforcement, disintegrated local authorities, lack of
standard technical references, and lack of domain knowledge) discussed above seem
to be pervasive in developing countries who have been actually conducting TIA. It
can be said that Vietnam will encounter similar problems. In Vietnam, for the time
being, a legal requirement on conducting TIA for the infrastructure development is
not yet available. Like Thailand and the Philippines in the initial period, the impacts
of traffic in Vietnam are only considered as a component under the Environmental
Impact Assessment (EIA). The power in reviewing EIAs is currently delegated to
Ministry of Natural resources and Environment (MONRE), while the Ministry of
Traffic Impact Assessment of Infrastructure … 931

Transport (MOT) LVVHWRQHRIWKH³LQ-OLQH´PLQLVWULHV>4]. Unfortunately, there is little

HIIHFWLYHFRRUGLQDWLRQEHWZHHQUHODWHGLQVWLWXWLRQVLQWKHSURFHVVRISURMHFWV¶SUHSDUa-
tion, appraisal and implementation [5].
Given the lack of legal binding on TIA, issuance of construction permits of many
projects do not require an obliged assessment of traffic impacts via a standard TIA.
Similarly, given the lack of a technical guidance, the study of traffic impacts of de-
velopments (if any) do not gain a necessary depth. Nevertheless, the perception on the
necessity and criticality of TIA is getting clearer nowadays by experts and manage-
ment authorities and the public [6].

3. Structure and contents of a TIA Report

Although the TIA content differs according to the type of development, a standard
TIA report includes the following structure and basic contents:
x Specific attributes of the development project, in particular transport features:
study purpose and objectives, description of the site and study area;
x Conditions of existing traffic and transport infrastructures surrounding the de-
velopment site: existing traffic conditions in the area of development;
x Forecasted traffic: determination of assessment years, trip generation, distribu-
tion, modal split, and traffic assignment. Presentation of forecasted results;
x Impacts of project on the future transport network: assessment of impacts of
development traffic on the level of service, and evaluation of critical junctions;
x Internal traffic: provide adequate facilities for internal traffic and circulation,
including parking, pick-up/drop-off provision, and truck deliveries;
x Mitigation measures to mitigate negative impacts: public transport, pedestri-
ans, cyclists, site access. Traffic management and traffic operation plan.
The overall process in conducting a TIA is presented in Fig 2. This, however,
should be customized under various contexts depending on the project type and nature
of background transport.

1. Development Project

2. Survey and analyse the 3. Forecast the future

existing transport conditions traffic demands

4. Analyse the future

transport conditions

No Yes
5. Internal circulations Conditions 6. Consideration for
and mitigation measures Acceptable? approval

Fig. 2. Overall process of TIA

932 T. D. Toan and D. V. Dong

4. Conclusions

In Vietnam, there witness rapid development of infrastructures, especially high-rise

residential buildings in Hanoi and HoChiMinh cities. These infrastructures generate
high traffic demand. Given the overlook in evaluation of traffic impacts in the project
appraisal process, there are many cases the project causes unbearable traffic conges-
tion problems on the surrounding network as soon as it enters the operation, thus se-
verely affect sustainability of the urban development.
In developed countries, TIA is often statutorily required before an investment deci-
sion is made. In many developing countries, including the South-East Asia, TIA has
been institutionalized and executed to a great extent. In Vietnam however, a formal
TIA is not yet institutionalized, thus TIA is not requested as an integral part of the
project proposal that the developer must submit to planning authorities in seeking for
approval. In other words, there is actually not yet legal obligation in assessment of
WUDIILFLPSDFWVLQWKHSURMHFWV¶decision-making process of urban infrastructure devel-
opments.
For TIA to be actuated, there should be a statutory document that institutionalizes
the requirements on conducting TIA in the project appraisal, stipulates the type and
size of the project that need to conduct TIA, the role of the government and obligation
of the related stakeholders. Apart from that there should be a technical guide that
prescribes the structure, content, procedure and method for TIA to help developers
prepare an impartial TIA report on the proposed development. This is among vital
steps that significantly contribute towards sustainability for urban growth in Vietnam.

References
1. Land Transport Authority. Transport Impact Assessment Guidelines for Devel-
opments, Singapore (2017).
2. Transport impact guidelines for site development: Literature review. Land
Transport New Zealand, Research Report 327 Part II, New Zealand (2007).
3. Karin, L., Cresencio, M.M., Kazunori, K., and Atsushi, F. The implementation
of traffic impact assessment in Southeast Asian Cities: Case study of Thailand
and the Philippines. Journal of the Eastern Asia Society for Transportation Stud-
ies, Vol. 6, pp. 4208 - 4223 (2005).
4. Hai, N. Environmental Impact Assessment in Vietnam. Presentation at the Min-
istry of Natural resources and Environment MONRE (2015).
5. Toan, T.D., Dong, D.V. Integrated Transport Planning for Sustainable Urban
Development ± 6LQJDSRUH¶$SSURDFKDQG/HVVRQVIRU Vietnam. Congrès Inter-
national de Géotechnique - Ouvrages - Structures (CIGO 2019).
6. http://www.baogiaothong.vn/can-danh-gia-tac-dong-giao-thong-cac-du-an-cao-
oc-do-thi-d253574.html, last accessed 2019/02/26.
 Reasons for cyclists denying new-built bicycle lanes

Nguyen Duc-Nghiem[1], Nguyen Viet-Phuong[2] and Hisashi Kubota[3]

1,2
Highway and Traffic Engineering Department, National University of Civil Engineering,
No.55 Giai Phong street, Hai Ba Trung district, Hanoi, Vietnam
3
Graduate School Science & Engineering, Saitama University (255 Shimookubo, Sakura
Ward, Saitama, 338-8570, Japan)
nghiemdhxd@gmail.com

Abstract. Recently, cycling has been considered as one of the most potential
sustainable transport modes in cities. Therefore, a great deal of bicycle facilities
has been built in them. One of the most popular facilities could be the bicycle
lanes (BL). However, in some cases, cyclists tend to deny using the new-built
BLs. This can be viewed as a significant waste of resources. The present study
aims to indicate some factors that prevent cyclists from using the BLs. To do
so, initially, the effects of the impacts factors on the facility choice of the cy-
clists were investigated, based on an objective camera-observation survey in the
daily cycling environment. Then, the impact trends were validated and ex-
plained based on an independent subjective stated preference survey. The re-
sults showed that perceived safety was the most important aspect that decides
whether they choose the BL or not. Several factors, including high vehicle vol-
ume, the existence of real-time stopping vehicles on the BL, gender, narrower
width of traffic lanes, narrower width of BL, and wider sidewalk were found
that negatively affected the probability of the cyclists on their choice to ride the
BLs. These findings can help developers, planners, and designers to adopt more
reasonable investment decisions as well as better designs while developing bi-
cycle facilities.

Keywords: Deny Bicycle Lane, Bicycle Lane Preference, Facility Preference

1 Introduction

Facing with many challenges such as air pollution and traffic congestion that resulted
from unsustainable transportation development in the past, many cities have been
introduced some measures for promoting sustainable transportation for the future
cities. In which, increasing bicycle ridership has been considered as one of the most
promising alternatives. Since transportation infrastructures built in the past mainly
serve for motor vehicle, but not for bicycle. Therefore, the central task of many city
governments now is to develop bicycle infrastructures first.
Because of the space limitation in the cities, it is normally difficult to build sepa-
rate bicycle paths. Instead, on-street bicycle lane (BL) could be the most popular bi-
cycle facility which has been built. BL is widely defined a part of a road that is sepa-
rated by a line from the rest of the road, for the use of cyclists [1]. Normally, cyclists

© Springer Nature Singapore Pte Ltd. 2020 933

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_149
934 N. Duc-Nghiem et al.

tended to prefer the BLs. However, in some cases, they are diverted from using them,
for example in case of San Francisco [2] and Texas [3, 4]. This can be viewed as a
significant wastage of resources.
In the literature, several factors affecting F\FOLVWV¶SUHIHUHQFHWR%/ have been indi-
cated. Aultman-Hall, L. and Adams, M. [5] found that cyclists often prefer BLs and
other on-street facilities in minor streets. Related to infrastructure conditions, the
presence of bus stops [6], the existence of nearby parking facilities [7] and street func-
tion [8] were found that DIIHFW F\FOLVWV¶ IDFLOLW\ SUHIHUHQFH Traffic condition factors
such as traffic volume can also DIIHFWF\FOLVWV¶SUHIHUHQFHWR BLs [5, 6, 9]. Some other
VWXGLHV LOOXVWUDWHG F\FOLVWV¶ FKDUDFWHULVWLFV VXFK DV gender [10, 11], age [12] and cy-
cling experience [4] showed the significant correlations with their facility preference.
To some extends, although previous studies illustrated several factors that can neg-
atively affect on BL usage, a comprehensive study focusing deterrents for cyclists to
use BLs has not been found in the literature; and the explanation for the reason why
cyclists denying BLs also had its limitation. Moreover, the above studies shared a
similar point is that they mainly based on self-stated preference surveys to validate the
hypothetical trends, but did not consider the real-time conditions while the cyclist
deciding to choose BL in the daily cycling practice environment.
Take this into account, the current study investigated a series of factors, including
exterior real-time factors that can affect F\FOLVWV¶IDFLOLW\FKRLFHLQGDLO\F\FOLQJSUDc-
tice environment that aims to find out what preventing cyclists from using BLs. An
effort was then made to validate and explain the exposed impact trends by an inde-
pendent subjective stated preference survey. Results of this study can be beneficial to
developers, planners, designers, etc. to adopt more reasonable investment decision as
well as better design to avoid abandoned BLs the cities.

2 Methods

2.1 Survey and Data Extraction

Camera Observation Survey

The camera observation surveys were carried out in streets in Saitama city, Japan. All
selected sites had a typical cross-section of a medium size urban street with 2 or 3
lanes. On these streets, both BLs and sidewalks were available for cyclists to ride.

Data Extraction
This study investigated seven independent variables which are shown in Table 1.
Whereas, variables related to real-WLPH WUDIILF FRQGLWLRQ DQG F\FOLVW¶V FKDUDFWHULVWLFV
were extracted from the recorded videos, geometric dimensions of the surveyed seg-
ments were measured on the sites.

Questionnaire Survey
Questionnaires were designed to find out the main reason why cyclist prefer or deny
Reasons for cyclists denying new-built bicycle lanes 935

Table 1 Considered factors for BL usage analysis

Variable Explanation
SW.Width Effective sidewalk width (m)(a)
TL.Width Traffic lane width (m)(a)
BL.Width Bike lane width (m)(a)
Veh.Vol. Vehicle volume of the nearby traffic lane per hour (veh./h.lane) (b)
Stop.Veh. 6WRSSLQJ YHKLFOH ³\HV´ LI ZLWKLQ WKH RQH-minute interval there was any vehicle
VWRSSLQJRQWKH%/³QR´RWKHUZLVH(c)
Wait.Veh. :DLWLQJ YHKLFOH ³\HV´ LI ZLWKLQ WKH RQH-minute interval there was at least one
YHKLFOHZDLWLQJIRUWUDIILFOLJKWRUSHGHVWULDQ³QR´RWKHUZLVH(c)
Gender Gender of cyclist
Note:
(a)
Values of variables were measured along the surveying segment.
(b)
The volumes were counted as the vehicles passed an assumed reference point on the video.
(c)
Because of the limitation of camera view distance, stopping and waiting events could only be ob-
served within about 40m along the considering segment (not the whole segment).

BLs and to validate the objective impact trends found from the camera survey. Ques-
tionnaires were posted in mailboxes of families which were located in the same dis-
trict where camera surveys were conducted. In total, 3,600 questionnaires were posted
and 421 of these were answered and sent back to our office by the respondents.

2.2 Data Analysis

Assessing and Comparing BL Usage Level

To assess and compare the level of BL usage among different groups, this study used
a figure called ³%/8VHG3URSRUWLRQ´%/83ZKLFKLVGHILQHGDV
௡ಳಽ
‫=ܷܲܮܤ‬ (1)
௡೟೚೟ೌ೗

Where: nBL is the number of cyclists who chose BL and ntotal is the total number of
cyclists in the considering group. If BLUP is big this suggests that the BL attracts
cyclists well and vice versa.

Proportion Test
In the analyses, the respondents were divided into different groups (e.g. man group,
woman group). Each group had a specific value of BLUP. To clarify whether the
differences in BLUP values were statistically different or just randomized the propor-
tion tests were conducted.
Given two groups with the number of respondents are n 1 and n2. The numbers of
cyclists who choose BL are x1 and x2. We have two proportions BLUP1 = p1 = x1/n1
and BLUP2 = p2 = x2/n2. Statistically, we can say the difference between two groups d
= p1 - p2 is normally distributed with mean = 0 and variance Vd is as follows
ଵ ଵ
ܸௗ ൌ ቀ ൅ ቁ ‫݌‬ሺͳ െ ‫݌‬ሻ (2)
௡భ ௡మ
936 N. Duc-Nghiem et al.

௫భ ା௫మ
Š‡”‡ǣ‫ ݌‬ൌ  (3)
௡భ ା௡మ

So, z = d/Vd is normally distributed, with mean = 0 and variance = 0. In other

words, z2 is chi-squared distributed. The proportion test is now equivalent to Chi-
squared test. All analyses were conducted in R software version 3.2.5 [13].

3 Results

Overall, there were 931 valid respondents, in which just over half of 544 ones chose
BL to ride (BLUP = 58.4%), the others chose sidewalk instead. The respondents were
divided into groups depending on the considering variable, then BLUPs of each group
were calculated. Fig. 1 illustrates overview of the distribution of BLUPs by variable.

Fig. 1 BLUP distribution by variable

To clarify the differences in BLUPs between groups are statistically different or

just randomized, proportion tests were carried out. Table 2 shows the results.
In the stated preference survey, questionnaires were designed to ask the cyclists
about their preference level to BL and the reasons why they choose or deny using BL.
Table 3 shows the questions and the corresponding surveyed results.

4 Discussion and Conclusion

One of the most remarkable results of this study was that identified the most im-
SRUWDQWDVSHFWWKDWGHFLGHGZKHWKHUF\FOLVWVFKRVH%/RUQRW7KLVZDV³3HUFHLYHG
Reasons for cyclists denying new-built bicycle lanes 937

Table 2 Results of proportion tests

Variables Note Proportion test results

Gender Man/Woman Ȥ2 = 42.1, df = 1, P < .01
SW.Width Meter Ȥ2 = 30.2, df = 3, P < .01
TL.Width Meter Ȥ2 = 8.8, df = 2, P = 0.012
BL.Width Meter Ȥ2 = 27.1, df = 1, P < .01
Veh.Vol. Veh/lane.hour Ȥ2 = 20.4, df = 4, P < .01
Wait.Veh. Yes/No Ȥ2 = 0.02, df = 1, P = 0.89
Stop.Veh. Yes/No Ȥ2 = 4.6, df = 1, P = 0.03

Table 3 Questionnaire survey results

Question: How often do you choose BL to ride? (by gender)

Almost always Frequently Sometime Rarely Almost never
Woman 13.6% 16.5% 31.3% 25.0% 13.6%
Man 13.6% 21.5% 32.2% 21.5% 11.2%
Question: What is the main reason for you to choose BL or sidewalk to ride?
Perceived safety Comfortableness Just habit Others
73.2% 11.9% 5.7% 9.3%
Question: While cycling on-street how following factors affect your perceived safety
Very much Fairly Some Little No effect
Width of kerb/BL 57.7% 24.5% 9.4% 3.1% 5.3%
Close to vehicle traffic 59.4% 23.8% 7.9% 3.2% 5.7%
Car/bus parking on kerb/BL 63.0% 21.3% 6.1% 4.8% 4.8%
Question: How does sidewalk width affect your perceived safety while you cycling on there?
Very much Fairly Some Little No effect
33.3% 46.8% 12.1% 4.5% 3.3%

VDIHW\´ ZKLOH DOPRVW WKUHH TXDUWHUV  VXSSRVHG WKDW WKLV ZDV WKH PDLQ factor
affecting their choice between BL and sidewalk (Table 3). This trends was consistent
with the findings of previous studies that the cyclists often choose where they feel
safer to ride [14, 15]. Therefore, one can say that if a BL is denied by cyclists, this is
more likely because cyclists feel unsafe while riding on it. Based on this, following
sections discusses about impact trends of considering factors and also try to explain
why the trends become like that.
As for cycliVW¶VJHQGHUmen tended to choose BL more frequently. While men cy-
FOLVWV¶ BLUP was 68.3%, the figure of women was by far lower, at 47.1% (Fig. 1).
This trend was consistent with the questionnaire survey result, while there was 21.3%
of men said they frequently used BL, compared to 16.5% of women (Table 3).
The difference between men and women could be because of the difference in per-
ceived safety between men and women. Several previous studies have illustrated that
women were often more concern about safety under effects of traffic condition factors
938 N. Duc-Nghiem et al.

whiling traveling on-street (e.g. high traffic volume, high speed vehicles) compared to
men [10, 16]. This could be the reason why women less frequently chose BL (an on-
street facility) compared to men.
Regarding vehicle volume, the literature indicated that high vehicle volume can
impair F\FOLVWV¶ perceived safety as they riding on-street [6, 9]. This could be the rea-
son why the proportion of cyclists chose BL showed a downward trend when vehicle
volume increased (Fig. 1). This propounds that constructing BLs on heavy vehicle
traffic streets could not be effective. In this situation, protective devices like fence or
kerb for separating BL and nearby vehicle traffic could work to improve BL usage.
As for the other two traffic condition factors, while ³Stop.Veh.´ shows significant-
ly negative impact on BLUP, ³Wait.VHK´is not a significant impact factor (Table 2).
The results of questionnaire survey illustrated that there was a very high proportion
(63.0%) of respondents stated that stopping vehicle on BL impaired their perceived
safety ³9HU\ PXFK´ (Table 3), in comparison with 4.8% of ones VDLG ³1R HIIHFW´
This suggests planners, designers, etc. that to promote BL usage, stopping vehicle on
BL should be limited or prohibited.
7KHLPSDFWRI³6WRS9HK´FRXOGEHH[SODLQHGDVIROORZV:KHQ a cyclist is riding
on a BL, a stopping vehicle on the BL is an obstacle to the cyclist. To overpass the
vehicle, he/she has to reroute and run into nearby traffic lane. In this case, he/she
might feel unsafe. Therefore, riding on sidewalk in this case might become a better
alternative for the cyclist.
In term of the effects of infrastructure characteristic, many previous studies [e.g. 5,
6, 14] illustrated that width of facility made significant impacts on cyclists¶ facility
preference. In this study, width of BL, traffic lane and sidewalk was taken into ac-
count. The impact trends of those are as follows.
Firstly, bigger BLs showed higher figures of BLUP compared to that of the smaller
ones, while 1.0-meter-width BLs accounted for by far higher BLUP, at 67.6%, com-
pared to that of 0.8-meter-width BLs, at 50.5% (Fig. 1). This suggests that if possible,
we should increase BL width to attract cyclists more. If we plan to construct narrow
BLs, the probability that the BLs could be ineffective (low BLUP) should be consid-
ered.
Similarly, traffic lane width also showed positive correlation to BLUP (Fig. 1).
+RZHYHUWKHZD\WUDIILFODQHZLGWKLPSDFWHGF\FOLVWV¶IDFLOLW\FKRLFHFRXOGEHGLIIHr-
ent from BL width did. In stated preference survey, there were almost 60% of re-
spondents supposed that cycling close to traffic impaired their perceived safety ³9ery
much´. This suggests that when traffic lane was bigger, the gaps between the riders to
the nearby motor traffic also bigger. So, cyclists could feel safer while cycling nearby
bigger traffic lanes compared to narrower ones.
On the other hand, sidewalk width showed the reverse relation with BLUP (Fig. 1).
There were about three quarters of respondents stated that while cycling on sidewalk,
sidewalk ZLGWKDIIHFWHGWKHLUSHUFHLYHGVDIHW\DW³9HU\PXFK´RU³)DLUO\´ level (Ta-
ble 3). So, one can say that sidewalk width has an indirect impact on BLUP, because
bigger sidewalks can attract cyclists more, therefore decreases BL usage.
To sum up, the most important reason why cyclists deny BL is that they feel unsafe
DVULGLQJRQLW7KLVSURSRXQGVWKDWWRSURPRWH%/XVDJHZHQHHGWRLPSURYHF\FOLVWV¶
Reasons for cyclists denying new-built bicycle lanes 939

perceived safety first, especially for women because they are often more concern
about risks of accident on the street compared to men.
This study also indicates some specific deterrents that preventing cyclists from us-
ing BLs such as heavy vehicle volume, the existence of real-time stopping vehicle on
the BL, narrower traffic lane, narrower BL, and wider nearby sidewalk. This suggests
measures to adopt more reasonable plans to promote BL usage, for example separat-
ing cyclists from vehicle traffic, limiting or prohibiting stopping vehicle on BL, etc.
This could somehow contribute to avoid abandoned BLs in the future cities.

Acknowledgments

This study was supported by Japan Society of Traffic Engineers, Research Grant of
The General Insurance Association of Japan

References

1. Cambridge Dictionary: https://dictionary.cambridge.org/vi/dictionary/english/bike-lane.

2. City of San Francisco: Fell Street Bike Lane (Scott to Baker) and Tow-Away Zone
Proposal. San Francisco, CA. (2004).
3. Taylor, D. and Mahmassani, H.: Transportation Research Record: Journal of the
Transportation Research Board (1556), 86-95 (1996).
4. Sener, I. N., Eluru, N. and Bhat, C. R.: Transportation 36 (5), 511-539 (2009).
5. Aultman-Hall, L. and Adams Jr, M.: Transportation Research Record: Journal of the
Transportation Research Board (1636), 71-76 (1998).
6. Li, Z., Wang, W., Liu, P. and Ragland, D. R.: Transportation Research Part D: Transport
and Environment 17 (3), 256-261 (2012).
7. Stinson, M. and Bhat, C.: Transportation Research Record: Journal of the Transportation
Research Board (1828), 107-115 (2003).
8. Godwin, A. and Price, A. M.: (2016).
9. Parkin, J., Wardman, M. and Page, M.: Accident Analysis & Prevention 39 (2), 364-371
(2007).
10.Garrard, J., Rose, G. and Lo, S. K.: Preventive medicine 46 (1), 55-59 (2008).
11.Emond, C., Tang, W. and Handy, S.: Transportation Research Record: Journal of the
Transportation Research Board (2125), 16-25 (2009).
12.Ford, K. M. and Brown, J. M.: presented at the Transportation Research Board 90th Annual
Meeting, Paper No. 11-0956, 2011 (unpublished).
13.Team, R. C.: R: A language and environment for statistical computing. R Foundation for
Statistical Computing, Vienna, Austria. (2016).
14.Bohle, W.: presented at the Traffic Safety on Two Continents Conference, 2000
(unpublished).
15.Tilahun, N. Y., Levinson, D. M. and Krizek, K. J.: Transportation Research Part A: Policy
and Practice 41 (4), 287-301 (2007).
16.Harris, C. R., Jenkins, M. and Glaser, D.: Judgment and Decision Making 1 (1), 48-63
(2006).
Sustainable and Health-Oriented Transport Planning and
Urban Planning

Karin Menges1 and Manfred Boltze1

1
Technische Universität Darmstadt, Otto-Berndt-Str. 2, 64287 Darmstadt, Germany

Abstract. The proportion of people living in cities is increasing in developed as

well as developing countries. Besides the positive effects of urbanisation there
are also negative impacts on human health, for example due to heat islands, air
pollution and noise. Health is of high importance for the citizens, and in our so-
cieties the awareness for health impacts is increasing due to more welfare.
However urban planning as well as transport planning are not considering
health impacts sufficiently, so far. Consequently, a sustainable, health-oriented
and integrated urban and transport planning approach is urgently needed to pro-
vide a high quality of life in a healthy urban environment. Upcoming changes in
vehicle technology and transport services will induce and require new urban
forms. The growing individual interest in improving the personal health situa-
tion will furthermore contribute to the high demand for an integrated approach.
In order to derive adequate processes and measures in the field of urban and
transport planning to improve human health and environmental conditions, the
interdependencies between urban form, transport system and health are elabo-
rated.

Keywords: Transport Planning, Urban Planning, Health, Interdependencies.

1 Introduction

A large part of the population lives in cities, and forecasts estimate that the number
will continue to increase. More than 70% of the population in Germany lives in cities
today, and it is expected that this proportion will rise to just fewer than 80% till 2030.
[1] In Vietnam, the urbanisation process is also steadily proceeding. In 2018, about
35% of the population lived in cities, and it is assumed that the proportion of people
living in cities will increase to about 55% till 2030. [2] Cities, on the one hand, are
engines of the economy, through which work and social life gets influenced in a posi-
tive way; on the other hand, they can negatively affect people's health, for example
through heat islands, air pollution and noise, resulting in increased morbidity and
premature deaths. [3]
The issue of health is of great importance to citizens, and awareness of the health
impact is also increasing due to the growing prosperity of our societies. Therefore,
this aspect will be more and more important for urban and transport planners. The fact
that health has a great value for the German population is shown by surveys of the
inhabitants. In 2014 and 2016, health was the most important issue for respondents,

© Springer Nature Singapore Pte Ltd. 2020 941

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_150
942 K. Menges and M. Boltze

and in 2018, health was still ranked on second place after nature. [4] Above all, the
problem of air pollution and the associated premature deaths turn out to be a global
problem. For example, the European Air Quality Report counts 422.000 premature
deaths due to air pollutants. [5] Air pollution and the associated health impacts are
also a current deeply discussed topic in Germany. The Federal Environment Agency
estimates that air pollution by particulate matter can be attributed annually 600,000
lost years of life, plus health impairments by NO2 and ozone. [6]
So far, there was a focus on improvements in vehicle technology to reduce emis-
sions. However, urban as well as transport planning are not considering health im-
pacts sufficiently, so far. Consequently, a sustainable, health-oriented and integrated
transport and urban planning approach is urgently needed to achieve a high quality of
life in a healthy environment. The growing individual interest in improving the per-
sonal health situation will furthermore contribute to the high demand for an integrated
approach.

2 Interdependencies Between Urban Form, Transport, and

Health

2.1 Overview

Fig. 1. Interdependencies between Urban Form, Transport, and Health.

The three fields of urban form, transport and health are connected by a variety of
interactions (see Figure 1). The urban form is defined by the four elements size, den-
sity, land use and layout. The field of transport, as a result of transport planning and
traffic engineering, can be separated in transport supply, transport demand, and traffic
flow. The field of health consists of exposure and health-related effects of urban and
transport planning. Different elements of urban planning can influence transport sup-
Sustainable and Health-Oriented Transport Planning … 943

ply and transport demand and the other way around. There are also direct influences
of urban planning on exposure. Existing transport supply and transport demand influ-
ence traffic flow which results in different levels of exposure. In the urban environ-
ment, people are exposed to various positive and negative effects of traffic, which are
summarized under the heading of exposure in figure 1. This exposure results in dif-
ferent health effects.

2.2 Interdependencies between Urban Form and Transport

Various factors of urban form affect transport supply as well as transport demand,
and there are many detailed investigations of these different interrelations in litera-
ture. [7] The density of a city, for example, has an effect on the distance travelled, and
thus, on mode choice. Diversity is a measure of the land use mix (different types of
use like residential buildings, shopping ...). If there are more shops near the place of
residence, then people have to travel shorter distances, and they are less dependent on
car and able to use different modes of transport. [8] A compact city is characterized
by a high density, high land use mix and a good accessibility of the public transport
system. This results in a change in modal split with a reduction of the share of trips by
motorised individual transport and a shift to public transport, cycling and walking
which has a positive effect on the health of the population. [9] Urban layout can be
separated in urban structure, access structure and housing and building/type. There are
different types of network: linear, radial, ring, grid and shifted grid network which
can influence the modal split and specifically the amount of motorised kilometers
traveled. [10] Housing and building/type can have a direct influence on exposure due
to ventilation dependent on different building development. For example, an urban
canyon with high-rise buildings presents worse ventilation than detached houses. [11]

2.3 Interdependencies between Transport and Health

Different elements of the transport system are influencing the scale of exposure,
and hence, the health impacts on inhabitants and road users. In many cities the car is
still the main transport mode, and public space is dominated by cars. With different
elements of transport supply ± such as a well-developed public transport system with
dense route network and schedule, or a well-established network of cycling routes ±
the traffic demand can be influenced and the levels of environmental stressors such as
air pollution can be reduced.
Peak-hours of commuting usually are the high-exposure periods during the day.
And especially in high vehicle-density metropolitan areas people are exposed to a
high pollution level on their way to work. There are several studies investigating the
influence of mode choice on exposure. [12] Boogaard et al. consider the exposure of
motorists and cyclists to ultrafine particles, fine particles and noise on different routes
in eleven Dutch cities. [13] Rank et al. did a similar study in the City of Copenhagen,
investigating the exposure of cyclists and car drivers on different routes during morn-
ing traffic. [14] The result of both studies is that the exposure of motorists was higher
than of people driving by bicycle. Beside the smaller proportion of exposure to air
pollution, cycling has also a positive health effect due to exercise. However, although
944 K. Menges and M. Boltze

active transport like cycling or walking increases the intake of pollution, the benefit is
still bigger than the negative impact. [15]
Beside mode choice, also route and time choice are influencing factors. Due to dif-
ferent traffic loads and other factors, emission levels are different on various routes
and over the day. For example, the share of ultrafine particle exposure could be re-
duced by 50% if main roads are avoided during commuting by bicycle. Research has
shown that ultrafine particulate exposure for all modes of transport is highest during
the morning peak hour. [16] In order to reduce the personal daily burden of air pollu-
tants, it is useful to adjust the time-of-day for travel not only with respect to improved
traffic flow, but also to health aspects.

2.4 Interdependencies between Urban Form and Health

With the continuous growth of population and especially of urban areas around the
world, urban planning is now recognised to be able to provide solutions for different
adverse health effects. Besides influencing health via transport planning and traffic
engineering, there are also possibilities for urban planning to influence health directly.
According to the World Health Organisation, physical inactivity is a major public
health risk. [17] An increase in green space and a better proximity to green park areas
can influence physical activity and air quality which results in better health conditions
for city residents. Coppel and Wüstmann investigated the impact of urban green space
on health in Berlin. They figured out that the coverage of urban green space and the
distance from the place of residence to green areas influence the subjective well-being
of inhabitants. [18] Besides green spaces, the orientation of streets in relation to wind
direction may have an effect on air quality. The highest proportion of air pollution is
expected when the wind comes vertical to the road axis and buildings because the
ventilation of the road gets reduced. If the wind direction is parallel to the road direc-
tion, it benefits from a better ventilation and shows about 50% lower levels of air
pollutants. When compiling a development plan and the associated definition of the
orientation of the road, it is therefore recommended from an urban planning point of
view, to align the streets parallel to the main wind direction to provide better ventila-
tion. [11]

2.5 Influence of Health Effects on Urban Form and Transport

Increasing awareness about the health impacts of urban form and transport will cer-
tainly affect the way in which we are planning cities and transport systems. Currently,
the health effects of transport are not taken into account in urban planning and
transport planning in a sufficient way. However, as the results of the studies men-
tioned above show, there are various ways to reduce the negative impacts of urban
form and transport on the health of citizens and road users. The objective should be to
integrate the health effects as important criteria in urban planning and transport plan-
ning. Furthermore, health aspects should also play a role at the level of transport us-
ers' individual choices in terms of transport demand. For example, choices of travel
route and travel time choice should be made under consideration of air pollutant ex-
posure.
Sustainable and Health-Oriented Transport Planning … 945

3 The SHOTUP Project

:LWKLQWKHSURMHFW³6XVWDLQDEOHDQG+HDOWh-Oriented Transport and Urban Planning

± 6+2783´, which is funded by the German Federal Ministry of Education and Re-
search and the Vietnamese Ministry of Science and Technology, different identified
interdependencies between Urban form, transport and health will be elaborated in
detail. The systematic description, as outlined in chapter 2, will be further developed
and used for a comparative study of Germany, as a developed country, and Vietnam,
as a developing country. The aim is to identify the differences between these coun-
tries and show the need for action and possibilities for improvements. For a more
detailed case study, different quarters in Frankfurt am Main (Germany) and Ho Chi
Minh City (Vietnam) will be selected. With the help of comparative measurements of
different air pollutants in the selected quarters, a database will be provided to investi-
gate the conditions in different countries and situations. Several expert workshops in
both participating countries will be utilized to extend the existing framework and to
develop recommendations for a future health-oriented urban planning and transport
planning.

4 Conclusion

Due to the progressing urbanisation and the large number of people affected by
transport, this topic will continue to be of great interest in the future. Numerous indi-
vidual studies have already investigated the effects of different elements of urban and
transport planning on health. The aim is to combine these findings within a generic
framework and to provide a better basis for an integrated urban and transport planning
approach. For urban as well as transport planning it has to be taken into account that
there are major changes ongoing regarding technologies and travel behaviour (e.g.
automated driving, low emission vehicles, active transport). In order to enable sus-
tainable development, such changes in the development of the urban structure and
transport infrastructure must be carefully considered. In addition to the recommenda-
tions for the planning levels, however, the awareness of road users should also be
sharpened in order to make individual decisions, such as route or time choice, with
due regard to health effects. With the help of the targeted comparison of the countries
Germany and Vietnam the differences of a developed and developing country are to
be worked out and the need for action as well as effective approaches for improve-
ments are pointed out.

References
1. Das Statistik-Portal. Anteil der in Städten lebenden Bevölkerung in Deutschland und
weltweit von 1950 bis 2010 und Prognose bis 2030,
https://de.statista.com/statistik/daten/studie/152879/umfrage/in-staedten-lebende-
bevoelkerung-in-deutschland-und-weltweit/, last accessed 2019/02/03
946 K. Menges and M. Boltze

2. World Urbanisation Prospects - Population Division - United Nations. Online available

under https://population.un.org/wup/Country-Profiles/, last accessed am 07.02.2019.
3. Bettencourt, L. M.Lobo, J., Helbing, D., Kühnert, C., & West, G. B.: Growth, innovation,
scaling, and the pace of life in cities. PNAS, 104, 7301-7306 (2007).
4. Werte-Index 2018: Natur und Familie sind den Deutschen jetzt wichtiger,
https://www.kantartns.de/presse/presseinformation.asp?prID=3609, last accessed
2019/02/03
5. European Environment Agency: Air Quality in Europe ± 2018 Report. (2018).
6. Schneider, A., Cyrys, J., Breitner, S., Kraus, U., Peters, A., Diegmann, V., Neunhäuserer,
L.: Quantifizierung von umweltbedingten Krankheitslasten aufgrund der Stickstoffdioxid-
Exposition in Deutschland. Umweltbundesamt 1/2018 (2018).
7. Ewing, R., Cervero, R.: Travel and the Built Environment ± A Synthesis: Transportation
Research Record 1780, Paper No. 01-3515, p. 87-114 (2010).
8. Nieuwenhuijsen, M., Khreis, H.: Urban and Transport Planning, Environment and Health.
Integrating Human Health into Urban and Transport Planning - A Framework. Springer, 3-
16 (2019).
9. Stevenson, Mark; Thompson, Jason; Sá, Thiago Hérick de; Ewing, Reid; Mohan, Dinesh;
McClure, Rod et al. (2016): Land use, transport, and population health: estimating the
health benefits of compact cities. In: The Lancet 388 (10062), S. 2925±2935. DOI:
10.1016/S0140-6736(16)30067-8.
10. Snellen, Danielle; Borgers, Aloys; Timmermans, Harry (2002): Urban Form, Road Net-
work Type, and Mode Choice for Frequently Conducted Activities: A Multilevel Analysis
Using Quasi-Experimental Design Data. In: Environ Plan A 34 (7), S. 1207±1220. DOI:
10.1068/a349.
11. Kastner-Klein, Petra (1999): Experimentelle Untersuchung der strömungsmechanischen
Transportvorgänge in Straßenschluchten.
12. Karanasiou, Angeliki; Viana, Mar; Querol, Xavier; Moreno, Teresa; Leeuw, Frank de
(2014): Assessment of personal exposure to particulate air pollution during commuting in
European cities--recommendations and policy implications. In: The Science of the total
environment 490, S. 785±797. DOI: 10.1016/j.scitotenv.2014.05.036.
13. Boogaard, Hanna; Borgman, Frank; Kamminga, Jaap; Hoek, Gerard (2009): Exposure to
ultrafine and fine particles and noise during cycling and driving in 11 Dutch cities. In: At-
mospheric Environment 43 (27), S. 4234±4242. DOI: 10.1016/j.atmosenv.2009.05.035.
14. Rank, Jette; Folke, Jens; Homann Jespersen, Per (2001): Differences in cyclists and car
drivers exposure to air pollution from traffic in the city of Copenhagen. In: Science of The
Total Environment 279 (1-3), S. 131±136. DOI: 10.1016/S0048-9697(01)00758-6.
15. Rojas-Rueda, David; Nazelle, Audrey de; Tainio, Marko; Nieuwenhuijsen, Mark J. (2011):
The health risks and benefits of cycling in urban environments compared with car use:
health impact assessment study. In: BMJ (Clinical research ed.) 343, d4521. DOI:
10.1136/bmj.d4521.
16. Ragettli, Martina S.; Corradi, Elisabetta; Braun-Fahrländer, Charlotte; Schindler, Chris-
tian; Nazelle, Audrey de; Jerrett, Michael et al. (2013): Commuter exposure to ultrafine
particles in different urban locations, transportation modes and routes. In: Atmospheric
Environment 77, S. 376±384. DOI: 10.1016/j.atmosenv.2013.05.003.
17. World Health Organization (2009): Global health risks. Mortality and burden of disease at-
tributable to selected major risks. Geneva, Switzerland: World Health Organization.
18. Coppel, Gero; Wüstemann, Henry (2017): The impact of urban green space on health in
Berlin, Germany: Empirical findings and implications for urban planning. In: Landscape
and Urban Planning 167, S. 410±418. DOI: 10.1016/j.landurbplan.2017.06.015.
 Integrated Transport Planning for Sustainable Urban
Development ± 6LQJDSRUH¶Approach and Lessons for
Vietnam

Trinh Dinh Toan 1[0000-0002-9113-1975] and Dao Van Dong 2[1111-2222-3333-4444]

1
Department for Transportation Engineering,
ThuyLoi University, 175 TaySon, DongDa, Hanoi
Trinhdinhtoan@tlu.edu.vn
2
University of Transport Technology
54 Trieu Khuc, Thanh Xuan, Hanoi.

Abstract.
7KLV SDSHU H[DPLQHV 6LQJDSRUH¶ DSSURDFK WR LQWHJUDWHG WUDQVSRUW SODQQLQJ IRU
sustainable urban development, and draws lessons that may be applicable for
large cities in Vietnam. The paper explores strategic planning approaches
Singapore applies for transport and land use integration, and the planning
instruments Singapore deploys to integrate different transport modes physically
and institutionally. It learns the way Singapore plans a city to make the city
liable and sustainable, and proposes relevant points under the subject of
integrated transport planning for Hanoi and HoChiMinh cities in Vietnam.

Keywords: Integrated Transport Planning, Land Use Plan, Sustainable Urban

Development.

1. Introduction

Transport planning is the process of defining future policies, goals, investments, and
designs to prepare for future needs to move people and goods to destinations.
Integrated transport planning refers to a well-functioning transport system that
pursues an integrated approach across various themes such as: (i) planning objectives;
(ii) transport and land use; (iii) transport modes; and (iv) operation [1].
Sustainable development commonly refers to a development that meets the needs
of the present generation without compromising the ability of future generations to
meet their own needs [2]. Sustainable transport refers to the broad subject of
transport that is sustainable in the senses of social, environmental and climate
impacts. Sustainable transport aims at a proper balance between current and future
environmental, social and economic qualities. Sustainable transport is a key element
in development of sustainable cities. A city with sustainable transport should be a city
that promotes integrated planning between transport and land use to reduce the need
for travel, to maximise the utility and value of spaces, and to enhance efficiency of
energy use and the city liveability [3].

© Springer Nature Singapore Pte Ltd. 2020 947

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_151
948 T. D. Toan and D. V. Dong

Singapore Transport System

Singapore is a small, densely populated country at the southern tip of the Asian
Continent. It has a land area of 720 sqkm, and the population of 5.6 million (2017).
The road network consists of 164 km of expressways, 662 km of arterial roads, 571
km of collector roads, and 2,055 km of local access roads, with the road density of 5
km/sqkm. Singapore has the total motorized vehicle population of 961,842, including
546,706 private cars; 23,140 taxis; 19,285 buses; 162,172 goods and other vehicles;
and 141,916 motorcycles. About 60% of Singaporean working population use public
transport [4].
The role of integrated transport planning for sustainable urban development is
clear, and is particularly essential for a densely populated country like Singapore.
Singapore Government considers development of an integrated transport system a
fundamental principle, among others, to achieve sustainable and liable city. The
integrated transport has been achieved by various approaches, notably close
integration between transport and land use planning, integration between transport
modes, integration with urban developments, and integration among institutions. It is
integrated transport planning that keeps Singapore moving.

2. 6LQJDSRUH¶ Approach in Integrated Transport Planning

2.1. Land Use and Transport Planning Integration

Long-term land use and transport plans, known as Concept Plan, represent strategic
plans that guide the overall physical development of the country for a 40-50 year
planning horizon. The plans are prepared in collaboration between various land use
agencies and the Land Transport Authority to accommodate anticipated population
and economic growth. The 2011 Concept Plan, for example, incorporated proposals
from more than 20 government ministries and agencies into a national development
policy document. The Concept Plan are translated into operational details in the
Master Plan, known as Medium-term land use and transport plans that provide early
physical arrangements for developments and consider transport network solutions that
shape the land use for 10-15 year horizon [5]. At the local level, urban design
guidelines consider physical integration between transport nodes and surrounding
developments. Over the years, this integrated framework for land use and transport
development continues to provide seamless connection for transport users.

2.2. Other Types of Integration

Network integration required a fundamental evaluation of the role of railed-based
transport and buses in providing public transport services [6]. The railed-based
transport includes mass rapid transit (MRT) and light rail transit (LRT). The two
modes complement to each other in the manner that MRT is the skeleton component
that provides high capacity, mobility and travel time reliability for heavy transport
corridors, LRT serves as feeders to the MRT network, and the bus system (SBS)
serves less heavy corridors to supplement both of them [4]. There are also convenient
Integrated Transport Planning for Sustainable … 949

linkages between rail stations and bus stops or taxi stands, by covered footpaths or
overhead bridges within short distances (Fig. 1).

Fig. 1. Convenient linkages between different modes of transport (Source [4])

Integration with Urban Developments

In Singapore, transport facilities are fully integrated with building developments
around the MRT stations as integrated public transport HUBs to reduce the need for
travel, and to give maximum accessibility to housing, commercial and recreational
facilities. This integrated framework for land use and transport development continues
to evolve to meet changing needs. Through the years, large complexes of
developments outside the CBD, including residential and social infrastructures are
clustered proximally around the HUBs [7], within a walking distance to encourage
greater commuting via public transport. Building more homes near work places and
more work places in residential areas reduce demand for transport.
Institutional Integration: The integration of land use and transport in Singapore is
greatly fostered by the government policies, planning and management instruments,
including a well-defined governance structure of institutions with clear roles and
responsibilities [6]. There is close interactions between government agencies, in
particular between the Land Transport Authority (LTA), the Urban Redevelopment
Authority (URA), and the Housing and Development Board (HDB) in solving
integrated issues between land use and transport. Within the LTA, there are close
operational interactions between the responsible agencies in various aspects,
including fare integration and information sharing [5].

3. Issues in Vietnam Urban Transport

The urban transport in big cities in Vietnam is predominantly undertaken by land

transport. In Hanoi Capital for example, the road network consists of nearly 600 road
sections with the total length of 1,200km. The road network is characterised by radial
arterials and ring roads, and predominantly at-grade intersections. There are
950 T. D. Toan and D. V. Dong

approximately 5 million motorcycles, 500 thousand cars, and 1.2 million other means
of transport. The average traffic growth is 10.2%/year, while the increase in the land
area for transport is only 0.25%/year. The primary means public of transport are bus
and taxi, whose contribution to the total ridership is as low as 15%, while there is
almost no rail-based transit for the time being [4]. Over the years, problems of traffic
congestion, traffic safety, and environmental pollution are getting pervasive in Hanoi
and HoChiMinh cities. Pervasive traffic congestion leads to low urban mobility ± one
of critical constraints to urban economic growth. These problems of urban transport
are attributed to many factors, among which integrated transport planning under this
study is considered a crucial factor that has great impact on the sustainable transport,
and is elaborated in the following section.
Lack of integrated planning: In Vietnam, there are 4 types of Plans that have
influences on transport system, including: General Socio-Economic Development
Plan (QHTT-KTXH), General Construction Plan (QHCXD), Land Use Plan
(QHSDD), and Transport Master Plan (QHGTVT). By law, QHGTVT must comply
with QHCXD and QHSDD for the same planning period. In theory, the regulation
that QHTT-KTXH guides the other Plans seems to ensure the consistency in the
development objectives, however, in actuality the interdependency between the Plans
causes serious consequences to the reliability and validity of the Plans.
As an example, regarding the planning period, the QHTT-KTXH and QHSDD
usually have the planning horizon of 10 years, but the QHCXD and QHGTVT usually
have the planning horizon of 15-20 years [8]. Therefore, the QHTT-KTXH and
QHSDD planning horizon is not sufficient as the basis for development of QHGTVT.
Another example, the preparation and approval of the Plans are very time consuming,
and must comply the top-down procedure as stated. Consequently, the approved
QHTT-KTXH may be no longer valid or reliable to set a ground for the other Plans.
This results in a paradox that the prescribed statement on the necessity and the
magnitude of the transport infrastructure do not base on technical grounds, such as the
forecasted travel demands from a 4-stage transportation modeling procedure. Thus,
the rigid constraints between the Plans in the interdependency manner cause problems
in reliability and validity of the planning documents.
Lack of physical and service integration: There is poor physical connection
between transport facilities, transport modes, and the integration of transport services.
Hanoi Bus Rapid Transit Project (BRT) is a typical example: The BRT is a
component of the Hanoi Urban Transport Development Project that aims at increasing
use of public transport and promoting environmentally sustainable transport for
Hanoi. After several PDMRU FKDQJHV LQ +DQRL¶V SXEOLF WUDQVSRUW VWUDWHJ\ the final
output was cut down to only a single BRT route of 14.7 km in length. The route began
its operation since January 2017, but has been basically under-utilized so far with
only 20 buses per hour, equivalent to 1,200 passengers per hour in each direction [9],
significantly lower than the capacity if the route would be dedicated to private
vehicles.
Integrated Transport Planning for Sustainable … 951

Fig. 2. The BRT corridor (white) and the Metro Line 2A corridor (yellow). Source [9].

Major causes of the problems is primarily attributed to a number of factors: (i)

improper corridor selection: a high portion of the BRT line runs through farmland
and low density residential area, and located only 1.3km apart [9] from the Metro
Line 2A of CatLinh - HaDong (Fig. 2); (ii) lack of integration with regular buses: the
single BRT line become unattractive for the limited coverage since it has only 5
connections to regular bus routes; (iii) Poor BRT station access and lack of modal
integration that lead to the fact that the BRT cannot accommodate a significant shift
from motorcycle and other transport modes.

4. Recommended Solutions for Vietnam

6LQJDSRUH¶VWUDQVSRUWVLWXDWLRQDQGSODQQLQJDSSURDFKDUHVLJQLILFDQWO\GLIIHUHQWIURP
that of Vietnam. However, being a densely populated country which historically
encountered similar urban transport problems, Singapore successful story can provide
good lessons. A review of Singapore experiences in integrated transport allows the
following recommended solutions for large cities in Vietnam to target a sustainable
transport:
Development of an integrated public-oriented transport system (TOD): The
transport system should be public oriented to ensure large capacity and mobility but
efficient energy use. It should be a comprehensive mass transit system, whose
backbone is the rail-based transit, closely integrated with BRT and regular buses.
Development of a clear comprehensive hierarchical road network: This includes
grade separation for critical intersections to promote transport mobility and
connectivity. The road network should be closely integrated with the rail-based
transport, and subordinated with waterways and airways as alternatives for inter-city
transport.
Multi-modal transport integration: The various modes of public transport systems
should be conveniently physically connected. The system should be extended to non-
952 T. D. Toan and D. V. Dong

motorised transport facilities such as walking and bicycles to provide seamless travel,
and to reduce unnecessary trips.
Integration with urban developments: To reduce the need to travel, the Singapore
integration model between transport and land use and urban developments is
recommended: distributing commercial and economic activities by development of
sub-regional and fringe centers that aim for a proper mix of residential, industrial, and
institutional developments of high densities around the mass transit stations.

5. Conclusions

Being a small densely populated country with scarcity of land and historically
experienced similar urban transport problems as Vietnam, Singapore has dramatic
physical transformation from an automobile-dependence to a livable and sustainable
city in the past 50 years. 7KH6LQJDSRUH¶VVWRU\RQVustainable transport today do not
come about by chance, but is attributed to many factors, of which the integrated
transport planning plays a crucial role. While these strategies have been successfully
used in Singapore, in applying to developing countries like Vietnam, some of them
may work, others may not. Therefore, it is important to thoroughly study the
approaches to work out what is applicable for cities in Vietnam.

References
1. https://en.wikipedia.org/wiki/Transportation_planning, last accessed 2019/2/16.
2. https://en.wikipedia.org/wiki/Sustainable_development, last accessed 2019/5/10.
3. http://www.wikiwand.com/en/Sustainable_transport, last accessed 2019/2/16.
4. Toan, T.D. 0DQDJLQJ WUDIILF FRQJHVWLRQ LQ D FLW\ $ VWXG\ RI 6LQJDSRUH¶V
experiences. Proceeding/ International Conference on Sustainability in Civil
Engineering. Hanoi, Vietnam (2018).
5. Gwee, Evan. Integrated Land Use - Transport Planning & Transport Energy Use
- 6LQJDSRUH¶VH[SHULHQFH$3(5&$QQXDO&RQIHUHQFH7RN\R
6. Chin, H.C. and Foong, K.W. Issues in transportation planning ± the Singapore
experience. WIT Transactions on State of the Art in Science and Engineering,
vol. 22, pp127-158. WIT Press (2005).`
7. Kuang, L.C. 6LQJDSRUH¶V $SSURDFK WR 'HYHORSLQJ D 6XVWDLQDEOH ,QWHJUDWHG
Transport System. Workshop µ7UDQVIRUPLQJ 7UDQVSRUWDWLRQ ,QVWLWXWLRQV
Washington DC, USA (2007).
8. Hung, K.V. https://--tich-hop-cac-quy-hoach-tong-the-phat-trien-do-thi-tren-
nen-tang-quy-hoach-he-thong-giao-thong-van-tai-pdf.htm, last accessed
2019/2/16.
9. World Bank's first BRT in Asia is designed to fail.
https://www.fareast.mobi/en/feature/hnbrtjan17, last accessed 2019/2/16.
 Transitioning Different Stages of Transport Planning in
Urban Areas: Experiences of Singapore and Vietnam
Rojas Lopez, M. C.1*[0000-0002-4718-0649], Trinh, Dinh Toan2[0000-0002-9113-1975] and Wong,
Y. D.1[0000-0001-7419-5777]
1
Nanyang Technological University, Singapore; 2 ThuyLoi University, Hanoi, Vietnam;
* Corresponding author: mariacec001@e.ntu.edu.sg

Abstract. Level of urbanisation ± as population density and transport demand ±

is rapidly increasing in many cities. Planning approaches vary from city to city
and authorities everywhere are being challenged to provide sustainable infra-
structure that meets social needs, maximise space and benefits. Transport plan-
ning, especially in urban areas, is crucial as it influences sustainable city-growth
and space-usage. Transport planning can be classified into three stages, which
evolve with level of urbanisation and other city¶V characteristics: (1st) vehicle-
based, (2nd) person or trip-based, and (3rd) liveable-city.
This study presents an overview of the three transport planning stages and rele-
vant examples. For each stage, the study discusses factors such as capacity, social
needs, different modes of transport, features of the built-environment, emerging
technologies (where applicable), and sustainability impacts. The focus is on plan-
ning approaches from Singapore to enhance the long-term vision for sustainable
urban development of big cities in Vietnam. Singapore, being currently in the 3rd
stage of planning, is focusing efforts in providing inclusive infrastructure and
promoting sustainable modes of transport. Ho-Chi-Minh City and Hanoi are still
focused on motorised transport with low rates of walking and cycling. Lessons
from Singapore to Vietnam are delineated accordingly.

Keywords: Stages of Transport Planning, Vehicle-Based, Person or Trip-

Based, Liveable City, Singapore, Vietnam.

1 Introduction

Transport planning has been evolving in line with social requirements and changes in
the urban landscape. Changes occur globally, yet practices in planning vary. Authorities
select the approach that is deemed to: maximise available space and resources, meet
social needs, and protect the environment. The task is even more intrinsic in urban ar-
eas, where demands of a globalised society triggers changes in inhabitants needs and
requirements [1]. Urban transport planning has been classified into 3 stages: (1st) vehi-
cle-based approach (VBA), (2nd) person or trip-based approach (TBA), and the most
recent, (3rd) activity-based approach (ABA) also known as liveable-city approach [2].
Cities commonly transition through these stages in the quest for quality of lives, sus-
tainability, and healthier lifestyles. This paper presents an overview of these stages and
take examples on Singapore and main cities of Vietnam.

© Springer Nature Singapore Pte Ltd. 2020 953

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_152
954 M. C. R. Lopez et al.

Singapore and Vietnam are countries in Asia. Singapore is an island country with
one of the highest population densities in the world (5.6 million inhabitants and a land
area of 716 km2 [3]). The island-country can be placed in the 3rd stage of urban transport
planning as authorities and planners are working towards developing a µliveable city¶
E\IRVWHULQJDµFDU-OLWH¶VRFLHW\[4]. Vietnam is a country with 33 cities, of which Ho-
Chi-Minh city (HCMC) and Hanoi are the largest and most densely populated [5].
HCMC hosts 8.4 million people in an area of 2,062 km2 and Hanoi hosts 7.5 million
people in an area of 3,329 km2 [6]. In these 2 cities, and the country in general, planning
for transport is mostly focused on motorised vehicles and there is limited public
transport and provisions for active modes of transport (e.g. walking and cycling) [7].
This study aims to explore the three transport planning stages and present relevant ex-
amples of Singapore and Vietnam.

2 The Three Stages of Transport Planning

The 1st stage is a VBA focusing on traffic growth. At this stage, approaches that can
help to maximise capacity for vehicles are prioritised. Vehicle-centric engineering ap-
proaches and infrastructures are often developed and even land-uses are planned to fa-
vour motorised vehicles [8]. This stage has been considered suitable for cities with low
density and wide land space such as some cities in the United States of America and
South America [2]. However, with population, vehicular, and urban growth, the VBA
quickly becomes hard to sustain. Planners soon realised the negative effects of motori-
sation, such as congestions, air pollution, and traffic accidents.
The negativities of motorised modes of transport eventually triggers the 2nd stage of
urban transport planning ± TBA. At this stage, the main aim is to maximise the trips
made by a person (i.e. person-trips approach) [2]. Herein, options to facilitate people
movements in the most efficient way are considered. Countries/cities in this stage (e.g.
Western Europe, Japan, big cities in China ± Beijing, Shanghai) commonly pay atten-
tion to developing public mass transits, such as rail and bus systems. In addition to
infrastructure, changes in policies are also introduced to favour mass transport (e.g.
road pricing) [9]. While being more sustainable than the previous stage, 2 nd stage plan-
ning often fails to understand the factors thDWDIIHFWWULSV¶PRWLYDWLRQDQGILUVWODVWPLOH
(FLM) trips to/from public transport stations are often disregarded. Thus, planning
travel demand prediction usually does not match users¶ constantly shifting needs.
With advances of research and technology, authorities and planners gained
knowledge on travelling motivation and characteristics of FLM trips. This caused a
shift to the 3rd and latest stage of transport planning or ABA. In this stage, the main aim
is to develop a liveable city, where much interest and investment are focused on inves-
WLJDWLQJSHRSOH¶VQHHGVVRFLDOHFRQRPLFFXOWXUDODQGDFWLYLW\SDUWLFLSDWLRQ[10]. This
stage takes a holistic approach to urban transport planning, incorporating spaces for
socialisation, creating an eco-friendly environment, considering health impact of dif-
ferent modes of transport, and aiming to improve overall quality of life. As such, the
role of active trips in enhancing liveability and improving health are acknowledged,
and infrastructure for active travelling is provided accordingly [11].
Transitioning Different Stages of Transport Planning … 955

3 The Case of Singapore

A three-tier planning framework has been developed and implemented since the early
days of 6LQJDSRUH¶Vnation building. In the top-tier, or concept plan, land uses are in-
vestigated with a 50- \HDUV¶ KRUL]RQ7KHQLQWKH PLGGOH-tier, or master plan, the
transport network features are considered with a horizon of 15-10 years in accordance
WRVRFLHW\¶VQHHGs. Finally, in the lowest-tier, or detailed plan, specific features of near-
term developments, policies, and programmes are considered [12]. This planning
framework provides physical arrangements for developments of a transport network
that shapes the land use for long-term planning horizon. Backed up by planning, Sin-
gapore made transport infrastructure the centre of interest for urban development. As
such, it can be said that right from early nation building, Singapore started off at the 2nd
stage or TBA. The 1st concept plan, developed in 1971, presented the ring concept,
where residences, commercial areas, and industries were planned to be developed in a
ring formation around a water catchment, with an interconnected expressway and rail
network, known today as the Mass Rapid Transit (MRT). This has evolved to an MRT
centric planning of self-sufficient towns that can cater for the targeted population of
close to 7 million by 2030 [12].
The towns have been planned so that these are served by regional centres in central,
north, east, south, and west regions of Singapore, connected by MRT rail lines. In the
towns, infrastructure is based on the current and targeted population, with 1/3 for resi-
dential, 1/3 for education, industrial, and commercial, and 1/3 for transport, green
spaces, and recreational activities. A public bus system of feeder buses is also available
to further enhance connectivity. MRT and buses in Singapore follow disciplined oper-
ating practices, always following established stops and rarely experiencing delays. Pub-
lic transport is accessible for the majority of the population, being the most commonly
used modes of transport. It has been reported that 62% of the population ride public
transport during peak periods and expected to increase to 75% by 2040 [13].
Moreover, 3 types of road systems are available in each town for buses and other
motorised vehicle movements. They are the expressways ± connecting towns with each
other, major arterial roads ± connecting with town centres, and minor arterial roads ±
connecting with main residential buildings [14]7RGD\ DURXQG  RI 6LQJDSRUH¶V
available land area is dedicated to transport infrastructure [15]. To restrain vehicle
growth, a certificate of entitlement (COE) is needed to acquire a vehicle, which can
have a cost of over half of the total cost of the car [16]. In addition, an electronic road
pricing system (ERP) is in operation to manage congestion [17].
In recent years, much effort has been placed in promoting active transport modes in
Singapore. This can be seen as the transition from 2nd to 3rd stage, or ABA. Inhabitants
are being encouraged to use these active modes as part of health campaigns, environ-
mental movements, and for social interaction. Also, personal mobility devices (PMDs)
such as electric scooters, hoverboards, and unicycles are gaining popularity. Singapore
presents a very unique layout whereby cyclists and PMD users share off-road facilities
since late 2016. Footpaths are ubiquitously available alongside most roads in Singapore
[18]. Today, walking trips account for 11% of the mode share, with most of these trips
being made for FLM purposes. On the other hand, after a long period of low popularity,
956 M. C. R. Lopez et al.

cycling resurfaced in the last decade and was boosted with the introduction of dock-
less bicycle-sharing system in the country. From traffic counts conducted in recent stud-
ies, it is estimated that prevailing cycling rate is close to 5%, while that for PMDs is
estimated to be around 2%. To cope with the increasing demand for cycling and PMD
riding, authorities are investing efforts to provide off-road cycling paths and other ded-
icated facilities [17]. These developments are likely to further elevate the active mobil-
ity rates in the country that is fostering active lifestyles.

4 The Case of Vietnam

Being a developing country, initial transport planning in Vietnam around 1960-70s was
focused on a VBA. The approach attempts to arrange the road network configuration
to maximise traffic stream. Between 6% and 15% of the available land in the country
is dedicated to transport [7]. The road transport planning is made in compliance with
the socio-economic development strategies, and is integrated with planning for other
transport modes. It is made for at least 10 years, oriented for the next 10 years, and is
adjusted accordingly to the socio-economic situation in each period.
HCMC and Hanoi are the biggest and most densely populated cities in the country.
It is not surprising that they have a very high number of vehicles (mostly motorcycles),
at 7.4 and 5.8 million in HCMC and Hanoi, respectively [7]. This number of motorised
vehicles is expected to further increase regardless of finance policies to control the in-
crease that have been implemented, such as taxes and registration fee [7]. This rapid
growth in inhabitants and vehicles is the major cause of overloading of transport infra-
structures [18]. It is not surprising that pervasive traffic congestion occurs daily, and
negative environmental consequences by vehicle emissions are overwhelming [19]. Ef-
forts have been made to extend the road network [20]. While extension of roads can
help to alleviate traffic, it has also been suggested that providing infrastructure for pri-
vate motorised vehicles only begets more traffic [18]. Rather than catering for unre-
strained vehicle usage in the cities, the primary objective of transport planning should
focus on catering for growing travel demand in another manner, such as TBA.
Around 1980-90¶s, Vietnam aimed to shift towards the 2 nd stage of transport plan-
ning (TBA) with the development of a public transport system. This planning approach
was primarily introduced by the national transport institutions and international organ-
isations who conducted formal comprehensive transport studies for major urban
transport development projects [21]. Public bus services are available in HCMC and
Hanoi, yet, these can meet only 7% and 14% of the demand, respectively [7]. The Hanoi
Master Plan indicated that the bus system would be improved, and the target is to
achieve 35% mode share by 2020 [20].
Regarding active mobility, the rates of walking and cycling in Vietnam are still low.
There are some footpaths available in HCMC as well as in Hanoi, yet these are mostly
used for business purposes. The unavailability of infrastructure, the big size of the cities
as well as travel behaviour have deterred people from active travelling. Provision of
infrastructure and enforcement of traffic regulations could help to further elevate the
walking habit and cycling rate. Further, creation of pedestrian-only streets promote the
Transitioning Different Stages of Transport Planning … 957

liability of the cities substantially. Indeed, the idea of a liveable city has been partly
actualised in some streets in the CBDs. HCMC and Hanoi are the principal economic,
political, and cultural centres of the country. Thus, meeting people's activity participa-
tion should be among primary concerns in transport planning. There are increasing re-
quirements on promotion of public transport and provision of facilities for active modes
and enhancing public space. This could help the main cities in Vietnam to advance
towards the 3rd stage or ABA.

5 Lessons from Singapore to Vietnam

Singapore¶V transport situations and planning approaches are very much different from
that of HCMC and Hanoi of Vietnam. Being a small densely populated country with
scarcity of land and historically encountered similar urban transport problems as Vi-
etnam, Singapore success story becomes prominent example for developing countries
who look for ways to transform from vehicle-dependent to livable and sustainable city.
Cities in Vietnam could take some positive SRLQWVIURP6LQJDSRUH¶VSODQQLQJapproach.
The main point can be the three-tier planning architecture. This procedure allows for
close collaboration between land use and transport planning in a manner that transport
shapes the structure of the urban space while accommodating anticipated population
and economic growth. This ensures sufficient land and other resources for transport-
related developments. While Vietnam¶Vproportion of land area dedicated to transport
is compatible, the rate of public transport usage is minimal. In Singapore, a substantial
proportion of land is dedicated to public transport and active mobility. Singapore de-
veloped an integrated public-oriented transport system (TOD) to ensure large capacity
and mobility. The TOD represents a comprehensive system or rail and buses fully inte-
grated with urban developments. This helps to reduce the need for travel, and to max-
imise accessibility to housing, commercial and recreational facilities. In this way, the
integrated transport system provides seamless and comfortable travel for people, as a
vital function for a liveable and sustainable urban growth [22].
Vietnam could aim to review existing plans and prioritise projects/developments that
favour usage of public transport. Noticing that public transport will entail some active
travelling for the FLM purposes, regulations that can be of aid to active mobility users
should be considered. Also, considering the increasing number of private motorised
vehicles, transport planning should be supported by traffic management and control
such as travel demand management (e.g. COE), and economic instruments (e.g. ERP),
that have been applied effectively in Singapore.

6 Conclusion

Features of the 3 stages of transport planning are presented with examples of Singa-
pore and 2 cities in Vietnam. Singapore is in the 3rd and latest stage of urban transport
planning, where much interest is placed in understanding factors that encourage de-
mand for different modes and providing adequate infrastructure to foster a liveable city,
while cities in Vietnam are basically in the 1st stage of transport planning pertinent to
958 M. C. R. Lopez et al.

vehicle-centric cities. Some experiences from Singapore applicable to Vietnam are

identified and discussed accordingly, indicating that in general, urban transport plan-
ning should always consider existing features of the area and should be constantly re-
viewed to incorporate innovative approaches to cater for the future.

References
1. Costa, P.B., Morais Neto, G.C., Bertolde, A.I., 2017. Urban Mobility Indexes: A Brief Re-
view of the Literature, in: World Conference on Transport Research 10-15 July 2016. Else-
vier B.V., Shanghai, China, pp. 3645±3655. doi:10.1016/j.trpro.2017.05.330.
2. Jones, P., 2014. The evolution of urban mobility: The interplay of academic and policy per-
spectives. IATSS Research 38, 7±13. doi:10.1016/j.iatssr.2014.06.001.
3. SINGSTATS, 2015. Population Trends 2015, Department of Statistics, Ministry of Trade &
Industry, Singapore.
4. CLC, 2018. Streets for all: Designing multimodal streets for a car-lite Singapore. Centre for
Liveable Cities, Singapore.
5. WPR, 2019. Population of cities in Vietnam. World Population Review.
6. GSOV, 2012. Area, population and population density by province, General Statistics Office
Of Vietnam - Statistical Yearbook of Vietnam 2012.
7. Ngoc, T.B., 2015. Challenges and solutions for sustainable urban transport in cities of Vi-
etnam, Department of transport - Vietnam Ministry of Transport.
8. Lam, S.H., Toan, T.D., 2006. Land transport policy and public transport in Singapore. Trans-
portation (Amst). 33, 171±188. doi:10.1007/s11116-005-3049-z.
9. Koh, P.P., Wong, Y.D., 2015. Walking and cycling as an urban transport option in Singa-
pore. Procedings ICE - Munic. Eng. 168, 106±114. doi:10.1680/muen.13.00033.
10. Rojas López, M.C., Wong, Y.D., 2017a. Attitudes towards active mobility in Singapore: A
qualitative study. Case Stud. Transp. Policy 5, 662±670. doi:10.1016/j.cstp.2017.07.002.
11. Meng, M., Zhang, J., Wong, Y.D., 2015. Integrated foresight urban planning in Singapore.
Proc. ICE - Urban Des. Plan. 169, 1±13. doi:10.1680/udap.14.00061
12. LTA, 2019. Land Transport Masterplan 2040. Land Transport Authority, Singapore.
13. LTA, 2016. Road line plan explanatory notes. Land Transport Authority, Singapore.
14. Bin, T.S., 2013. Long-term land use planning in Singapore, Lee Kuan Yew Sch. of Policy.
15. LTA, 2017. COE bidding results. Land Transport Authority, Singapore.
16. LTA, 2017. Electronic Road Pricing. Land Transport Authority, Singapore.
17. Chin, K.K., Menon, G., 2015. Transport accessibility and infrastructure in Singapore± pe-
destrian facilities. Procedings ICE - Munic. Eng. 168, 133±139. doi:10.1680/muen.14.00013
18. Thomas, M., 2002. Urban Studies Out of Control: Emergent Cultural Landscapes. Urban
Stud. 39, 1611±1624. doi:10.1080/0042098022015168.
19. Tung, H.D., Tong, H.Y., Hung, W.T., Anh, N.T.N., 2011. Development of emission factors
and emission inventories for motorcycles and light duty vehicles in the urban region in Vi-
etnam. Sci. Total Environ. 409, 2761±2767. doi:10.1016/j.scitotenv.2011.04.013.
20. VietNamNet, 2016. Hanoi announces transport plan to 2030. News VietNamNet.
21. JICA. Comprehensive Study on the Sustainable Development of Transport System in Vi-
etnam (VITRANSS 2). Final Report, Hanoi, Vietnam (2010).
22. Toan, T.D. and Dong, D.V. Integrated Transport Planning for Sustainable Urban Develop-
ment ± 6LQJDSRUH¶$SSURDFKDQG/HVVRQVIRU9LHWQDP&RQJUqV,QWHUQDWLRQDOGH*pRWHFK
nique - Ouvrages - Structures (CIGO 2019).
 Analysis of Mode Choice Behavior under Adverse
Weather Conditions Using RA and SA Surveys
- A Case Study from Ho Chi Minh City (HCMC), Vietnam

1[0000-0001-8859-5392]
VU Anh Tuan ,
2[0000-0001-7099-6278]
NGUYEN Thi Thanh Huong
1Vietnamese-German Transport Research Centre (VGTRC), Vietnamese-German University (VGU)

drtuan.va@vgtrc.vgu.edu.vn
2Section of Transport Planning and Management, University of Transport and Communications Campus in Ho Chi Minh City, Vietnam

ntthanhhuong@utc2.edu.vn

Abstract. This paper analyzes mode choice behavior in different weather con-
ditions, with population groups, on travel purposes and travel distances to sup-
port the formulation of weather responsive traffic management measures. Re-
vealed and stated adaptation surveys are carried out with 400 respondents and
100 respondents, respectively. The survey shows that the rate of mode change
due to adverse weathers is much smaller than the figure in developed countries.
The major patterns of mode change are shifts from motorcycle to car or taxi,
accounted nearly half of mode changes, thus causing severer traffic conges-
tions. The mode choice modeling show travel time, travel cost, gender, income,
specific jobs, motorcycle ownership, bus ticket holding, provision of weather
information prior to the trip are significant factors to mode choice decision un-
der bad weather conditions. The results suggest focusing on ³DYRLG RU Ueduce
unnecesVDU\PRELOLW\GHPDQG´DQG³LPSURYHpublic transport´measures.

Keywords: mode choice, adverse weather, road flood, revealed and stated ad-
aptation survey, Ho Chi Minh City (HCMC).

1 Introduction

Traffic congestion in Ho Chi Minh City jointly caused by recurrent traffic and road
flooding might cause a loss of 97 billion USD for the period of 2015-2045 (ARUP
and Siemens, 2014). Therefore, city-wide flooding mitigation and traffic management
have become crucially important for alike cities. Traffic management strategies in-
clude advisory, control and treatment measures to help address road congestions
caused adverse weathers [1]. Among these, data collection and analysis of behavioral
mechanisms or human factors are essential for formulating weather-responsive traffic
management strategies. Unfortunately, Vietnam has a lack of practice and research on
adverse weather responsive traffic management. Control measures at network and
road section levels are implemented manually by policemen. As a result, proper and
timely responses to incidents including adverse weathers are often very limited. Pre-

© Springer Nature Singapore Pte Ltd. 2020 959

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_153
960 A. T. Vu and T. T. H. Nguyen

liminary observations show that during a heavy rain the number of cars and taxes may
increase by up to 70-80%, meanwhile, the number of motorcycles might decrease by
30-40% as compared to a normal condition. It is interesting to understand how and
why people change travel mode when an adverse weather event happens. Such under-
standing is necessary to help predict and evaluate the impacts of bad weather events
on transport system, and determine the benefits of implementing advisory, control and
treatment strategies [3]. This paper is aimed to present the results of a deep research
on mode choice behavior and decision-making factors under adverse weathers and
flooding conditions. This paper is structured in five chapters. The first chapter pre-
sents the need of understanding mode choice changes. The second chapter describes a
methodology and data collection. The third chapter presents the results of analysis on
mode change patterns. The fourth chapter presents the results of binary logit mode
choice model workable under adverse weathers. The fifth chapter concludes with
solution recommendations and further works.

2 Methodology and Data Collection

Based on literature review and focus group discussions, two questionnaires were de-
signed. One was for a revealed adaption (RA) survey to capture mode choice behavior
changes in response to adverse weather conditions. Another was for a stated adapta-
tion (SA) survey, to prepare data for modeling of mode choice behavior. The RA
survH\ZDVQ¶WGHVLJQHGIRUPRGHOLQJ, but capturing the mode choice changes and the
SA survey was to model mode choice behavior.

Which option do you like to choose in each of the following scenarios?

No Trip purpose Circumstance Options
Road flood during the motorcycle 26,000 VND 36 minutes
1 Commuting
trip taxi 192,000 VND 28 minutes
Congestion just before motorcycle 13,000 VND 18 minutes
2 Shopping
departure time taxi 100,000 VND 14 minutes
Congestion during the motorcycle 38,000 VND 52 minutes
3 Leisure
trip car 189,000 VND 43 minutes
Heavy rain just before bus 19,000 VND 62 minutes
4 Other
departure time mc taxi 112,000 VND 42 minutes

Fig. 1. A sample questionnaire in the stated adaptation survey

Revealed Adaptation (RA) Survey: Each respondent was asked to reveal how
they would eventually change their travel in four cases, namely (1) encountering a
heavy rainfall just before the trip, (2) encountering a heavy rainfall during the trip, (3)
encountering a road flood just before the trip, and (4) encountering a road flood dur-
ing the trip. For each trip purpose (e.g.., commuting, shopping, leisure, and other),
each respondent is given a number of choices, including (1) cancelation, (2) post-
ponement, (3) destination change, (4) route change, (5) mode change to car, motorcy-
cle, bicycle, e-bike, and bus, and (6) no change. About 400 respondents were success-
fully interviewed, including 160 motorcyclists and e-bike riders, 80 car drivers, 80
bus riders and 80 cyclists. To ensure an optimal correspondence between the surveyed
Analysis of Mode Choice Behavior under Adverse … 961

samples and the actual population, the sample were weighted. The observations were
weighted to match the sample distribution and the population distribution.
Stated Adaptation (SA) Survey: Each respondent was presented with two options
of mode use in a hypothetical situation for making a choice ± combinations of trip
purpose, circumstance, and trip cost and trip time (see Figure 2). There are four trip
purposes considered, including commuting, shopping, leisure and others. There are 6
types of circumstance taken into account, namely (1) heavy rain just before moving,
(2) heavy rain encountered during moving, (3) road flooding just before moving, (4)
road flooding encountered during moving, (5) congestion due to heavy rain and road
flooding just before moving, (6) congestion due to heavy rain and road flooding en-
countered during moving. There are seven transport modes used in choice experiment,
including car, motorcycle, bicycle, e-bike, bus, taxi and motorcycle taxi. About 100
respondents were successfully interviewed, including 70 motorcyclists, 15 car drivers
and 15 bus passengers. A binary logit model was deployed to analyze the mode
choice decision-making mechanism as stated by individuals or road users.

3 Mode change behavior under Adverse Weathers

Patterns of mode change were captured based on the RA survey. Rates of mode
change and specific modal shifts are presented in the following sections. As shown in
Table 1, rates of mode change under heavy rain and flood in HCMC look rather small
in comparison with the figures in developed countries. Cools et al. (2010) found the
rates of mode change caused by heavy rains and snow, for examples, 15%-24% of
commuting trips changed mode, 14%-22% of shopping trips and 16%-26% of leisure
trips did so [2]. This study found that under the conditions of heavy rains and floods
the rates of mode change were less than 5%, 0.5%, and 3% for commuting, shopping,
and leisure trips, respectively. The lack of alternative transport modes in the city may
make change mode unfeasible. It can be observed for few cases that people park their
private vehicles and take a bus, taxi or even motorcycle taxi due to a heavy rain or
road flooding during their trips when they feel unsafe to continue driving themselves.

Table 1. Rates of mode change due to a heavy rain or flood by trip purpose
TP HRD HRT FBD FDT TTW
Rate 2.6% 1.4% 5.0% 0.1% 803 TP=Trip purpose;
Commuting HRD=Heavy rain before
Number 21 11 40 1
dep; HRT=Heavy rain
Rate 0.5% 0.0% 0.0% 0.1% 803
Shopping during the trip; FBD=Flood
Number 4 0 0 1
before dep; FDT= Flood
Rate 3.0% 0.9% 2.6% 0.9% 803
Leisure during the trip; TTW=Total
Number 24 7 21 7 number of respondents
Rate 0.7% 1.7% 0.9% 0.4% 803 (weighted).
Others
Number 6 14 7 3

Under the adverse weathers, the greatest mode change was 5.6% of the respond-
ents switching from motorcycle to car, followed by 3% of them changing from mo-
torcycle to taxi and 2.5% shifting from bus to car. Totally, 12.6% of the respondents
switched from space-saving modes (e.g., motorcycle and bus) to space-intensive car
962 A. T. Vu and T. T. H. Nguyen

and taxi. Further, while 1.9% of the respondents shifted to bus, 2.7% of the shifted
from bus to car and motorcycle. While only 2.4% of the respondents shifted from
bicycle and car to motorcycle, 10.5% of them switched from motorcycle to other
modes. This fact calls for the need to improve public transport services in order to
meet the increasing mobility demand as well as to provide more choices for people
moving around under bad weathers.

Table 2. A matrix of mode change rates among the mode-changed trips

C MC Bi B W O Total C: car, taxi
C 1.30% na na 2.58% na 3.88% MC: motorcycle,
M 38.69% na 13.57% 2.84% na 55.10% ebike, motorcycle
Bi 10.33% 18.08% 2.58% na na 30.99% taxi
B 4.84% 5.17% na na na 10.01% Bi: bicycle
W na na na na na na B: bus, shuttle bus
O na 0.01% na 0.01% 0.02% W: walking
Ȉ 53.86% 24.55% 0.00% 16.16% 5.42% na 100.00% O: other

As shown in Table 2, within the mode-changed trips, about 54% changed to car
and taxi, 25% to motorcycle and motorcycle taxi, 16% to bus and shuttle bus, and 5%
to walking. Among the mode-changed trips, 55% shifted from motorcycle, ebike, and
motorcycle taxi, followed by 31% from bicycle, 10% from bus and shuttle bus, and
4% from car and taxi.
Among the mode-changed respondents, more than nearly 40% decided to change
mode actually shifted to car and/or taxi. While road capacity is usually reduced be-
cause of heavy rain or flood, the shift from space-saving mode (motorcycle) to space-
consuming mode (car/taxi) has caused more serious traffic congestions in rainy or
flooding day. The shift to public transport (from motorcycle and car users) is only
about 10%, much lower than the shift to car/taxi. It is important to conduct an analysis
of influential factors to mode choice in heavy rain and urban flooding for a deeper
understanding and to support the formulation of strategies. The following chapter
presents the modeling of mode choice behavior under adverse weathers based on the
stated adaptation survey data.

4 Stated Mode Choice Adaptations to Adverse Weathers

In each hypothesized choice scenario, the respondent was asked to choose between
two available modes (i.e., Mode i and Mode j). In total, there were 400 hypothesized
choices made by the respondents (Table 3). Based on the hypothetical choice data,
three binary logit models were estimated using the R software. Results of model esti-
mation are presented in Table 4.

Table 3. The number of choice samples by choice set

Choice set (Mode i vs Mode j) Samples Mode I, % Mode j, %
Car/Taxi vs Other(Motorcycle, Motorcycle Taxi, Bicycle, Ebike, Bus) 238 33% 67%
MC/MC taxi vs Other (Car, Taxi, Bus, Bicycle, Ebike) 236 69% 31%
Bus vs Other (Taxi, Motorcycle, Motorcycle Taxi, Bicycle, Ebike) 158 53% 47%
Analysis of Mode Choice Behavior under Adverse … 963

Key findings from the model estimation can be summarized as follows. Firstly,
³WUDYHOWLPH´DQG³WUDYHOFRVW´DUHstatistically negative factors to mode choice, mean-
ing that people tend to choose a faster and cheaper mode. People choosing motorcy-
cle/motorcycle taxi and car/taxi are sensitive to both cost and time factors while peo-
ple choosing bus are not sensitive to both travel time and travel cost. The insignifi-
cance of travel time and travel cost to the choice of bus is because travel time by bus
is always much longer than by car and motorcycle and travel cost by bus is usually
fixed. Secondly, males tend to choose FDU RU WD[L WKDQ IHPDOHV 7KLV WHQGHQF\ LVQ¶W
observed in case of choosing motorcycle and bus. Thirdly, personal income has a
negative effect on choosing bus, meaning that lower income people tend to choose
EXV7KLVWHQGHQF\LVQ¶WREVHUYHGLQFDVHRIFKRRsing car and motorcycle. It suggests
that free ticket of public transport might be considered in adverse weathers to facili-
WDWHWKH³VKLIW´WUDIILFPDQDJHPHQWVROXWLRQV
Table 4. Estimated binary logit models of mode choice in adverse weathers
Model Car/Taxi vs Other MC/MC Taxi vs Other Bus vs Other
variable Est. Std.E. z-val P(>|z|) Est. Std.E. z-val P(>|z|) Est. Std.E. z-val P(>|z|)
¨Trip cost -0.007 0.002 -2.977 0.003 -0.012 0.002 -6.094 1.1e-9 -0.004 0.003 -1.259 0.21
¨Trip time -0.052 0.020 -2.637 0.008 -0.041 0.016 -2.573 0.010 -0.025 0.022 -1.122 0.26
Gender 0.578 0.321 1.801 0.071
Income -0.125 0.071 -1.755 0.07
Career dummy
JE 2.630 1.138 2.310 0.021
SW 1.398 0.798 1.752 0.08
Unskilled labor 1.241 0.417 2.975 0.002
Retired 1.706 1.117 1.527 0.13
Vehicle Ownership dummy
Motorcycle 1.929 1.072 1.800 0.07
Bus ticket 1.810 0.830 2.181 0.029
Trip Purpose dummy
Commuting -3.022 1.153 -2.621 0.009
Shopping -2.940 1.142 -2.574 0.010
Leisure -2.433 1.135 -2.142 0.032
Other -2.599 1.141 -2.278 0.022
Weather Condition dummy
Heavy rain before departure time -1.004 0.386 -2.603 0.009 0.801 0.458 1.747 0.080
Flood during the trip 0.606 0.452 1.343 0.18
Summary statistics
N°of Obser. 238 236 158
Null deviance 329.9 327.2 219
Res. deviance 269.4 271.7 193.6
Res.Df 228 233 150
Log-likelihood 269.4 271.7 193.6
AIC 289.4 277.7 209.6
6LJFRGHVµ ¶µ ¶µ ¶µ¶µ¶1; JE: Junior engineer, trainee; SW: Skilled worker, artisan

Fourthly, several career groups have positive effects on choosing car/taxi, particu-
larly young engineer, skilled workers, and artisans. Meanwhile, unskilled labors have
a tendency of choosing bus. This finding might suggest that trip reduction through
964 A. T. Vu and T. T. H. Nguyen

flexible work schedules could be considered for specific job groups to influence their
mode choice. Fifthly, motorcycle ownership has a positive effect on choosing car,
meaning that because a motorcycle is more vulnerable in adverse weathers, motorcy-
clists tend switch to a taxi or a car for being more comfortable and safer. Finally, it is
VR LQWHUHVWLQJ WKDW ³+HDY\ UDLQ EHIRUH GHSDUWXUH WLPH´ KDV D QHJDWLYH LQIOXHQFH RQ
FKRRVLQJPRWRUF\FOHEXWLWKDVDSRVLWLYHLQIOXHQFHRQFKRRVLQJEXV³)ORRGLQJGXULQJ
WKH WULS´ KDV D SRVLWLYH HIIHFW RQ FKRRVLQJ EXV EXW QRW VWDWLVWLFDOO\ VLJQLILFDQW 7KLV
suggests that if people know a heavy rain will happen soon just before the departure
time, they tend to shift from motorcycle to other mode or might consider choosing
bus. If accurate weather and traffic information are provided prior to a trip, it would
LQIOXHQFHSHRSOH¶VPRGH choice.

5 Conclusion

This study has provided an understanding of the patterns of mode changes and factors
to mode choice decision in adverse weathers, which can be useful for designing traffic
management solutions. Several key conclusions can be drawn from the study. Firstly,
public transport improvement and private transport control are important measures for
mitigating the impacts of adverse weathers on traffic system because the transfer to
car and taxi is notable while the shift to city bus and shuttle bus is very modest.
Secondly, as the values of saved travel time become much higher in adverse weathers,
pricing policy would be effective in influencing mode choice of car and motorcycle
users. Thirdly, provision of weather and traffic information prior to the trip and
provision of discounted or free tickets would encourage people to choose public
transport and discourage the choice of motorcycle. Fifthly, flexible working regims
shall be offered to groups of junior engineers, trainees, skilled workers, artisans and
motorcycle owners so that they would refrain from choosing a car in adverse
conditions. Finally, future works might include improving stated adaptation data and
analysis and design solutions and testing its impacts by simulation models.

6 Acknowledgements

The authors would like to acknowledge the financial support of the Ministry of Edu-
FDWLRQ DQG 7UDLQLQJ 02(7 IRU WKH RYHUDOO UHVHDUFK SURMHFW ³7UDIILF 0DQDJHPHQW
0HDVXUHV LQ 5HVSRQVH WR 8UEDQ )ORRGLQJ &DXVHG E\ +HDY\ 5DLQ DQG 7LGDO 5LVH´
Thanks to this sponsorship, the research team was able to conduct the questionnaire
survey on revealed weather event adaptations by road users in HCMC in 2018.

7 References
1. 86'HSDUWPHQWRI7UDQVSRUWDWLRQ³:HDWKHU5HVSRQVLYH7UDIILF0DQDJHPHQW´
2. Mario Cools and al., 2010. Changes in travel behavior in response to weather conditions:
whether type of weather and trip purpose matter? https://trrjournalonline.trb.org/doi/ref/10.3141/2157-03.
3. Hyland and al., 2018. Riders on the storm: Exploring weather and seasonality effects on
commute mode choice in Chicago. Travel Behavior and Society, 13, 44±60.doi:10.1016/j.tbs.2018.05.001
A Study on Motorcycles Overturning due to Strong Wind
and its Warning System

Vo Duy Hung1, Do Anh Vu2

1,2 The University of DaNang ± University of Science and Technology, 54 Nguyen Luong Bang
Street, Danang city, Vietnam
vdhung@dut.udn.vn

Abstract. Motorcycle is a typical transportation mean in Vietnam. In present,

motorcycles travel in strong wind conditions is always difficult and can lead to
unacceptable accidents that may have an adverse effect for people who ride the
motorbike. In this case, so many accidents are in relation to motorcycle because
of strong wind when the resident move on the bridges. In general, there are
various researches about car overturning in development country but the
overturning of motorbikes is quite new and not so much researches mentioned
about this. In current study, the authors give an overview of the effects of
motorcycles when subjected to win. Furthermore, authors try to build the risk
curves, analyze the data and provide a regulation system for motorcycle in strong
wind conditions. The warning work is based on wind speed and wind direction.
Then, based on the relationship of the wind speed and wind direction with
motorbikes velocity, the system will send the warning messages for motor-riders.
Especially, the renewable power also uses to operate the warning system which
can make it become more efficiency and sustainable.

Keywords: Motorcycle, overturning, warning system, wind speed, vehicle

velocity, risk curve

1 Introduction

Currently, along with the development of technology in construction, many long span
bridges are being built in many countries around the world. These works are often in a
airy places, it is therefore the vehicles travel along the bridges are often influenced by
horizontal winds, causing expectable overturning of vehicle. Especially, when the
vehicle passes through bridge tower, wind speed will be changed dramatically due to
wind flow separation, which greatly changes the equilibrium state of vehicle and make
it overturns. Countries around the world have also cared and studied this issue [1-5].
As in Japan, there have been researches and measures to regulate traffic given when
there is wind, such as restricting traffic of crossing bridges, speed restrictions, traffic
stopping and so on.
Vietnam is a tropical country and we always have effects by strong wind. However,
the research on wind effect still limited, especially the effects of wind to the traffic
through the bridge. In strong wind condition, many motorcycles can be overturned

© Springer Nature Singapore Pte Ltd. 2020 965

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_154
966 V. D. Hung and D. A. Vu

when crossing the bridge in Vietnam, like Tran Thi Ly Cable-stayed Bridge - Da Nang
City- Vietnam (Fig.1). The reason is that the impact of wind on dynamic balance of
motorcycles. Especially when traveling through locations where the wind velocity
changes abruptly like the bridge tower, this dynamic balance will change continuously.
Another reason can be attributed to wind characteristics that often have high-speed
gusts that endanger to the traffic travelling. In Da Nang city, the accident occurred when
people in the traffic moving through the bridge. This situation is still occurring
frequently and there are no remedial measures to make citizen traveling safely.
Therefore, it is unsettling to bridge operators and local government. This paper will
investigate the relationship between motorcycle speed with wind velocity and direction.
From that relation, this study will focus on developing the warning system for
motorcycle travel during strong wind condition.

Fig. 1. Overturning of the motorbike on Tran Thi Ly Bridge caused by strong wind

2 Aerodynamic force acting on a running vehicle

2.1 Side force acting on vehicle

To establish the relation between wind effect and motorbikes, this study considers
normal motorbike traffic with velocity Vr and wind acting on motorbike, which is
referred to wind speed Uc and wind direction (Ec). U is the combined wind velocity
effect on the motorcycle with the E angle. (Figure 2). Here, the maximum speed of the
vehicle is 70km/h, the critical wind speed is 25m/s (wind speed is forbidden to circulate
in Japan and Europe code). Wind direction (Ec) varies with values from 0-180 degrees.
In this paper, the wind-motorcycle relation will be consider and ignore the effects rain
conditions and motorcycle rider psychology.

Fig. 2. The effect of wind to vehicle

A Study on Motorcycles Overturning due to Strong … 967

Side force acts on motorcycle can be calculated by:

ଵ
ܲ ൌ ൈ ߩ ൈ ܷ ଶ ൈ ‫ݏܥ‬ሺߚሻ ൈ ‫ ܩ‬ൈ ‫݊ܣ‬ (1)
ଶ
Where:
An : Projected area of vehicles.
G : Gust factor (assumed to be 1.2).
‫ݏܥ‬ሺߚሻ : Side force coefficient. [6]
U : General velocity (Between wind speed and vehicle speed).

U U c2  Vr2  2U cVr cos(E c )

ߩ : Air density (kg/m3).

To ensure safety for vehicles involved in traffic, the standards of Japan and Europe
are often chosen for wind speed Uc = 25(m/s) making thresholds to close traffic. At the
speed of 25(m/s) corresponding to the big windstorm (Level 10 on Beaufort wind
scale), here we take Uc = 25(m/s) as the critical wind speed. Therefore, we take the
case that Uc = 25(m/s) and Vr = 0(km/h) is a dangerous limit case. Then, if the
horizontal force acts on the vehicle to overcome the limit, the probability of overturning
is very high (P < Pgh). At the same time, the limited horizontal side force (P gh) is
FDOFXODWHGFRUUHVSRQGLQJWRWKHKRUL]RQWDOZLQGDQJOHVRȕF žWKHFRHIILFLHQWRI
shock G = 1.2 and An: wind shield area according to the vehicle direction instead of
formula (1). We have: Uc=25(m/s), Vr=0(m/s)
ͳ
ܲ ௚௛ ൌ ൈ ߩ ൈ ሺܷܿ ൈ ‫ܿߚ݊݅ݏ‬ሻଶ ൈ ‫ݏܥ‬ሺߚܿ ൌ ͻͲ݀݁݃ሻ ൈ ‫ ܩ‬ൈ ‫ ݊ܣ‬ൌ ͷͻ͸Ǥͷ͸ሺܰሻሺʹሻ
ʹ

2.2 Risk curve for overturning

The risk curve for overturning for wind velocity from 10(m/s) to 25(m/s) can be
summarised in Figure 3. In general, the relationship between the side force, wind angle
and motorcycle velocity shows in each graph. It is clear that the overturning limit do
not cut the risk curve, so the motorcycle can travel safe in this range of wind speed.
However, when wind speed increases to 16 m/s, the side force cut the ³RYHUWXUQLQJ
OLPLW´ when travel with Vr= 50-70km/h, it mean that motorcycle can be overturn in this
case. Therefore, motorcycle should across bridge with speed under 40km/h. The same
scenario also found when wind speed increase to 20 m/s- 25 m/s. Especially, when wind
speed increase to around 22m/s and above, the risk of overturning is very high. It means
that these cases are very dangerous for motorcycles. Traffic shoud be closed when wind
speed exceed 20m/s for safety reason.
968 V. D. Hung and D. A. Vu

Fig. 3. Side force on a moving motorcycle

A Study on Motorcycles Overturning due to Strong … 969

3 Warning system for motorcycles

3.1 Wind measurement

In order to know the wind speed and wind direction, it is important to set up the wind
measurement system and real-time data transmission. From the wind data, the traffic
operator can regulate the velocity of motorcycles. The Figure 4 illustrates the wind
measurement system and data transmission. System includes anenometers, storgae
system (Internet cloud), display software, control device, and warning panel.
The anenometer system was developed based on founddation of Atmel ARM Cortex-
M3 CPU. The divices can capture the wind speed from 0-50m/s with accuracy to
0.1m/s, Wind direction from 0°-360°. The sampling frequency is 01Hz.
The anenometer can measure the real-time wind speed and dicrection. Based on the
wind data, the warning system can be figured out by the relationship between wind and
motorcycles, which was discussed in section 2.2. Furthermore, the input energy for this
system is solar energy.

Fig. 4. Wind measurement system and data transmission

3.2 Simulation software and wind speed prediction

To deliver the warning message, it is vital to build the warning gate through internet.
This study focused on developing on-line warning gate for citizen. The warning gate
include two main modules, the first is real-time wind data gate, which can display the
real-time wind data for the places where anenometers installed.
The second part of system is the wind map which can update all information to traffic
operaror as well as citizen. The wind map was built on platform of Google Map Api
(Application Programming Interface) which can update the wind data from
anenometers. Then, by calculating the relationship of wind and motorcycle, the warning
message can be delivered. The detail of warning information can be seen in Figure 5.
It showed the location, time, average wind speed and warning velocity for motorcycles.
Currently, the system was installed on Han river- Danang city, Vietnam. The wind map
address is at http://windmapdanang.tapit.vn/. Beside the online warning, the reguation
of mortorcycle can also be done by electronic panel in each end of bridges.
970 V. D. Hung and D. A. Vu

Fig. 5. Wind map interface ( Based on Google Map Api )

4 Conclusions

In this study, the relationship between motorcycle velocity, wind speed and wind
direction were investigated. In addition, the warning system for motorcycle also
developed. Important results obtained from this study are as follows:
x There is a strong relationship between motorcycle speed, wind velocity and wind
direction.
x When wind speed is 20 m/s or above, motorcycles should not travel during this
condition. For the other wind speed, motorcycle should travel safety during strong wind
when it runs with eligible speed.
x This study also successfully built the warning system under strong wind conditions
successfully. This system can be applied for many places, especially for the bridges.

References
[1]. Simiu E., Scanlan R.H(1996),Wind Effects on structures ± Fundamentals and Application
to Design, New York; John Wiley & Sons, 3rd ed.
[2]. ASCE (2010) , Minimum Design Loads for Buildings and Other Structures ± ASCE/SEI
7 ± 10.
[3]. Davenport A.G (1968), The Dependence of Wind Load upon Meteorological Parameters,
Proceedings of the International Research Seminar on Wind Effects on Buildings and
Structures.
[4]. Nakaguchi H., Hashimoto K. & Muto S.(1968), An experimental study on aerodynamic
drag of rectangular cylinders, Journal of Japan Aeronautics Space Science, Vol. 16(168).
[5]. Clough R.W, Penzien J. (1993), Dynamcs of Structures; New York; McGraw-Hill. 2nd
ed.
[6]. Hiromichi Shirato, et al., Fundamental Study on Warning for Vehicles Overturning due
to Crosswinds, ICWE14 CD Room.
 Bus signal priority by active signal program
± A case study in Ho Chi Minh City (HCMC), Vietnam

NGUYEN Thi Thanh Huong1[0000-0001-7099-6278]

1
Section of Transport Planning and Management
University of Transport and Communications, Campus in Ho Chi Minh City, Vietnam
ntthanhhuong@utc2.edu.vn

Abstract. The bus signal priority (BSP) has been applied in many cities world-
wide. However, it has not been implemented in Vietnam, particularly in Hanoi
and HCMC where traffic congestion happens every day due to i) the limitation
of space for expand road infrastructure; ii) the limited level of technology inte-
gration in Vietnam; iii) the budget constraint; iv) the poor traffic management
system; v) the traffic particiSDQWV¶ DZDUHQHVV DQG VR IRUWK Although the BSP
has been suggested by many experts, no feasible solution of BSP for HCMC
has been proposed. Therefore, this research shall focus on the application of
ASP (active signal controller program) aiming to prioritize buses in mixed traf-
fic flow in HCMC. The traffic management theory is combined with, the smart
signaling system technology (automatic vehicle location (AVL) and advanced
signal control systems) in this research. The BSP is expected to improve the bus
quality service, which will encourage people to use bus in daily activity instead
of motorbike. Besides, the economic loss and the air pollution caused by time
delay and petrol consumption is also expected to reduce.

Keywords: Bus in reserved lane, Bus signal priority (BSP), active signal pro-
gram (ASP), Ho Chi Minh City (HCMC)

1 Introduction

Buses can help to reduce the pollution, and be considered as a safe and efficient
transportation. They are an indispensable means of transportation in densely populat-
ed cities. In a motorcycle-dependent city as HCMC, the bus service quality is poor
since it is affected by mixed traffic and traffic congestion. Buses need to be treated as
priority vehicles in HCMC, where Green Transport project was recently approved by
Government. Priority measures for buses are classified in three main groups: i) priori-
ty signal measures; ii) lane measures; and iii) bus stop measures [1]. These measures
aim to eliminate or reduce the external disturbances at intersections (by priority traffic
VLJQDOVRQZD\VRIFRPPXWLQJDQGDWEXVVWRSV'HYHORSHGFLWLHV¶H[SHULHQFHVVKRZ
that these priority measures have minimized the delay and improved the speed, the
punctuality and the reliability of the bus. As the result, this helps to improve the bus
service quality [2]. Therefore, the urban transport problems shall be reduced if bus
service quality is improved by these priority measures.

© Springer Nature Singapore Pte Ltd. 2020 971

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_155
972 T. T. H. Nguyen

The lack of serious research on the topic in the specific HCMC conditions is one of
the reasons for the absence of bus priority measures in HCMC. The conditions in-
clude the road infrastructure, the traffic features, the driving conditions, and forth.
The suitability and effectiveness of each priority measure varies depending on each
FLW\¶VFRQGLWLRQ%XVSULRULWy studies must be tailored to the city's actual condition.
Therefore, this research focuses on BSP solutions and manual technique of calcula-
tion methods for HCMC. This research is structured in 4 chapters. The 1st chapter
introduces +&0&¶VWUDQVSRUWV\VWHm issues and BSP strategies in previous research.
The second chapter describes methodology and data collection. The third chapter
presents the results of bus priority signal solution and estimated travel time and fuel
consumption saving. The 4th chapter draws conclusions and recommendations of this
study.

2 Methodology

Based on the BRT experiences in other countries and in Hanoi, application of a bus
UHVHUYHG ODQH ZLWK VLJQDO SULRULW\ RQ +&0&¶V URDGV VKDOO EH FRQVLGHUHG 7KH FDVH
study ± Dien Bien Phu main street (DBP) is chosen to propose solution and manual
technique calculation of signals, then to evaluate the effectiveness.
DBP is the road on the East side of HCMC. Its length is 7.3 km. Congestion fre-
quently occurs from 7am -8 am, from 11:30 am -12:30 pm and from 4pm -6:30 pm.
DBP is one-way road from the Nga Bay roundabout to the Dien Bien Phu-Dinh Tien
Hoang intersection. The distances between intersections along DBP are about from
700-1000 m. Many intersections have affected the travel time of passengers on DBP
road. The intersection Dien Bien Phu ± Nam Ki Khoi Nghia (DBP-NKKN) is chosen
for applying manual technique calculation. The results of this studying will be used
for further study of BSP for all along DBP Street and other roads in HCMC. DBP is
9.6 m of width, 3 lanes of 3.2 m, 1 pavement of 4 m and another of 2 m. Nam Ki Khoi
Nghia (NKKN) is 16 m of width, 4 lanes in two ways, 1 lane divider of 0.5 m and 2
pavements of 4 m. Traffic and infrastructure survey was undertaken on DBP road and
at DBP-NKKN intersection. Data of the signal cycle, flow speed, traffic volume, ve-
hicle occupancy rate, traffic composition, and bus services are collected. Road capaci-
ty at DBP-NKKN intersection in Table 1 is calculated according to HCM 2000.

Current state Traffic survey (filming and Proposing of traffic

studying of road data analysis) organizing in priori-
and bus network Measurement of current tizing bus signal
Case study- infrastructure dimensions
Literature review ing DBP Signal cycle calcula-
of bus signal Observing of traffic behavior
main road & and analyzing tion
priority strategy DBP-NKKN
Studying of travel Collecting of signal cycle, Evaluating the effec-
intersection flow speed, traffic volume, tiveness of proposed
demand and
traffic features of vehicle occupancy rate, solutions supported
HCMC traffic composition by VISSIM

Fig. 1. Studying framework

Bus signal priority by active signal program … 973

Table 1. Current lane capacity at intersection

Lane So Fw Fbb Fa % TL % TR Flt Flr Si S C

T1 2000 1 0.98 0.9 0.6 0 0.97 1 1718
T2 2000 1 0.98 0.9 0.1 0 0.99 1 1756
6997 3304
T3 2000 1 0.98 0.9 0.1 0 0.99 1 1756
T4 2000 1 0.98 0.9 0 0 1 1 1767
N1 2000 0.99 0.96 0.9 0.4 0.3 0.98 0.99 1667
N2 2000 0.99 0.96 0.9 0.4 0.3 0.98 0.99 1667 5033 2079
N3 2000 0.99 0.96 0.9 0 0.3 1 0.99 1699
D1 2000 0.51 0.98 0.9 0 0 1 1 892
D2 2000 0.7 0.98 0.9 0 0 1 1 1229 3350 1582
D3 2000 0.7 0.98 0.9 0 0 1 1 1229
B1 2000 0.99 0.96 0.9 0 0 1 1 1707
B2 2000 0.99 0.96 0.9 0 0 1 1 1707 5121 2134
B3 2000 0.99 0.96 0.9 0 0 1 1 1707

Note that: So: saturation flow of an urban road lane 3.5-3.75 m of width in stand-
ard conditions (pcu/h); Fw: coefficient adjusted according to lane width; Fbb: coeffi-
cient adjusted due to bus stopping at intersection; Fa: coefficient adjusted according
to area (urban center 0.9, others 1); %TL: % vehicle turn left at intersection; %TR: %
vehicle turn right at intersection; Flt: coefficient adjusted due to turn left flow; Frt:
coefficient adjusted due to turn right flow; Si: saturation flow of the lanes at the inter-
section. Si is calculated by the formula: Si = So.n.fw.fhv.fg.fp.fbb.fa.flt.frt
(HCM2000) (pcu/h); S: saturation flow of the legs in the intersection. S is calculated
by sum of Si; Ca: capacity of the legs at signal intersection which is calculated by
ୗ‫כ‬୥೔ ௣௖௨
formula: Ca = ሺ ሻ, in which gi: effective green time of phase i (s), Cy: time
େ୷ ௛
cycle (s).
Cycle time signal prioritizing bus and effective green time (gi) of phase i are calcu-
lated by using the Webster formulas (1) and (2):
ଵǤହ௅ାହ ௬௜ሺ஼ௗ௘௦ି௅ሻ
COP = (1) gi = (2)
ଵିσ ௒ ௬

In which: Cop: cycle length in seconds; L: total lost time = (no. of phases)*(lost
time per phase) + (all red time); Y: sum of the critical lane flows/ (1900); yi: Traffic
volume of critical lane i; y: Total traffic volume of critical lanes of all phases; Cdes:
Cycle time designed.
The bus signal control program to prioritize buses was proposed as in the Figure 2.
To measure the effectiveness of traffic organization solution in prioritizing bus pro-
posed, 2 groups of indicators are used, which are the technical group including ca-
pacity, occupancy rate of capacity (OCC), time delay and level of service (LOS); and
the socioeconomic group including time saving for bus passengers and fuel consump-
tion reduced. The occupancy rate of capacity is calculated by the following Webster
௏
formula: OCC = , in which: V: designed traffic flow, Ca: capacity of intersection
஼௔
leg. The average flow speed, the time delay and the LOS are estimated by the simula-
tion model. The results help to evaluate whether the intersection is over or under-
saturated in the existing condition. The simulation model of NKKN-DBP in VISSIM
was built, calibrated and validated with the goal of obtaining the same results in the
traffic survey data at current state, including the traffic volume and the traffic speed of
974 T. T. H. Nguyen

each direction. The parameters of driving behavior and the distributions of traffic
were inputted into the model and modified in order to calibrate the model. This pro-
cess would continue until the model results met the validation targets and the traffic
model reached a level of fidelity of more than 85%. Results will be presented in fol-
lowing chapter.

Doing nothing
Detecting of bus, transmission of

a•X
information to the control center

Examination of signal on DBP

Remaining
Green Green time lasting
green time a (s)
a<
Street after y (s)

X-a (s)

Return to green Turn on yellow 3(s) then

Yellow UHG ;¶-3 (s) for NKKN
X (s)
Street. Then, turn on
green X (s) for DBP
c•;¶ Street. After that, return-
Remaining red
Red ing to normal cycle
time c (s)
c<;¶ Doing nothing

Fig. 2. Bus priority signal control program at intersection

3 Results

To insert a bus reserved lane on DBP Street, DBP cross section is modified as in the
following figures 3 and 4. At the same time, it is proposed changing from the large
bus to smaller bus in order to fit the bus reserved lane of 2.6 m. Because the average
occupancy rate of buses on DBP street is only about 0.5, we can change all buses
from B80 to B50 having dimensions as (D × R × C) 7080 × 2035 × 2780 (mm). Be-
sides, we propose change bus schedule of 5 lines assuring the frequency of 3 minutes
per bus passage in maximum because the cycle timing is about 72-76s. It is noted that
the width of a standard urban lane in Vietnam is 3.25 m (from 2.75 to 3.75 m accord-
ing to design speed from 30-100 km/h). However, in Hanoi and HCMC, there are
many limited-width routes of 2-2.75 m, where are only good enough for low-speed
traffic of 15-30 km/h. Observed average flow speed on DBP street is about 19 km/h in
peak hours and 22 km/h in normal hours. In addition, on the 2.5 m-of-width lane, cars
can go in a row at speed of about 30 km/h. On 1-1.7 m-of-width lane, motorcycles can
go in one row. On over-1.7 m-of-ZLGWKODQHPRWRUF\FOHVFDQJRLQWZRURZV7KDW¶V
why, considering the lack of space to expand routes for bus-reserved lane in HCMC
downtown, 2.5 m-of-width lane for cars and 2 m-of-width lane for motorcycles are
proposed. Indeed, the standard is just for reference, appropriate design solutions could
be applied depending on the specific conditions. The private transport traffic is not in
priority, so that we can only guarantee for them a just enough comfort lane.
The parameters in the bus priority signal control program showed in the Figure 2
were calculated as in the Table 2. After prioritizing the private lane for the bus, the
traffic volume of individual motorized vehicles on DBP road would be changed be-
Bus signal priority by active signal program … 975

cause of three main reasons: some people would use the bus instead of private means;
others would choose alternative routes due to heavy traffic; some even canceled the
unimportant trips due to congestion. The forecasting of the number of motorcycle
trips transferring to bus and the number of private motorized users would choose an-
other route will be presented in another study. In this paper, it is assumed that 10% of
private-motorized-vehicles trips would shift to bus and 5% of private motorized users
would change to another route. This assumption will be discussed in another study.
We recalculate the lane capacity of DBP Street at intersection and the traffic volume
in DBP Street after inserting bus reserved lane. New optimal signal cycle timing is
recalculated by the Webster formulas.
Pavement 2m Pavement 4m

Fig. 3. Current state of DBP cross section

Bus reserved lane

Pavement 2m Pavement 4m

Fig. 4. DBP cross section proposed with bus reserved lane

Table 2. BSP program calculating results

Exit length path of bus at intersection (1) 13.25 m

Exit speed of bus at intersection with signal priority (2) 30 km/h
Bus safety override time at intersection (3) = (1)/(2) 2.31 s
Cushioning time for vehicles on DBP over the node and vehicles on NKKN enter the node (4) 1.48 s
Cushioning time for vehicles on NKKN over the node and vehicles on DBP enter the node (5) 1.77 s
Bus detecting time (6) 1s
Information processing time at control center (7) 1s
Information transmission time to control center (8) 1s
Information transmission time to signal lamp (9) 1s
Scheduled stop time at bus stop (10) 15s
Distance from nearest bus stop to signal lamp (11) 46m
Minimum distance of bus detecting (12)=min[(3s+5+6+7+8+9);(11)] = min(73;46) 46m
Y = (10) + (11)/(2) ± (3s+(5)+(7)+(9)) 13.75s
X = (3)+(7)+(9) 4.31s
;¶ V 4.77s

The effectiveness evaluation of BSP application at DBP Street is showed in the

Table 3 and 4. In case that BSP is applied to all along of DBP Street, DO consump-
tion of running bus reduce 0.37 l per trip due to the saved delay time at the intersec-
tion, and travel time of passenger by bus reduce 5 minutes per trip. This means that
we could save in total about 55 billion VND per year for fuel consumption by bus and
77 billion VND per year for waiting time saving at intersection by passengers (See in
976 T. T. H. Nguyen

the Table 3). Besides, the level of service (LOS) in private motorized lanes would be
almost the same as current state (See in the Table 4).

Table 3. Evaluation of the effectiveness using socioeconomic indicators

After application of In which:
Before solution
solution 1 USD2016 =
Number of bus trip per hour (1) 20 20 22655
Number of passenger per bus trip (2) 60*0.6*2.5*40%=36 40*0.91*2.5*40%=36 VND2016;
VOC: bus
Average time delay (s) (3) 49.6 0
vehicle opera-
GDP per capita 2016 (USD) (4) 2215 2215 tion cost
Bus fuel consumption (l/h stopping at stopping at
3 3
intersection) (5) intersection;
Price of 1l DO (VND) (6) 13,480 13,480 VOT: value of
VOC = (1)*(3)*(6)*(5)/3600 (VND/h) 11,134 0 travel time of
passengers
ሺଷሻ ሺସሻ‫כ‬ଶଶ଺ହହ
VOT =(1)*(2)* * (VND/h) 97,416 0 stopping at
ଷ଺଴଴ ଷ଺ହ‫כ‬ଵସ intersection.

Table 4. Evaluation of the effectiveness using technical indicators

Capacity VCRL Average time
Direction OCC LOS Note that:
(pcu/h) (pcu/h) delay (s/pcu)
NKKN 3304 1822 0.55 40 E
OCC: occupancy
Current state DBP 2079 1312 0.63 49.6 E
rate of capacity
Total 1 5383 E
NKKN 3314 1822 0.55 31.5 E
After bus priority signal LOS: level of
DBP 1604 1053 0.65 41.7 E service
solution applying
Total 2 4961 E

4 Conclusions and recommendations

The bus priority signal control program proposed in this study is successfully calcu-
lated and evaluated the effectiveness on the technical and socioeconomic aspect. De-
spite the lack of space, road section is rearranged neatly, a lane is reserved to bus and
the service quality for private motorized flow is maintained as well. This study con-
tributes a manual technique calculation of BPS control program for a real case study
in urban traffic condition of HCMC. The value of this contribution will be improved
in further study for all intersections on DBP Street and generalize solution for other
streets of HCMC. Besides, further generalizations of the findings are also possible by
shifting the scope toward evaluating impacts of solutions for not only bus passengers
and traffic on DBP Street but also private motorized transport users and traffic on
NKKN, or toward coordinating priority signal of all intersections along the street.

Reference
1. Peter T. Martin and Milan Zlatkovic (2010) Evaluation of Transit Signal Priority Strategies
for Bus Rapid Transit on 5600 West Street in Salt Lake County, Utah
2. Nguyen Van Nam (2013) Bus priority in motorcycle dependent city. Department of Civil
Engineering and Geodesy, Technische Universität Darmstadt, ISSN 1613-8317.
 Modeling land use change based on Remote Sensing, GIS
and Algorithm Cellular Automata decision support sys-
tem for urban sustainability planning in Quy Nhon, Binh
Dinh province central Vietnam
Tho Van Phan1[0000-0002-0887-5966] Tu Anh Ngo1[0000-0002-8909-7755] Huyen Thi Nguyen1[0000-0001-5780-
8869]
Rute Sousa Matos2[0000-0002-7715-0256] and Trang Thuc Dang3[0000-0002-5097-4095]
1
4X\QKRQ8QLYHUVLW\$Q'XRQJ9ѭѫQJ4X\QKRQFLW\9LHWQDP
2
University of Évora - CHAIA, Largo Marquês de Marialva 8, 7000-809 Évora, Portugal
3
Scene Plus Co. Ltd, 20 Dao Duy Anh, Phu Nhuan District, HCMC, Vietnam
phanvantho@qnu.edu.vn
ngoanhtu@qnu.edu.vn
nguyenthihuyen@qnu.edu.vn
trang.dangthuc@sceneplus.vn

Abstract. A study on the prediction of land use change and its spatial and tem-
poral variability had been studied for Quy Nhon city for a period of 18 years via
land change model with approaches based on remote sense images obtained
from sensors Landsat and Markov string was used to model land use change. In
order to predict land use changes, the study conducted on the creation of map-
ping land use classification from remote sensing image and to assess the accu-
racy of the classification result by the Kapa index. Validation will be essential
to the prediction of land use changes and future urban expansion, until 2027 and
2035, contributing to the orientation of future land use planning.

Keywords: Modeling, Land use change, Quy Nhon city, Urban growth.

1 Introduction

In the period of rapid urbanization and overpopulation, the land has become every
nation's problem and sometimes one of tKHFULWHULDXVHGWRDVVHVVDQDWLRQ¶VGHYHORp-
PHQWLVWKHODQGXVHFKDQJHRIWKDWFRXQWU\,W¶VWKHSRSXODWLRQJURZWKWKHGHYHORp-
ment of urbanization and industrialization that put great pressure on land use. Landuse
FKDQJH¶V D PDWWHU FRQFHUQHG E\ ERWK UHVearchers and managers because of its close
relationship with social and ecological systems [1].
4X\1KRQ¶VDFRDVWDOFLW\LQ%uQKĈӏQK3URYLQFH,W¶VFRPSRVHGRIZDUGVDQG
communes with a total area of 28,536.62ha. In recent years, the speed of urbanization
in Quy Nhon city has raised dramatically. Agricultural land is rapidly shrunk for resi-
dential areas, industrial parks, and other public works. A long-term and large-scale
IRUHFDVWRIODQGXVHFKDQJH¶VQHFHVVDU\for the city's governing bodies,W¶VDFRPSOH[
task that requires much time, high expense. Moreover, the limit in the area of the
update implementation and unified information creates a huge obstacle.

© Springer Nature Singapore Pte Ltd. 2020 977

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_156
978 T. V. Phan et al.

2 Study Area General Setting

Quy Nhon has a varied topography, being extremely diversified (see Fig. 1).

Fig. 1. The study area of Quy Nhon city

3 Data and methods

3.1 Research data
The Landsat satellite images was collected over the time in 2000, 2009 and 2018. The
details of images were illustrated in Table 1. To serve the image interpretation, 80
samples were collected by GPS.

Table 1. Table captions should be placed above the tables.

Year Code Date taken Quality Cloud

2000 LE07_L1TP_123051_20001108 11/08/2000 92% 8%

2009 L71123051-05120090322 22/03/2009 98,8% 15%

2018 LC08_L1TP_123051_20181102 02/10/2018 99% 10%

3.2 Research methodology

Remote sensing image classification method. The research used the object-oriented
image classification methods of remote sensing images in the years of 2000 - 2018
with the support of eCognition software. Classification results have checked the accu-
racy through Kappa coefficient formula [2].

Methods of fuzzy logic. 7KH IX]]\ ORJLF PHWKRG¶V RIWHQ XVHG LQ WKH PXOWL-criteria
evaluation as well as spatial information modeling problems [4]. This study has iden-
tified that the unit type of land use: 1) Urban land; 2) Unused land; 3) Forest land; 4)
Modeling land use change based on Remote Sensing … 979

Agricultural land; 5) Water surface are the key factors that related directly to the ur-
ban development of Quy Nhon city.

Markov - CA model method. 0DUNRY¶VDVHTXHQFHRIX0, X1, X2,... including random

variables:
P(Xn1 x | X 0 , X1 , X 2 ,..., X n ) P( X n 1 x | X n ) (1)
A simple way to visualize a specific type of Markov chain is through a finite
Otomat [4],IWKHV\VWHP¶VLQ-state y at the time n, the probability that the system will
switch to state x at time n + 1 does not depend on the value of time n only depends on
the current state y:
P(Xn+1 = x| Xn = y) (2)
In which element x, y has a value and inde-pendent on the time index n.

Probability Matrix Construction of Land Use Change. The nature of the Markov
FKDLQ DQDO\VLV¶V WR EXLOG WKH UHODWLRQVKLS EHWZHHQ WKH WZR ODQG XVH PDSV DW WKH WZR
points of assessment to create a scientific basis for the modeling process in the next
steps [5].

3.3 Framework for the study

In order to forecast land use changes, the study conducted an assessment of changes
in land use type of the studied area in the period of 2000 ± 2018.

Accuracy
Image remote sensing data Image classification Markov chain
assessment

Reference 2018 Simulated 2018 Probability

Maps of driving factor Validate
land use map land use map conversion matrix

Initial land use map Forecasting land use map CA – Markov

Fig. 2. The framework for the study

4 Results and Discussion

4.1 Classification of Remote Sensing

In this study, we interpreted images into land use classes as follows: urban land; water
surface; agricultural land and unused land.
980 T. V. Phan et al.

Fig. 3. The results of remote sensing image classification in 2000, 2009 and 2018

4.2 Construction Markov chain

In this study, we use the results of image interpretation from 2000 to 2009.

Table 2. Probability matrix of land use change from 2000 to 2009

1. Unused 2. Agricul- 3.Urban 4. Forest 5. Water

2000 - 2009
land tural land land land surface

1. Unused land 0.7220 0.0840 0.0938 0.0830 0.0172

2. Agricultural land 0.0579 0.7416 0.1308 0.0478 0.0219

3.Urban land 0.0562 0.0039 0.9100 0.0150 0.0149

4. Forest land 0.0503 0.0222 0.0517 0.8707 0.0051

5. Water surface 0.0420 0.0149 0.0360 0.0112 0.8960

4.3 Modeling land use change

The factors affecting land use change are also affected by the characteristics of the
uncertainty of GIS data, so to determine the thresholds of data we used Fuzzy logic by
dividing the data into levels to assess the appropriateness for each type of land use
that that factor impacted.

Table 3. Reclassification of appropriate factors affects land use types.

Degree of suitability for urban development

Factors affecting Not suitable Least suitable Suitable Most suitable
(S=0) (S=75) (S=128) (S=255)
Slope (degree) > 25 15 ± 25 8 ± 15 0±8
Distance to roads (m) >1000 500 ± 1000 200 ± 500 0 ± 200
Distance to residential (m) > 1000 500 ± 1000 200 ± 500 0 ± 200
Distance to tourist destina-
>15000 5000 ± 1500 2000 ± 5000 0 ± 2000
tions and services (m)
Modeling land use change based on Remote Sensing … 981

5DVWHUL]HGGDWDKDVDJUD\YDOXHIURPWRDQGLW¶VLQFOXGHGLQ,'5,6,VRIWZDUH
to evaluate multiple criteria. Therefore, when appropriately hierarchizing these data in
IDRISI, depending on the hierarchical number, WKHVFDOH¶VGLYLGHGLQWRWKHUDQJHIURP
0 to 255, in which 0 is the nonconforming value and gradually increases up to 255 for
adaptive levels of future urban development.

Fig. 4. Demarcation of factors: a. residential area; b.tourist spots- services; c. traffic; d. slope
These areas will be eliminated in the process of selecting the input for the model
when calculating. Since then, the model results will have a high level of accuracy for
the next years when forecasting the speed of space development of the city. Restricted
and unsuitable areas for development will carry zero values and areas that affect ur-
ban areas will have the value of one.

Fig. 5. A suitable hierarchical image for different types of use a, unused land; b, agricultural
land; c, urban land and d, forest land; e, river and stream land; g, limited development area
The modeling results achieved a fairly high accuracy rate (more than 95%) with re-
sults of Landsat image classification in 2018. Statistical results of forecast land use
types are compared with 2018 classification results.
982 T. V. Phan et al.

Fig. 6. Forecasting land use map; a, 2027; b, 2035

The land has changed greatly with the increase in urban land area (accounting for
37.44% of the natural area) in 2035. The area of forest land is the largest land (ac-
counting for 46.86% of the total natural land area). Meanwhile, the area of unused
ODQGDFFRXQWVIRUWKHORZHVWUDWHRIWKHWRWDOQDWXUDOODQGDUHDZKLFK¶VQearly
equal to the area of rice land (3.68%).

5 Conclusion and recommendations

The project has used technology to model land use change in Quy Nhon city in 2018
based on the results of classification in the period of 2000 - 2009 and the factors af-
fecting the land use change, verified the comparison between the results of modeling
in 2018 and the classification of 2018, which showed that the modeling results had a
high accuracy rate, nearly 95% compared to the results of Landsat image classifica-
tion in 2018.
Through Markov chain analysis model combined with Automatic Network algo-
rithm, the study has predicted the land use change of the studied area up to 2027 and
2035, with the decrease of vacant land, forest land, and rice land while urban land
increased sharply.

References
1. Tang, Z., et al.:'Forecasting land use change and its environmental impact at a watershed
scale'. Journal of environmental management, 35-45 (2005).
2. Nobre, R., et al.:'Groundwater vulnerability and risk mapping using GIS, modeling and a
fuzzy logic tool'. Journal of Contaminant Hydrology, 277-292 (2007).
3. Chi, V.K., et al., Coastal urban climate resilience planning in Quy Nhon, Vietnam.
International Institute for Environment and Development (2015).
4. Aaviksoo, K.:'Simulating vegetation dynamics and land use in a mire landscape using a
Markov model'. Landscape and Urban Planning, 129-142 (1995).
5. Sui, D.Z.:'A fuzzy GIS modeling approach for urban land evaluation'. Computers,
environment and urban systems, 101-115 (1992).
 Modeling and applying heuristics for optimization of
solid waste collection under consideration of vehicle
capacity

Hai Dung Dinh1

1 Vietnamese-German University, Binh Duong, Vietnam

Abstract. The main contribution of this paper is to propose a general network

flow optimization model for the problem of solid waste collection related to
garbage management and transportation route planning. A model is formulated
for a specific case in a big city in Vietnam. With the continuous increase of ur-
ban population in big cities of a developing country and the arising problem of
waste management over the years, attempts have been made to organize the
process of collecting solid waste from households and a series of tasks to ensure
the best logistical steps. Considering the network of depots and information
about the vehicles, we propose to apply heuristic algorithms from Operations
Research to optimize the routes for vehicles to travel to depots and back. The
research question thus concentrates on sequencing and routing with similar so-
lution concepts as a classical Traveling Salesman Problem. For generating solu-
tion and implementation, we apply Evolutionary Algorithm using a commercial
Solver tool. In general, network problem include cost of moving materials
through a network involving varying demands, parameters, and constraints de-
pending on the locations that the materials are being brought to. Problems of
these type are characterized network flow optimization. The consideration of
the connections between different parts of the network with inputs and outputs
is what makes these problems difficult, but on the other side quite practical and
applicable. The visual effect of a network flow model makes it useful for plan-
ning activities on the management level. Using numerical examples, we show
that the efficiency of the evolutionary algorithm in the defined problem instanc-
es is better than any greedy approach. Finally, we discuss the similarity between
the proposed model for solid waste collection and the problem of routing order
pickers in multiple block warehouses.

Keywords: Operations Research, Network Flow Optimization, Heuristics, Gar-

bage Management, Routing Problem

1 Introduction

In business, there are many problems that require the manager to find the best route
to solve the issue. Some of those are simple problems which can easily be solved with
basic mathematical skills. However, others take the form of equations with various
variables, even up to tens or hundreds. Mathematical programming approaches max-

© Springer Nature Singapore Pte Ltd. 2020 983

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_157
984 H. D. Dinh

imize or minimize an objective function for improvement, to ensure operational effi-

ciency and also serve as decision support tools. Solid waste management includes
various issues such as the generation, collection, transport, treatment, and disposal of
solid waste. With the rising fuel costs, labor, vehicle maintenance, and increased vol-
ume of solid waste from population and consumption growth, solid waste collection
and disposal costs constitute 75±80% of a city's solid waste management budget.
Therefore, even a small improvement in the waste collection and transportation can
lead to significant savings in costs. Trucks collect the garbage from different locations
of cities and transport to depots which are incinerators, landfills and transfer stations.
This is the reason why projects such as "Solid Waste Transport Planning" gain greater
importance. In this paper a general network flow optimization model for the problem
of solid waste collection related to garbage management and transportation route
planning is formulated and solved. Finally, we discuss the similarity between the
proposed model for solid waste collection and the problem of routing order pickers in
multiple block warehouses.

2 Literature review

Solid waste management. For a better overview of the modeling of solid waste gen-
eration see for example the research of Beigl et al. [3]. There is also a relationship
between the cost of collection and frequency of collection, and especially the density
of population in an area [5], [6]. The model estimation has a variety of techniques
such as the time-series analysis [7] or system dynamics [8]. The input for such model
can be stochastic or deterministic, and can be done with diverse forecasting tech-
niques [9]. Having input data for the model, search procedures are conducted to find
optimal solutions through algorithms and given computing power. The routing opti-
mization problem in solid waste management has typically been addressed with dif-
ferent types of mathematical algorithms. Routing algorithms use a measuring system
called a path length to determine the ideal route to a defined destination. The optimal
routes are then determined by comparing the different paths.

Solid waste management proved its importance for all countries and municipals, such
as in India [10], Iran [11], Kenya [12], Ghana [13]. The optimization of collection
vehicles and transportation routes were discussed in [14], [15]. Another interesting
study to be mentioned is the one of Jovicic et al. [2] which employed a software
called ArcGIS Network Analyst to calculate the shortest solid waste collection route,
with the goal to reduce overall fuel costs, for City of Kragujevac, Serbia. In that prob-
lem, a number of approximately 4,000 waste bins at 2,000 locations within 12 city
collection districts was considered. The optimization result is as follows: The original
collection route length was 30.9km, while the new method produced a waste collec-
tion route of 22.2km in length, for a 28.1% decrease in overall kilometers traveled.
7KLV FUHDWHG D SRWHQWLDO VDYLQJV RI NP GULYHQ SHU \HDU EDVHG RQ WKH YHKLFOH¶V
schedule of six collections per week. For the purpose of reducing environmental ef-
Modeling and applying heuristics for optimization … 985

fect, the decrease in CO2 emissions was also calculated, based on the reduction in
kilometers driven the city would produce 40 less tons of CO2 emissions each year.
Thus, similar results are expected with an optimization model for other cases.

3 Model and Solution

3.1 The model

Consider the problem of Ho Chi Minh City in Vietnam. For simplicity let us as-
sume that the central depot is located in Da Phuoc, Binh Chanh District in the South-
ern part of the city, from which trucks collecting solid waste depart. Without loss of
generality, consider a set of 9 other locations to be travelled through, each represents
a district of the city. Figure 1 illustrates the map and the network form of the problem.
We are given a matrix with pairwise distances or costs cij between any node i and
QRGHMIRUQQRGHV«Q7KHREMHFWLYHLVWRGHVLJQDWRXUWKURugh the n nodes be-
ginning and ending at the node of origin, denoted as node 1, such that the total dis-
tance or traveling cost is minimized.

Fig. 1. Map of the city and the network form of the problem.

3.2 The Solver

Optimizing the collection route with advanced routing software is a common ap-
proach to reducing travel costs. There are many routing programs available for con-
ducting route analysis and optimization. The Vehicle Routing Problem can be solved
in Spreadsheet form as presented in the next section. This has many advanced features
and allows decision makers to customize different parameters already built on a given
form, for example a cost matrix. This provides an expedient solution for many differ-
ent routing scenarios. The Solver is used to find the quickest or shortest route between
two or more points on a network. The Solver for the model in this paper only consid-
ers a single route, though in case of extensions it can solve multiple routes with more
parameter settings such as breaks, time windows, or vehicles with compartments.
Basically, the solver can reorder the set of locations to find the optimal route. Alt-
986 H. D. Dinh

hough this paper focused on evaluating the shortest routes, more problem instances
with objectives such as minimal traveling time or highest efficiency can be addressed.

3.3 Heuristic algorithm and the solution

It is obvious that the increase in the number of visited locations means an exponen-
tial raise in complexity of the optimization problem. In parallel to the development of
exact search methods, another stream of research deals with heuristic procedures for
solving these same problems [1]. The routing problem can be solved using different
types of heuristics. Some of the routing algorithms used include Simulated Annealing,
7DEX6HDUFK*HQHWLF$OJRULWKP$QW&RORQ\2SWLPL]DWLRQDQG'LMNVWUD¶VDOJRULWKP
[18]. Particle Swarm Optimization was applied in [16]. A detail analysis for a case in
the US was done by Clark and Gillean [17]. With the evolutionary algorithm using the
MS Excel Solver tool the problem is handled easily without concern on the number of
visited locations. The most important step is to build up a correct model for each prac-
tical implication of the proposed approach. We use this consequence of numbers (for
example 1, 2, 3, 4 and 5) to find the consequence of visiting locations. If we succeed
in making the formal description of the problem in MS Excel, the Solver will try to
find the solution. An MS Excel spreadsheet is used to enter the input data. In order to
solve the problem with the evolutionary algorithm in Solver, we must change it to the
standard form: All constraints are expressed in equations and all variables are set to
be non-negative. Here, we apply the Alldifferent constraint to satisfy the requirement
of stopping at each location once.

Fig. 2. The distances as a cost matrix and optimal solution.

A greedy solution which search for the next shortest route will deliver a distance of
160.4 km by travelling through 1-4-3-10-2-5-8-7-6-9-1. The optimal solution found
by the Solver is 1-4-3-10-2-5-6-7-9-8-1 with a total distance of 144.1 km as shown in
Figure 2, meaning a 10% reduction of travelled distance. Consider the capacity of the
waste collecting trucks. If a truck can carry a total weight of X tons and the demands
differ from location to location x1 to xN, we want to identify the best solution. If within
one smaller network of 3-5 locations a best route is to be found and the connectivity
of these sub-networks is given, the idea is then to solve each network as a sub-
problem.
Modeling and applying heuristics for optimization … 987

In the research of De Koster and Van der Poort [4], methods and algorithms for
routing order pickers in a warehouse with multiple cross aisles were presented and
discussed, due to the fact that warehouses are vital in a supply chain. Similarly to the
problem handled in this paper, an optimization task is then to improve the goods pick-
ing processes to reduce goods picking turnaround time and transportation costs, lead-
ing into an overall reduction of makespan. Figure 3 presents the problem and some of
the heuristics for routing methods in warehouse order picking process.

Fig. 3. Routing methods for warehouses.

We would like to identify the best solution that allows the shortest path from the
depot to all of the predefined positions of orders to be picked up and finally back to
the depot. If a distance matrix can be set up using the coordinate system, in which the
depot has position (x=0, y=0), the Solver can be applied to search for the best route.

4 Conclusion

One of the most important issues in research to be conducted is to find better cost
saving measures for transportation. This is due to the fact that there are high costs to
operate and maintain the fleet of vehicles. Furthermore, there is also a concern that
trucks used in solid waste collection have a negative impact on the environment. The
quantity of miles driven, fuel type, engine inefficiency, and exhaust gases emissions
must be consideration. In this paper, we focused on the routing optimization problem
as a classical traveling salesman problem and applied heuristic procedures. Here we
used the Solver which contains a sophisticated, general heuristic procedure known as
the Standard Evolutionary Engine (SEE). This procedure belongs to the class of pro-
cedures known as genetic algorithms, which is non deterministic and relies on random
mutations in its construction of feasible solutions. The example described in this pa-
per is useful for further extensions. This model can also be of significant use to make
short-term operational decisions in various scenarios. Other researchers and business
managers can easily extend their decision models by adapting the proposed model.
Finally it is suggested that the problem of routing order pickers in multiple block
warehouses has similarity and can also be modeled and solved the same way.
988 H. D. Dinh

References
1. Baker, K. and Trietsch, D. Principles of Sequencing and Scheduling, John Wiley & Sons,
New York (2009).
2. Jovicic, N., BRVNRYLF*9XMLF*HWDO³5RXWHRSWLPL]DWLRQWRLQFUHDVHHQHUJ\HIILFLHQ
F\DQGUHGXFHIXHOFRQVXPSWLRQRIFRPPXQDOYHKLFOHV´7KHUPDO6FLHQFHYROSS
67±78, (2010).
3. Beigl, P., Lebersorger, S. and Salhofer, S.: Modelling municipal solid waste generation: a
review. Waste Management. 28(1), 200-214 (2008).
4. De Koster, R., and Van der Poort, E., Routing orderpickers in a warehouse: a comparison
between optimal and heuristic solutions. IIE Transactions, 30, 469-480 (1998).
5. Bel, G. and Costas, A. Do public sector reforms get rusty? Local privatization in Spain.
The Journal of Policy Reform. 9(1), 1-24 (2006).
6. Dijkgraaf, E. and Gradus, R. Efficiency Effects of Privatising Refuse Collection: Be care-
ful and Alternatives present. Tinbergen Institute Discussion Papers 11-156/3, Tinbergen
Institute, Netherlands (2011).
7. Leao, S., Bishop, I. and Evans, D.. Assessing the demand of solid waste disposal in Urban
region by urban dynamics modeling in a GIS environment. Resources, Conservation and
Recycling. 33(4), 289-313 (2001).
8. Dyson, B. and Chang, N.-B. Forecasting municipal solid waste generation in a fast-
growing urban region with system dynamics modeling. Waste Management. 25 (7), 669-
679 (2005).
9. Navarro-Esbrí, J., Diamadopoulos, E. and Ginestar, D. Time series analysis and forecast-
ing techniques for municipal solid waste management. Resources, Conservation and Recy-
cling. 35(3), 201-214 (2002).
10. M. Sharholy, K. Ahmad, R. C. VaishyaDQG5'*XSWD³0XQLFLSDOVROLGZDVWHFKDUDF
WHULVWLFV DQG PDQDJHPHQW LQ $OODKDEDG ,QGLD´ Waste Management, vol. 27, no. 4, pp.
490±496, (2007).
11. 05$ODYL0RJKDGDP10RNKWDUDQLDQG%0RNKWDUDQL³0XQLFLSDOVROLGZDVWHPDQ
agement in Rasht City, IUDQ´ Waste Management, vol. 29, no. 1, pp. 485±489, (2009).
12. 5.+HQU\=<RQJVKHQJDQG'-XQ³0XQLFLSDOVROLGZDVWHPDQDJHPHQWFKDOOHQJHVLQ
developing countries - .HQ\DQ FDVH VWXG\´ Waste Management, vol. 26, no. 1, pp. 92±
100, (2006).
13. I. Monney, ³*KDQD¶VVROLGZDVWHPDQDJHPHQWSUREOHPVWKHFRQWULEXWLQJIDFWRUVDQGWKH
ZD\IRUZDUG´ Modern Ghana, (2014).
14. 6 5 $JKD ³2SWLPL]LQJ 5RXWLQJ RI 0XQLFLSDO 6ROLG :DVWH &ROOHFWLRQ 9HKLFOHV LQ 'HLU
El-Balah - *D]D6WULS´ IUG Journal of Natural Studies, vol. 14, no. 2, pp. 75±89, (2006).
15. 6 'DV DQG % . %KDWWDFKDU\\D ³2SWLPL]DWLRQ RI PXQLFLSDO VROLG ZDVWH FROOHFWLRQ DQG
WUDQVSRUWDWLRQURXWHV´ Waste Management, vol. 43, pp. 9±18, (2015).
16. /+6RQ³2SWLPL]LQJPXQLFLSDOVROLGZDVWHFROOHFWLRQXVLQJFKDRWLFparticle swarm op-
WLPL]DWLRQLQ*,6EDVHGHQYLURQPHQWVDFDVHVWXG\DW'DQDQJFLW\9LHWQDP´ Expert Sys-
tems with Applications, vol. 41, no. 18, pp. 8062±8074, (2014).
17. Clark, R. M. and Gillean -,³$QDO\VLVRI6ROLG:DVWH0DQDJHPHQW2SHUDWLRQVLQ&OHYH
lDQG2KLR$&DVH6WXG\´ Interfaces, vol. 6, no. 1-part-2, pp. 32±42, (1975).
18. Karadimas, N. V., Kolokathi, M., Defteraiou, G., & Loumas, V. Municipal Waste Collec-
tion of Large Items Optimized with Arc GIS Network Analyst. 21st European Conference
on Modelling and Simulation (2007).
Effective Capital Usage Solutions for Road Infrastructure
Maintenance in Vietnam
F(;IM9B.<=.IM9HIB;5B8F(;IM9B+IMB<-5B;

"5BC=F7<=H97HIF5@/B=J9FG=HM 0=9HB5A

/B=J9FG=HMC:.F5BGDCFH5B8CAAIB=75H=CBG 0=9HB5A
khuongdungkt@yahoo.com

Abstract. #BF979BHM95FG H<9FC58=B:F5GHFI7HIF9GMGH9A=B0=9HB5A<5G89J9@

CD98F5D=8@M GIGH5=B56@M =B5AC89FB8=F97H=CBK=H<?9MDC=BHG.<97CAD9H=

H=J9B9GG=B89L:CFH<9FC58EI5@=HM=BH<9D9F=C8C: 
 =GF5B?98F8 5B
=B7F95G9C:@9J9@G7CAD5F98HCH<9D9F=C8C: 
 7<=9J=B;GI7<F9

GI@HG  69G=89G =B7F95G=B; 7CBGHFI7H=CB =BJ9GHA9BH6CH< =B K=8H< 5B8 89DH<  H<9
FC@9C:FC58A5=BH9B5B79=G=B8=GD9BG56@9:H9F  K<9BH<9,C58'5=BH9

B5B79 IB8=GDIH=BHCCD9F5H=CB H<9HC@@GCBACHCF=N98J9<=7@9G5F97C@@97H98

5B8H<975D=H5@5@@C75H98:CFH<9B5H=CB5@<=;<K5MA5=BH9B5B79=BH<9D9F=C8C:
 
 =G 6=@@=CB0(M95FCB5J9F5;9 =H=G;F95H@M=B7F95G97CA

D5F=B;K=H<H<9DF9J=CIGD9F=C8"CK9J9F H<975D=H5@IG5;9A5B5;9A9BH=GGH=@@

=B589EI5H9 M 5B5@MN=B;H<9DC@=7=9G  GH5HIGC: A5B5;9A9BH :CFFC58 A5=BH9

B5B7975D=H5@IG5;9=B0=9HB5AIG=B;H<9F9GI@HGC:FIBB=B;-*--GC:HK5F9:CF
=BJ9GH=;5H=CB5B8GIFJ9M HKC;FCIDGC:GC@IH=CBG<5J9699BDC=BH98HC=ADFCJ9
H<9 A5B5;9A9BH C: H<9 75D=H5@ IG5;9 :CF FC58 A5=BH9B5B79 .KC ;FCIDG 5F9
7CAD@9H=CBH<95@@C75H=CBA97<5B=GA:CF@C75@FC58A5=BH9B5B79:IB85B8@9;5@
GMGH9A F9@5H98 HC 75D=H5@ IG5;9 A5B5;9A9BH HC 9BGIF9 H<9 9::=7=9B7M 5B8 GIG

H5=B56=@=HMC:FC58HF5::=7=B:F5GHFI7HIF9

Keywords:,C58A5=BH9B5B79:IB8 5D=H5@IG5;9A5B5;9A9BH 9BHF5@ IB8 

&C75@ IB8

1 Introduction

9:CF9    H<9 75D=H5@ :CF FC58 A5=BH9B5B79 K5G A5=B@M AC6=@=N98 :FCA H<9 GH5H9
6I8;9H5B8 ) :IB8 K=H<CIH5 DF=J5H9 :IB8 :CF A5=BH9B5B79.<=G 75D=H5@=G AI7<
@CK9FH<5BH<957HI5@89A5B8 FCA 
  H<95J9F5;975D=H5@5@@C75H98:CFH<9
B5H=CB5@<=;<K5MA5=BH9B5B79K5G 6=@@=CB0(M95F A99H=B;
 C:H<9
89A5B834
#AD@9A9BH=B; H<9 DFCJ=G=CBG C: ,C58 .F5::=7 &5K (C  +"  H<9 !CJ

9FBA9BH =GGI98 97F99 (C (

*   CB ,C58 A5=BH9B5B79 :IB8
:H9FH<=G H<9:99:CFFC58A5=BH9B5B79=GD5=86MH<=G:IB8.<9:IB8=G:CFA98:FCA
H<9:C@@CK=B;GCIF79Gcollection of annual road use fees for motorized vehicles; the
State budget allocates annually to the Fund and other sources of revenue related to
road use .<9 IB8 9GH56@=G<985H H<9 79BHF5@
;CJ9FBA9BH @9J9@=G75@@98H<9 9BHF5@
,C58 '5=BH9B5B79 IB8 <9F9=B5:H9F F9:9FF98 HC 5G H<9 9BHF5@ IB8 5B8 H<9 IB8

© Springer Nature Singapore Pte Ltd. 2020 989

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_158
990 N. T. T. Dung and N. Q. Sang

9GH56@=G<985HH<9DFCJ=B7=5@@9J9@=G75@@98H<9&C75@,C58'5=BH9B5B79 IB8<9F9=B

5:H9F75@@98H<9&C75@ IB8.<99BHF5@:IB8=GIG98:CFA5=BH9B5B79C:H<9B5H=CB5@

<=;<K5M GMGH9A &C75@ :IB8G 5F9 IG98 :CF H<9 A5=BH9B5B79 C: @C75@ FC58 GMGH9AG
.<FCI;<H<=G A97<5B=GA  H<9 FC58=B8IGHFM <5G5 B9K588=H=CB5@  G9D5F5H9 5B8=B89

D9B89BH:IB8:CFFC58A5=BH9B5B79 H<IGA5?=B;75D=H5@JC@IA9:CFFC58A5=BH9B5B79
57H=J=H=9G=B7F95G9G=;B=:=75BH@M
.<99LD9BG9F9GI@HGC:FC58A5=BH9B5B79:IB8CJ9FH<9M95FG5F95G:C@@CKG

Table 1. LD9BG9GF9GI@HGC:H<9FC58A5=BH9B5B79:IB8=B6=@@=CB8CB;Sources [2], [6])

No Content Year 2013 Year 2014 Year 2015 Year 2016 Year 2017
# LD9BG9             
LD9BG9G :FCA 9BHF5@
          
 IB8

 ,C58 A5=BH9B5B79 9L

        

D9BG9G

)H<9F9LD9BG9G           
 LD9BG9G:FCA&C75@ IB8G           
## 5FF=98CJ9FHCB9LHM95F          
.<9 9LD9BG9 F5H9 =B7F95G98
        
 CJ9FH<9M95FG

2 Current status of capital usage management for road

maintenance

1=H< H<9 :CFA5H=CB C:5 B9K :IB85B8 B9K F9J9BI9G  H<9 :C@@CK=B; M95FG 75D=H5@ =G
<=;<9FH<5BH<9DF9J=CIGM95F_G 6IHH<975D=H5@IG5;9:CFA5=BH9B5B79GH=@@:579GA5BM
G<CFH7CA=B;G =BK<=7<5F9
•.<98=J=G=CBC:F9J9BI9GCIF79G:FCAFC58IG9:99G=GBCB
GD97=:=75B8=B:@9L=6@9
 77CF8=B;HCF9;I@5H=CBG H<9DFCJ=G=CBGCB8=J=G=CBC:F9J9BI9:FCAFC58IG9:99G
5F9:CF 9BHF5@ :IB8G5B8  :CF&C75@ :IB8G.<9 8=J=G=CB =G65G98 CBH<F99
7F=H9F=5  H<9 @C75@ FC58 @9B;H<   H<9 GH5B85F8 BIA69F C: F9;=GH9F98 75FG =B H<9
DFCJ=B798=::=7I@HM7C9::=7=9BHC:F9J9BI9C:957<DFCJ=B7934 34 "CK9J9F  =:
CB@M 65G98 CB H<9G9 H<F99 7F=H9F=5  H<9 5@@C75H=CB C: 9LD9B8=HIF9 9GH=A5H9G :CF @C75@
:IB8G:579GA5BM8=::=7I@H=9G6975IG9GI7<7F=H9F=58CBCH7CADF9<9BG=J9@MF9:@97HH<9
HF5::=7 G=HI5H=CB  HF5::=7 JC@IA9  A5=BH9B5B79 B998G 5B8 GD97=:=7 7<5F57H9F=GH=7G C:
957<DFCJ=B79F9;5F8=B;HCDC;F5D<M ;9C@C;M <M8FC@C;M K95H<9F 9H7
• )BH<9CH<9F<5B8 H<9@9;5@GMGH9AF9@5H98HCH<9CD9F5H=CBC:H<9A5=BH9B5B79
IB8 <5G BCH 699B GMB7<FCBCIG  IB=:CFA  @958=B; HC H<9 8=::=7I@H=9G =B A5B5;9A9BH
5B8IG9C:75D=H5@ GD97=:=75@@M
 =FGH@M  :FCA $5BI5FM GH     =B 7CAD@=5B79 K=H< H<9 DFCJ=G=CBG C: H<9 &5K
(C +"CB:99G5B87<5F;9G5B8&5K(C +"CB-H5H9I8;9H 
F9J9BI9 :FCA FC58 IG9 :99G KCI@8 69 HF5BG:9FF98 HC H<9 -H5H9 .F95GIFM 5B8 A9F;98
=BHCH<99BHF5@I8;9H34 34.<9GD9B8=B;D@5BG5F958CDH98CB@M6M'=B=GHFMC:
=B5B79BC@CB;9F'=B=GHFMC:.F5BGDCFH =B577CF85B79K=H<H<97IFF9BH-H5H9I8;

Effective Capital Usage Solutions for Road Infrastructure … 991

9HA5B5;9A9BHF9;I@5H=CBG H<IG :IB8G5B88=G6IFG9A9BHIB89FA5BM5;9B7=9G_A5B

5;9A9BH 5F9 BCH H=A9@M =AD@9A9BH98 5B8 75D=H5@ GCIF79 75BBCH DFC57H=J9@M 5FF5B;9
8I9 HCH<97<5F57H9F=GH=7G C: FC58 A5=BH9B5B791<9F95G quarterly 6I8;9H5@@C75H=CB
75IG9G 8=::=7I@H=9G :CF GI6CF8=B5H9 ;CJ9FB5B79 6C8=9G =B 9BGIF=B; H<9 F9;I@5F=HM C:
:IB8G 9GD97=5@@M:CF5GG=;BA9BHC:588=H=CB5@D@5BGGI7<5GCJ9F7CA=B;7CBG9EI9B79G
:FCAH<9F5=B :@CC8 5B8GC@J=B;6@57?GDCHG5B8DCH9BH=5@F=G?GC:HF5::=7
 -97CB8@M 577CF8=B; HC 97F99 (C  (
*    CB 7CBGHFI7H=CB
=BJ9GHA9BH A5B5;9A9BH  H<9 DFC>97H A5B5;9A9BH IB=HG C: 9D5FHA9BH C: .F5BGDCFH
<5J9699BA9F;98CFHF5BG:CFA98=BHCH<9GD97=5@=N98DFC>97HA5B5;9A9BHIB=HGC:H<9
DFCJ=B79#HA95BGH<5H9D5FHA9BHC:.F5BGDCFHBC@CB;9F<5GBCB
6IG=B9GGIB=HG=B
7<5F;9 C: FC58 A5=BH9B5B79  H<=G H5G? =G 7IFF9BH@M 5GG=;B98 HC H<9 DFCJ=B7=5@ .F5::=7
'5B5;9A9BH =J=G=CBG "CK9J9F  7IFF9BH@M  K=H< @=A=H98 @56CF :CF79  H<9 .F5::=7
'5B5;9A9BH =J=G=CBG <5J9 A5BM8=::=7I@H=9G =B7<97?=B;5B8 GID9FJ=G=B; H<9 A5B

5;9A9BHC:75D=H5@IG5;9K<=7<@958GHC@CHGC:@CGG5B8K5GH9
 .<=F8@M H<9F95F9GCA9@9;5@8C7IA9BHGGH=DI@5H9H<9FC58A5=BH9B5B79M9HDFC79

8IF9G5F9BCHIB=:CFA A5?=B;=H7<5@@9B;=B;HC=AD@9A9BH CF9L5AD@9 577CF8=B;HC

97F99  (
*  CBH<9DFC8I7H=CB5B8GIDD@MC:DI6@=7DFC8

I7HG 5B8 G9FJ=79G  FC58 A5=BH9B5B79 =G CB9HMD9C:DI6@=7 ;CC8G 77CF8=B; HC CB

GHFI7H=CB&5K(C  +"5B8,C58.F5::=7&5K FC58A5=BH9B5B79=G69@CB;

=B;HC7CBGHFI7H=CB57H=J=H=9G77CF8=B;@M H<9G9@97H=CBC:A5=BH9B5B797CBHF57HCFG=G

8=::=7I@H8I9HCH<97CA6=B5H=CBC:H<9:CFAC:H9B89F=B;65G98CBH<9=88=B;&5K_G
DFC798IF9G 5B8H9B89F=B;5B8CF89F65G98CB97F99  (
*
 FCA H<9 56CJ9 G<CFH7CA=B;G  H<9 5IH<CFG DFCDCG9 G9J9F5@ GC@IH=CBG HC 7CAD@9H9
H<95@@C75H=CBC:F9J9BI9:FCAFC58IG9:99G5B87CAD@9H=CBH<9@9;5@GMGH9AF9@5H98
HCH<9A5B5;9A9BH5B8IG9C:75D=H5@:CFA5=BH9B5B79

3 Solutions to complete the capital usage management for road

maintenance

3.1 Solution for completion of the fund allocation mechanism to the local Fund
G 5B5@MN98 56CJ9  7IFF9BH@M  H<9 5@@C75H=CB C: F9J9BI9 :FCA FC58 IG9 :99G HC @C75@
:IB8G 65G98 CB H<F99 7F=H9F=5 =G H<9@C75@ FC58 @9B;H<  H<9 GH5B85F8 BIA69F C: F9;=G

H9F9875FG=BH<9DFCJ=B798=::=7I@HM7C9::=7=9BHC:F9J9BI9C:957<DFCJ=B79 =BK<=7<
5F9=FF5H=CB5@ BCB
GD97=:=75B8=B:@9L=6@9
 .<FCI;< H<9 F9G95F7< DFC79GG  H<9 5IH<CF <5G 89J9@CD98  ACF9 7F=H9F=5  B5A9@M
 .F5::=7 :@CK =B H<9 DFCJ=B79  #B F9@5H=CB HC G97IF=HM 5B8 89:9BG9  .CDC

;F5D<=75@ 5B8 <M8FC;9C@C;=75@ 7CB8=H=CBG C: H<9 DFCJ=B79  @=A5H9 5B8 K95H<9F
7CB8=H=CBG C: H<9 DFCJ=B79 #B CF89F HC 9J5@I5H9 K<9H<9F H<9 588=H=CB C: H<9G9 :CIF
7F=H9F=5=GF95GCB56@95B8:95G=6@9CFBCH5B8HC89H9FA=B9H<9=AD57H@9J9@C:5@@G9J9B
7F=H9F=5 CB H<9 5@@C75H=CB C: :IB8G  :CF H<9 5DDFCDF=5H9 5B8 C6>97H=J9 5@@C75H=CB C:
:IB8G H<95IH<CF7CB8I7HG5B=BJ9GH=;5H=CB5GIFJ9M 5@CB;K=H<H<9GIDDCFHC:-*--
GC:HK5F9 
GH9DF9G95F7<DFC79GG5G:C@@CKG
 - Step 1: Design a questionnaire..<FCI;<8CA9GH=75B8:CF9=;BGHI8=9G H<95IH<CF
<5G89J9@CD98H<9.9GH+I9GH=CBB5=F95G:C@@CKG
992 N. T. T. Dung and N. Q. Sang

 Test Questionnaire:
+/-.#)((#,
On the allocation of Central and Local funds

 *@95G9:=@@=BH<95DDFCDF=5H9BIA69F=BH<96CL577CF8=B;HCH<9897F95G=B;89

;F99C:H<9=AD57HGC:H<97F=H9F=5K<9B5@@C75H=B;9LD9BG9GHCH<9@C75@:IB8
*@95G9F5B?H<9=B7F95G=B;89;F99C:H<9=AD57HGBIA69FF9:9FGHCH<9<I;9GH
=AD57H BIA69FF9:9FGHCGA5@@9F=AD57HG
&C75@FC58@9B;H< =::=7I@HM7C9::=7=9BHC:F9J9BI9C:957<DFCJ=B79
 .CDC;F5D<=75@ 5B8 <M8FC;9C@C;=75@ 7CB8=H=CBG C:
-H5B85F8EI5BH=HMC:@C75@F9;=GH9F9875FG
H<9DFCJ=B79
#BF9@5H=CBHCG97IF=HM5B8B5H=CB5@89:9BG9 @=A5H95B8K95H<9F7CB8=H=CBGC:H<9DFCJ=B79
.F5::=7:@CK=BH<9DFCJ=B79 

- Step 2: Experimental research..<9DIFDCG9C:H<=GGH9D=GHC9J5@I5H95B858>IGHH<9

.9GH+I9GH=CBB5=F969:CF97CB8I7H=B;C::=7=5@F9G95F7<.<9.9GHEI9GH=CBB5=F9=GG9BH
HCF9GDCB89BHG=BHKC:CFAG8=F97H5B89A5=@98 5B8H<95IH<CF<5G9LD@5=B98=B89H5=@
H<9=GGI9GH<5HB998HC697CBGI@H98-I6>97HGG9@97H98:CFH<9=BJ9GH=;5H=CBGIFJ9M5F9
H<CG9K<C<5J9KCF?98:CF5@CB;H=A9=BGH5H9A5B5;9A9BH5;9B7=9G F9G95F7<=BGH=

HIH9G 7CBGI@H5B7M7CBHF57HCFGCF7CBGHFI7H=CB7CBHF57HCFGCD9F5H=B;=BH<9:=9@8C:FC58
HF5BGDCFH =B:F5GHFI7HIF9 77CF8=B;HCH<9 BIA69FC: M95FGC: 9LD9F=9B79 
  M95FG
577CIBH98:CF  :FCA
 M95FG577CIBH98:CF CJ9F M95FG577CIBH98:CF
 77CF8=B; HCH<9 KCF?=B;IB=H -H5H9 A5B5;9A9BH5;9B7=9G 5B8F9G95F7< =BGH=

HIH9GCBFC58A5=BH9B5B79577CIBH98:CF 7CBGI@H5B7M7CBHF57HCFG577CIBH98:CF
 7CBGHFI7H=CB7CBHF57HCFG577CIBH98:CF 77CF8=B;HC98I75H=CB@9J9@IB=

J9FG=HM577CIBH98:CF  DCGH
;F58I5H9577CIBH98:CF 
 .<9BIA69FC:EI9GH=CBB5=F9G8=GHF=6IH98=G.<95IH<CFC6H5=B98J5@=8F9

GDCBG9G .<9 -*-- GC:HK5F9 K=@@ 69 IG98 :CF ,9@=56=@=HM -H5H=GH=7G 5B8 7CFF9@5H=CB
GH5H=GH=7G69HK99B957<7F=H9F=CBK=H<5@@7F=H9F=5=BH<9;FCID#H9A
.CH5@-H5H=GH=7GHC
J9F=:MH<9F9@=56=@=HMC:.9GH+I9GH=CBB5=F9
 - Step 3: Official research.CB8I7H=B;5A5GGGIFJ9MIG=B;H<9GIFJ9MA9H<C8
:CFG5AD@9G9@97H=CB =BK<=7<H<9BIA69FC:G5AD@9GHC69GIFJ9M98=G H<95I

H<CFIG9GH<9-*--GC:HK5F9HC5GG9GGH<9=AD57H89;F99C:H<97F=H9F=5H<9<=;<9FH<9
A95BJ5@I9C:7F=H9F=5=G H<9;F95H9FH<9=AD57H89;F99=G34 .<9F9GI@HGG<CK98H<5H
5@@7F=H9F=5A9HH<9F9EI=F9A9BHG:CFF9@=56=@=HM5B87CFF9@5H=CB K=H<FCB657<_G@D<5
=B89L    5B8 CFF97H98 #H9A
.CH5@ CFF9@5H=CB    H H<9 G5A9 H=A9  H<9
F9GI@HGC:A95GIF98=AD57H89;F99C:7F=H9F=55F9DF9G9BH98in the table 2.
Based on Table 2, the author gives the following comments:

@@7F=H9F=5<5J9A95BJ5@I9 GCH<97F=H9F=5<5J9579FH5=B=AD57H89;F99CB
H<95@@C75H=CBC::IB8G"CK9J9F H<97F=H9F=5_GA95BJ5@I9<5G58=::9F9B79H<97F=H9F=5
K=H<A95BJ5@I9K=@@<5J95897=G=J9=AD57H=B7@I8=B;7F=H9F=5(C  =B.56@9
When planning the estimate allocation plan, priority should be paid to these criteria.


Effective Capital Usage Solutions for Road Infrastructure … 993

Table 2. .<9=AD57H89;F99=BH<95@@C75H=CBC:75D=H5@HCH<9@C75@:IB8
Impact degree (%) Se-
"I;
.<9 que-
#A
#A
#A

9GH #A
#A
@95GH nce of
No. Criteria D57H D57H D57H Mean
=A
D57H D57H =A
im-
(C (C (C
D57H (C (C D57H pact
  

(C [(C degree
 &C75@FC58@9B;H<                 
-H5B85F8EI5BH=HM
C:
                 
&C75@F9;=GH9F98
75FG
#BF9@5H=CBHCG97I

 F=HM5B8B5H=CB5@                 

89:9BG9
.F5::=7:@CK=B
               
H<9DFCJ=B79
=::=7I@HM7C9::=

 7=9BHC:F9J9BI9                 

C:957<DFCJ=B79
.CDC;F5D<=75@
5B8<M8FC;9C@C;

                 

=75@7CB8=H=CBGC:
H<9DFCJ=B79
@=A5H95B8
 K95H<9F7CB8=H=CBG                
C:H<9DFCJ=B79

• .<9:CIF7F=H9F=5=BHFC8I7986MH<95IH<CFCF8=B5@(C   =B.56@9<5J95
A95B G7CF9 C:   =B8=75H=B;H<5H H<9M5@@ 5::97HH<95@@C75H=CB C: :IB8G Therefore,
they need to be complemented in the principle of dividing the revenue of road use fees
for the Central Fund and the Local Fund
• GD97=5@@M F=H9F=CB(C"Traffic flow in the province" <5GH<9<=;<9GH89;F99
C:=AD57HA95B This criterion is not currently available in the State's regula-
tion on the allocation of funds to the local Fund, therefore, it should be studied to be
supplemented in the division principle.
• #B H<9 :CIF 588=H=CB5@ DFCDCG98 7F=H9F=5  H<9 7F=H9F=5 (C   75B 69 EI5BH=

:=98.<97F=H9F=5 (C   5F9 EI5@=H5H=J9 <CK9J9F H<9G9 5F9 B979GG5FM 7F=H9F=5HC
697CBG=89F988IF=B;H<95@@C75H=CBC::IB8G

3.2 Solutions for completion of the legal system related to capital usage
management for road maintenance
#B CF89FHC GC@J9 H<9 G<CFH7CA=B;G C: H<9@9;5@ GMGH9A A9BH=CB98=B G97H=CB  H<9
5IH<CF <5G GCA9 F97CAA9B85H=CBG 5G :C@@CKG '=B=GHFM C:.F5BGDCFH G<CI@8 DFCDCG9
HCH<9!CJ9FBA9BHHC=GGI95GD97=5@D=@CHA97<5B=GA:CFH<9FC58A5=BH9B5B79 GD9

7=:=75@@M
994 N. T. T. Dung and N. Q. Sang

• .<9F9G<CI@8695G9D5F5H9A97<5B=GA:CFH<9A5B5;9A9BHC:75D=H5@IG5;9C:
FC58 A5=BH9B5B79  GC H<5H H<9 75D=H5@ :CF A5=BH9B5B79 =G F9;I@5F@M 5B8 7CBH=BICIG@M
A9H:CFH=A9@MF9GC@IH=CBC:GI889BG=HI5H=CBG =B7F95G98H<99::=7=9B7M
• .CD9FA=HH<99GH56@=G<A9BHH<9,C58'5=BH9B5B79'5B5;9A9BH/B=H69@CB;=B;
HCDFCJ=B7=5@9D5FHA9BHGC:.F5BGDCFHHCD9F:CFAH<9A5B5;9A9BH A5=BH9B5B795B8
9LD@C=H5H=CB H<9 9BHFIGH98 B5H=CB5@ FC58G  DFCJ=B7=5@ FC58G IF65B FC58G 5B8 CH<9F 5G

G=;B98 FCIH9G .<=G 9GH56@=G<A9BH =G 7CAD@9H9@M 7CBG=GH9BH K=H< H<9 F95@=HM 5B8 =B

7F95G9G B9=H<9F DI6@=758A=B=GHF5H=CB KCF?:CF79 BCF 6I8;9H9LD9B8=HIF9 6975IG9H<=G

A5B5;9A9BH IB=H =G 5 :=B5B7=5@@M =B89D9B89BH CD9F5H=CB :FCA H<9 FC58 A5=BH9B5B79
DFC>97HGCIF79 IFH<9FACF9 =H=GBCH7CBHF5FMHCH<97IFF9BHF9;I@5H=CBGCBCF;5B=N5

H=CB5@GHFI7HIF9C:9D5FHA9BHC:.F5BGDCFH
• -CA9 @9;5@ 8C7IA9BHG7IFF9BH@MCJ9F@5D  K<=7<@958GHC 8=::=7I@H=9G:CF75D=H5@
A5B5;9A9BH  H<9F9:CF9  H<9 7CAD@9H=CB 5B8 IB=:=75H=CB C: H<9 GMGH9A C: @9;5@ 8C7I

A9BHGG<CI@8697CAD@9H98HCK5F8GH<9F95@=HM5B8=B577CF85B79K=H<H<97<5F57H9F=G

H=7GC:A5=BH9B5B79KCF?5B8FC58A5=BH9B5B79G<CI@869F97C;B=N985G57CBGHFI7

H=CB57H=J=HM

4 Conclusion

5G98CB5B5@MG=G5B85GG9GGA9BHC:H<9F95@=HMC:75D=H5@IG5;9A5B5;9A9BH:CFFC58
A5=BH9B5B79  =B8=75H=CB C: G<CFH7CA=B;G  =B H<=G 5FH=7@9  H<9 5IH<CF 89J9@CD98 HKC
;FCIDG C: GC@IH=CBG HC =ADFCJ9 H<9 9::=7=9B7M C: 75D=H5@ IG5;9 1=H< H<9 GC@IH=CB HC
7CAD@9H9 H<9 :IB8G 5@@C75H=CB A97<5B=GA HC H<9 @C75@ FC58 A5=BH9B5B79 :IB8  H<9
5IH<CF 58898 :CIF ACF9 7F=H9F=5 :CF H<9 577IF5H9 5B8 C6>97H=J9 5@@C75H=CB C: :IB8G
.F5::=7:@CK=BH<9DFCJ=B79#BF9@5H=CBHCG97IF=HM5B889:9BG9.CDC;F5D<=75@5B8
<M8FC;9C@C;=75@ 7CB8=H=CBG C: H<9 DFCJ=B79 @=A5H9 5B8 K95H<9F 7CB8=H=CBG C: H<9
DFCJ=B79HH<9G5A9H=A9 H<95FH=7@95@GCDFCDCG98GC@IH=CBGHC=ADFCJ9H<9@9;5@
GMGH9AF9@5H98HCH<969HH9FA5B5;9A9BH75D=H5@IG5;9C:FC58A5=BH9B5B79

References

'=B=GHFMC: =B5B79 =F7I@5F  ..

.85H98 ;I=8=B;H<9F9;=A9C:A5B

5;9A9BH IG9 D5MA9BH5B8G9HH@9A9BHC:C:FC58GA5=BH9B5B795B8A5B5;9A9BH:IB8G

'=B=GHFMC:.F5BGDCFH  =B5@,9DCFH         
 !CJ9FBA9BH  97F99 (C (
* 85H98   CBH<9 ,C58 '5=BH9B5B79 IB8
97F99(C (
*85H98  5A9B8=B;5B8GIDD@9A9BH=B;5BIA69FC:5FH=7@9G
C: 97F99 (C  (
* 97F99 (C    (
* 85H98     5A9B8=B; 5
BIA69FC:5FH=7@9GC:97F99(C (
*5B897F99(C (
*
 (;IM9B .<= .IM9H IB;  ,9G95F7<CBGC@IH=CBG:CF75D=H5@ IG9 5B8 AC6=@=N5H=CB:CFFC58
A5=BH9B5B79 *<H<9G=G /B=J9FG=HMC:.F5BGDCFH5B8CAAIB=75H=CBG 
(5H=CB5@GG9A6@M  995B8<5F;9&5K(C +" (CJ9A69F  
0=9HB5A ,C58 8A=B=GHF5H=CB  9BHF5@ ,C58'5=BH9B5B79 IB8    =B5@ ,9DCFH    
     
Perspectives of the use of GPS in travel survey: Research
on identification of missing trips in a GPS pilot survey in
Hanoi

NGUYEN Thanh Tu

Faculty of Transport Economics,

University of Transport and Communications, Hanoi, Vietnam
Telephone +84 989-194-794, Fax. +84 243-766-4053
thanhtunguyen.utc@gmail.com

Abstract. GPS-based data collection methods have become particularly popular

in travel behavior research, mainly because of the worldwide coverage and the
accuracy of the GPS system. This paper identifies the missing trip data obtained
by two methods survey reported and GPS recorded in the same days. This study
shows that the GPS survey can be used successfully to complete the conventional
transport surveys, but it is still too early to predict the complete substitution of
conventional survey by the GPS mobility survey.

Keywords: Missing trips, GPS, Travel diary, Logit model, Comparison

1 Introduction

Travel survey methods based on new technologies have evolved in the past few dec-
ades, shifting from limited experiments to large-scale travel surveys. GPS-based data
collection methods have become particularly popular in travel behavior research,
mainly because of the worldwide coverage and the accuracy of the GPS system.
A challenge in the GPS data post-processing is the development of methods to fill
GPS missing data and to reconstitute automatically continuous sequence, both in space
and time. The results obtained from the post-processing software are calibrated by com-
parison with conventional methods, particular a few questions when getting back the
GPS device, providing for a few trips, taken at random, additional information on the
reliability of the device and on more detailed characteristics (mode, purpose and ac-
companying persons).
The primary objective of this paper is to compare survey-reported trips and GPS
recorded trips for the same individuals to identify the reasons for differences between
two sources of data. It could be from the household, individual or travel reasons. Based

© Springer Nature Singapore Pte Ltd. 2020 995

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_159
996 T. T. Nguyen

on these results, we could evaluate the contributions and challenges of travel data col-
lected by GPS in travel survey. Besides, if we could identify the influencing factors on
missing trips, the quality of travel survey data will be improved.
To give full information, first, this paper introduces some statistics about both data
sources. In which, the percentage of the missing trips by different characteristics are
investigated. It means that all the trips recorded by GPS but not reported by respondents
and the reported trips with no GPS trace are concerned in this step. Further, these miss-
ing trips are modeled using a logit model to estimate the variables that affect the differ-
ence between the survey-reported and GPS-recorded trips, without making any a priori
assumptions about the accuracy of the data from each source. This finding could be
useful in further researches in order to improve the quality of travel data collected by
GPS or by travel diary. The data of this study was collected in Hanoi.

2 GPS technology in household travel surveys

In recent years, GPS technology has been increasingly used in travel survey research in
order to evaluate and to improve the travel data reported by diary.
As a main tool to collect the travel data, the comparison between GPS-recorded trips
in comparison with the respondent-reported trips is very important. GPS data show
firstly the rate of missing trips obtained from conventional methods (by CATI, CAPI,
face-to-face or self-administrated). This is a significant shortfall considering that house-
hold travel surveys are used for transport modeling and planning. In the research of [3]
they showed that in a CATI household travel survey in Laredo, Texas only 44 percent
of trips recorded by a GPS were able to be successfully matched with a trip reported by
CATI. By combining GPS and a prompted recall survey, Stopher et al. [5] found that
the rate of non-reported trips was only 7.4% but they ignored all the trips non-recorded
in the GPS for some reasons (forgetting to wear, no satellite signal, etc.).
Besides the comparison between travel data by GPS and diaries, many studies try to
identify the key factors characterizing the under-reported trips. Zmud and Wolf [10]
identified key demographic factors that may contribute to misreporting of trips. Bricka
and Bhat [1] have found that young adults (less than 30 years of age), men, individuals
with less than high school education, unemployed individuals, individuals working in
clerical and manufacturing professions, workers employed at residential, industrial, and
medical land-uses, and individuals in nuclear families are all more likely to under-re-
port trips than other respondents.
In addition to demographics, travel behavior characteristics have influenced on re-
ported missing trips. Wolf [6] claims that short duration trips are often forgotten or
omitted. The travel purposes and the number of trips undertaken on the survey day also
have impacts on reported missing trips. The authors of [1] and [5] found that individuals
with high mobility level tend to under-report their travel. Trips with shorter time dura-
tion and trip distance for visits and picking up or dropping off a passenger were more
likely to be under-reported.
Besides trip under-reporting, the situation where the GPS unit recorded fewer trips
than that reported by respondent, has been studied by some researchers. Wolf et al. [7]
Perspectives of the use of GPS in travel survey … 997

looked for the reasons why the GPS trips were missed in comparison with the trip dia-
ries. Forrest and Pearson [3] found some level of higher reporting from survey diaries
than recorded through GPS units. The authors of [4] also found that another reasons for
the trip not being recorded in the GPS but being reported in the survey data, are the
forgetfulness of respondents and the GPS device problems. Besides, the trips under-
taken at early morning or late evening, are easier to be under-recording by GPS units.
More details, the authors of [2] have examined the differences in measures of intensity
of travel between survey and GPS trips for two trip purposes (work and non-work).
They found that the GPS survey should be considered as the data collection method for
the younger, more technology experiences and have high travel propensities but for the
elderly and more leisurely travelers, the traditional survey method is recommended.

3 Data information and descriptive

The travel surveys were conducted in Hanoi city using two main tools at the same time:
a wearable GPS devices and a travel diary self-reported. The methodology is mentioned
as follows: all participants - who are author's relatives, friends, neighborhoods, and col-
leagues - were asked to wear the GPS on all their trips during one week. They were also
asked to report their trips in the diary at the end of the day. The data collected includes
start time, end time, mode of transport, travel purposes and travel distance. The house-
hold information and personal socioeconomic characteristics were self-reported.
In the travel survey test in Hanoi, we collected data from 95 participants with 1150
matched trips among 1872 reported-trips and 1680 GPS-recorded trips.

Table 1. Number of matched and un-matched trip in France and Vietnam

Number of trip Matched Un-matched Total Missing (%)

By Diary 1150 722 1872 39%
By GPS 1150 530 1680 32%

The missing percentage of a survey reported and GPS recorded trips is 39% and
32%. The distribution of missing trip by the different was divided into 2 groups: (1)
missing trips in travel diary i.e. the trip recorded in GPS but did not report in travel
diary; (2) missing trips in GPS, the trips reported in diary but no trace in GPS.
Some descriptive statistics of missing trips is following:
In terms of demographic, younger people tend to record more trip by GPS than the
elderly. This finding was confirmed by the authors of [2] in their research of factors
influencing differences in survey-reported and GPS-recorded trips.
In terms of mobility, the person with high travel frequency tends to miss more trips
by GPS than those who travel less (e.g. 43% missing GPS trips for people making more
than 6 trips per day, compared to 14% for those making 1-2 trip per day). The trip
distance seems also to have an influence with 50% missing trips under 2km while this
percentage is only 17% over 10km in the survey.
998 T. T. Nguyen

The elders are more likely to have diary missing trips, people who is more mobile
(from 30 ± 50 years old), has more missing trips. When people are more mobile and
make longer trips, rather for work or education than for shopping or personal purpose,
the number of the missing trips is lower. In this paper, we forgot the factor of speed
when studying the GPS trips, but the study of Pham [10] found that this unmatched rate
is the greatest (77%) for speeds below 7km/h which correspond to "Walk´. The reasons
would be the omission of important events by the respondent or a wrong imputation of
the mode of transport by GPS post-processing. This is the same result when we compare
with diary missing trip, the respondent would forget the non-motorized trips (the miss-
ing rate is 60% by bike and more 40% on foot in both surveys).

4 Factors influencing missing trips

To identify the factors influencing missing trip, a logit binary model was used, in which
the dependent variable is the possibility of missing trip (i.e. Yes - No), include:
the GPS missing trips and (ii) diary missing trips. The independent variables are
demographic and travel behavior of the respondents. 3 types of trips were identified.

4.1 Identification of GPS missing trips

The GPS missing trips are trips reported in the travel diary but have not been detected
by GPS devices. We previously counted 1872 trips in the travel diary, in which 722
trips missing data in GPS. To better understanding the factors influencing GPS missing
trips, we used a binary logistic regression with the dependent variable is GPS missing
trip (yes/no). The independent variables are characteristics of households, personal in-
formation and travel behavior. Table 2 shows the factors influencing GPS missing trips.
Trip distance, number of trips per day, mode of transport and starting time were signif-
icant variables.

Table 2. Personal characteristics and travel behavior in 2 surveys

Variables DF Wald Khi-2 Pr > Chi-2

Trip distance 3 107.60 <.0001
Nb. of trips per day 3 11.26 0.0104
Mode of transport 6 39.62 <.0001
Starting time 29.19 <.0001

More detail, trip distance has an important impact on GPS missing trips: the percent-
age of missing trips under 2km is 4.372 times higher than for trips over 10km. The
second factor is the trip frequency: the more trips are undertaken per day, the higher the
GPS missing trips is. In terms of mode of transport, we found that the that the GPS
devices would record well the motorcycle trips but tend to omit the walk or bus trips.
Perspectives of the use of GPS in travel survey … 999

4.2 Identification of reported missing trips

The reported missing trips are those detected by GPS but were not reported in the travel
diary. As shown in table, 530 trips omitted among 1680 GPS recorded trips. For a better
understanding the reason for these trips, we estimate a binary logistic regression on
each survey.

Table 3: Factors influencing reported missing trips

Variables DF Wald Khi-2 Pr > Chi-2

Number of GPS trips 3 41.69 <.0001
Duration of GPS trips 1 7.83 0.0051
Distance of GPS trips 3 33.38 <.0001
Starting time 5 20.92 0.0008

Among the influencing factors of missing trips in the diary, the number of trip is the
most important factor. Individuals who make frequent trips tend to have more reported
missing trips: the ratio of reported missing trips is 10 times lower for a day with 1-2
trips than for a day with 7 trips or more. Travel duration is also one of the factors: the
rate of reported missing for trips under 5 minutes is 11.918 times higher than for trips
over than 60 minutes. The personal occupation has only a light impact: employees and
retired persons have more reported missing trips. Besides, the start time could have a
light influence: trips started early in the morning are more usually reported.

4.3 Comparison between GPS recorded trip and survey reported trips

After identifying factors influencing the missing trips, we compare the matched-trips
i.e. those were recorded by GPS and also reported in the travel diary. We calculate the
difference between the trip characteristics recorded by two ways and we found that:
The departure time reported in the diary is earlier than that recorded by GPS for about
70% of trips. In which, the difference exceeds 10 minutes is for 37%.
The difference for ending time is lower than for departure time, in which 28% of the
difference in ending time between these surveys is more than 10 minutes. The smaller
difference for ending time than for departure time may be due to the Time necessary to
first fixe (TTFF) that is a measure of the time required for a GPS receiver to acquire
the satellite.
People tend to overestimate the travel duration, i.e. the reported duration is longer
than GPS recorded, the reported travel duration is over-estimated for about 80% of
trips.
The travel distance reported in the diary is usually greater than travel distance rec-
orded by GPS but the difference is usually less than 5 km (87% of trips).
1000 T. T. Nguyen

5 Conclusion

The paper aimed at two main objectives. First, it examined the influencing factors for
the missing trips by using a logistic regression. Secondly, it analyzed the differences in
the characteristics of travel for matched trips. The findings from this research can in-
form both survey-reported as well as GPS-recorded travel data collection approaches.
Overall, there is a large difference in the measurement of daily mobility between
GPS recorded and survey reported trips. These results suggest that improvements are
needed in both methods to yield more accurate and un-biased data. So we recommend
that the GPS-only survey should be conducted with caution. Future research could in-
volve the development of more complex choice models to explain other margins of
transport choice. Furthermore, the travel survey by GPS only should strongly consider
to develop the algorithms to impute trip purpose. In addition, the GPS seems to be a
method dealing the younger, more technology savvy individuals as well as those that
have high travel propensities or characteristics associated with trip chaining, in order
to ensure that all trip details are recorded. However, for the elderly and more leisurely
travelers, the traditional survey method is recommended.

References
1. Bricka, Stacey, and Chandra Bhat. Comparative Analysis of Global Positioning System-
Based and Travel Survey-Based Data. Transportation Research Record: Journal of the
Transportation Research Board 1972 (1): 9±20 (2006)
2. Bricka, Stacey G., Sudeshna Sen, Rajesh Paleti, and Chandra R. Bhat. An Analysis of the
Factors Influencing Differences in Survey-Reported and GPS-Recorded Trips. Transporta-
tion Research Part C: Emerging Technologies 21 (1): 67±88 (2012).
3. Forrest, T., and D. Pearson. Comparison of Trip Determination Methods in House- hold
Travel Surveys Enhanced by a Global Positioning System. Transportation Research Record
1917 (1): 63±71 (2005).
4. Stopher, P., and L. Shen. An In-Depth Comparison of GPS and Diary Records. In
https://trid.trb.org/view/1148676 (2011).
5. Stopher, Peter, Camden FitzGerald, and Min Xu. Assessing the Accuracy of the Sydney
Household Travel Survey with GPS. Transportation 34 (6): 723±41 (2007).
6. Wolf, J. Using GPS Data Loggers to Replace Travel Diaries in the Collection of Travel Data.
Dissertation, Georgia Institute of Technology, School of Civil and Environmental, 58-65
(2000).
7. Wolf, J., M. Loechl, M. Thompson, and C. Arce. Trip Rate Analysis in GPS- Enhanced
Personal Travel Surveys. In Transport Survey Quality and Innovation, 28:483±98. Oxfordௗ
Elsevier (2003).
8. Wolf, J. Applications of New Technologies in Travel Surveys. In Travel Survey Methods,
531±44. Emerald Group Publishing Limited (2006).
9. Zmud, J., and J. Wolf. Identifying the Correlates of Trip Misreporting ± Results from the
California Statewide Household Travel Survey GPS Study., Lucerne (2003).
10. 3KDP 7KL +XRQJ 7KDR $SSRUWV HW GLIILFXOWpV G¶XQH FROOHFWH GH GRQQpHV j O¶DLGH GH Up
cepteurs GPS pour réaliser une enquête sur la mobilité. Doctorat Dissertation. Paris-Est Uni-
versity. (2016).
 Satelite image application for assessing the effect of
urbanisation to temperature at Danang City, Viet Nam

Ngoc Duong Vo1, Quang Binh Nguyen1

1
The University of Danang ± University of Science and Technology

vo.ngoc.duong@dut.udn.vn

Abstract. The infrastructure development for economy is really imperative to-

wards developping countries. Therefore, this process brings several unexpected
impacts. Landcover change caused temperature increase is noticeble thing. This
impact along with global warming bring out significant consequences about the
health, the economy and the society. For the purpose of analysis the effect of
urbanisation on temperature variation, this study processes the Landsat satellite
image over forty years. The result is expected to find out the relation between the
urbanisation process and city warming. It might be the crucial basic for planning
the urban area more effective and more sustainable, as well for mitigating the
negative impact on the public health. In this paper, the authors will present sev-
eral results of the temperature variation due to urbanisation over the pats forty
years.

Keywords: Urbanisation; Landcover change; Temperature variation; Landsat

satellite image.

1 Introduction

Economic growth and urbanization are developing rapidly in the world, especially in
developing countries. Cities, special zones, economic centers are planned and built in
many places to meet the infrastructure and human resources for economic development.
In these places, transportation infrastructure, factories, houses, etc. are built a lot and
population density is also higher than in other places. This rapid and local change has
affected regional climate conditions, urban environment and socio-economic develop-
ment [1]. Land-use changes on surface temperature is the most significant impact of
urbanization process. Temperature is an important component in surface energy bal-
ance that helps regulate the boundary layer air temperature [2]. The cause is attributed
to the change in the area of surface cover, forest area, water surface and concrete re-
placement. Combined with economic activities, transportation, energy sources from
high-rise buildings make climate conditions, especially the temperature become more
severe. Therefore, studying the relationship between land cover changes and surface
temperatures will be necessary to help us better understand the impact of urbanization,
helping to develop appropriate strategies for management. sustainable urban areas.

© Springer Nature Singapore Pte Ltd. 2020 1001

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_160
1002 N. D. Vo and Q. B. Nguyen

Currently the study of the main temperature distribution mainly based on observed sta-
tion data [3]. However, the analytical results are still sporadic and only reflected in
some locations. With the development of high-resolution remote sensing imaging tech-
nology in space, detailing for many areas, surface temperature data of remote sensing
data such as Landsat Thematic Mapper (LTM) has been collected and analyzed since
1972 up to now. Based on this, remote sensing technology combined with Arcgis tool
will be used to analyze and map spatial temperature distribution. With the advantages
of this technique, so far, many authors have applied to assess the impact of the process
of urbanization on surface temperature in many different areas. Ahmed et al. [4] used
remote sensing imagery in combination with Arcgis tool to assess the impact of urban
temperature and the impact of climate change on economic activity of Suez Gover-
norate, Egypt. Hokao and Phonekeo [5] used Landsat images from 1994, 2000 and 2009
to evaluate the temperature characteristics of urban areas in Bangkok, Thailand by in-
vestigating the relationship between soil surface temperature (LST) and Normalized
Difference Vegetation Index (NDVI). Kumar et al. [6] investigated and identified the
type of land use that had the greatest impact on the increase in temperature around the
city of Vijayawada, India by remote sensing techniques. With the aim of provide some
crucial basics for planning the urban area more effective and more sustainable, as well
for mitigating the negative impact of temperature change due to urbanisation on the
public health, economy development, the study is launched to estimate the impact of
land cover change on temperature in past years in Danang city.

2 Methodology

The research method is based on the calculation of changing the temperature from the
grayscale on the thermal channel of Landsat satellite image. The process begins from
FRQYHUWLQJ'LJLWDO1XPEHU'1WRWKHYDOXHRIVSHFWUDOUDGLDWLRQ/ȜWKHQFRQYHUWLQJ
this radiation value to temperature value. This conversion is carried out according to
the expression of the United States Geological Survey (U.S. Geological Survey -
USGS):

LȜ= ((LMAX±LMIN) /(QCALMAX±QCALMIN))*(QCAL ± QCALMIN) + LMIN

With:
QCAL = Radiated values are calibrated and quantitative properties are in integer form
LMINs và LMAXs are spectral radiation values in integer form
The conversion of spectral radiation to temperature is carried out by the formula of
Planck:

7 .OQ./Ȝ

With: T = Effective temperature on satellites (K);

K1, K2: Correction factor
Satelite image application for assessing … 1003

2.1 Study Area

Danang City has a natural area of 128.342 km2, including the mainland and archipelago
in the East Sea. The mainland is located at 15055' to 16014' North latitude, 107018' to
108020' East longitude, the North borders Thua Thien - Hue province, the West and
South are Quangnam province, the East is adjacent to the East Sea. Located in the cen-
tral region of the country, on the North - South transport axis of road, rail, sea and air,
764km from Hanoi to the North, 964 km from Ho Chi Minh City. South connects the
Central Highlands region with National Highway 14B. The business and commercial
centers of Southeast Asia and Pacific countries are within a radius of 2000km from
Danang city.
Danang belongs to a tropical monsoon climate with high temperature. Each year has
two distinct seasons: dry season from January to September, rainy season from October
to December. The average annual temperature is about 25oC, the highest is in June, July
and August, averaging from 28oC - 30oC, the lowest in December, January and Febru-
ary averaging from 18 oC - 23oC.

Fig. 1. Area study

2.2 Material
To assess the more detailed effect, four remote sensing images correspond to the 4-year
period 2001, 2005, 2010 and 2017 (Table 1)

Table 1. Landsat image data

Spacecraft Sensor Date acquired Scene center time Grid cell size
Landsat 5 TM 07/4/2001 02:52:33 30x30m
Landsat 5 TM 20/05/2005 02:59:58 30x30m
Landsat 5 TM 05/07/2010 03:03:11 30x30m
Landsat 8 OLI 01/7/2017 03:06:14 30x30m
1004 N. D. Vo and Q. B. Nguyen

3 Results and discussion

The landsat satellite image is used for finding out the temperature change tendency in
Danang City. The tendency of temperature is then compared with the change tendency
of landcover in different periods. The comparison will bring out the relation between
the landcover change and temperature increasing.

Fig. 2. Land used in Danang city in different periods; (a) 2001; (b) 2005; (c) 2010; (d) 2017

Fig. 3. Temperature map of Danang city in different periods

From the analysis results, the forest area changes gradually over the years in both area
and percentage. Specifically, in 2001, 508182m2 (accounting for 46.92%), by 2017,
Satelite image application for assessing … 1005

455100 m2 (accounting for 42.02%). This trend is similar to the surface area of water.
In contrast, the area of the land cover is an increase in urban area and percentage. In
2001, 85822m2 (accounting for 7.92%), by 2017, 107685 m2 (accounting for 9.94%).
Analysing the Landsat satellite data, the preliminary result demonstrates that the
'DQDQJ¶VPHDQWHPSHUDWXUHKDVYDULHGVLJQLILFDQWO\LQSDVW\HDUV
The results of temperature analysis based on space over 4 years are shown in Fig. 3.
From the map, high temperature is mainly concentrated in areas with high levels of
urbanization, less trees and water surface. Detailed results in urban areas are shown in
Fig. 4 and 5.

Fig. 4. . Temperature at Danang city in different periods

The result showes that the temperature in Danang city changes from 28.77 0C to 32.240C
over seventeen years (2001 ± 2017). As shown from Fig. 3,4,5 the temperature varies
greatly according to the space between years, between urban areas, rivers and forests.
Specifically, the temperature in 2001 was less than 3 years in 2005, 2010 and 2017 both
in terms of amplitude and difference between regions. The average temperature in 2001
is about 22.470C, the difference between urban and forest areas is only 6.050C. In con-
trast, the distribution temperature of 3 years 2005, 2010 and 2017 is quite similar, the
average temperature is about 26.830C, 26.440C, 26.450C, respectively. In these 3 years,
the influence of urbanization has shown clearly, the temperature is higher in urban areas
and the difference between regions is also greater. The temperature difference between
urban and forest areas in 3 years is 7.4 0C, 9.180C và 9.570C, respectively.
1006 N. D. Vo and Q. B. Nguyen

Fig. 5. Temperature change in Danang center.

4 Conclusions

The result showes that the temperature in Danang city changes from 28.770C to 32.240C
over seventeen years (2001 ± 2017). The analysis also demonstrates that there is a re-
lation between the urbanisation process and city warming. It might be the crucial basic
for planning the urban area more effective and more sustainable, as well for mitigating
the negative impact of temperature change due to urbanisation on the public health,
economy development. Futhermore, the paper is hoped to present a new approach in
applying satelitte image for evaluation the impact of urbanisation on temperature
change for lack of observed data.

References
1. H. E. Landsberg, The urban climate, vol. 28. Academic press, 1981.
2. (.DOQD\DQG0&DL³,PSDFWRIXUEDQL]DWLRQDQGODQG-XVHFKDQJHRQFOLPDWH´Nature,
vol. 423, no. 6939, pp. 528±531, 2003.
3. 20$NLQERGH$2(OXGR\LQDQG2$)DVKDH³7HPSHUDWXUH and relative humidity
distributions in a medium-VL]H DGPLQLVWUDWLYH WRZQ LQ VRXWKZHVW 1LJHULD´ J. Environ.
Manage., vol. 87, no. 1, pp. 95±105, 2008.
4. 6 $KPHG ³$VVHVVPHQW RI XUEDQ KHDW LVODQGV DQG LPSDFW RI FOLPDWH FKDQJH RQ
socioeconomic over Suez GoverQRUDWHXVLQJUHPRWHVHQVLQJDQG*,6WHFKQLTXHV´Egypt. J.
Remote Sens. Sp. Sci., 2017.
5. . +RNDR DQG 9 3KRQHNHR ³$VVHVVLQJ WKH LPSDFW RI XUEDQL]DWLRQ RQ XUEDQ WKHUPDO
HQYLURQPHQW$FDVHVWXG\RI%DQJNRN0HWURSROLWDQ´Int. J. Appl., vol. 2, no. 7, 2012.
6. .6.XPDU38%KDVNDUDQG.3DGPDNXPDUL³(VWLPDWLRQRIODQGVXUIDFHWHPSHUDWXUH
WRVWXG\XUEDQKHDWLVODQGHIIHFWXVLQJ/$1'6$7(70LPDJH´Int. J. Eng. Sci. Technol.,
vol. 4, no. 2, pp. 771±778, 2012.
 Dambreak Simulation at Thuan Ninh Reservoir,
Binhdinh, Vietnam.

Ngoc Duong Vo1, Quoc Hoang Anh Huynh 2 and Trung Dung Vo3
1The University of Danang ± University of Science and Technology
2
BinhDinh Irrigation Works Operating Ltd Company.
3
Department of Agricultural and Rural development, Bin Dinh Province.

vo.ngoc.duong@dut.udn.vn

Abstract. With more than 160 reservoirs, Binhdinh is considered as the province
having the most reservoirs over the country. This system have contributed im-
portantly for provincal socio economic development. Therefore, the reservoirs in
Binhdinh comprehend many potential risks due to their ages, noticeably the
Thuan Ninh reservoir (Vhi= 32.26 106m3) which was contructed for 1970s. In
order to help the local authority to respond actively with urgent incident, the study
is to simulate dambreak incident for Thuan Ninh reservoir. The study is realizes
relied on MIKE FLOOD model. The modelling scenario is supposed following
the standard of TCKT 03:2015/TCTL. The simulation demonstrates that the
Thuan Ninh dambreak incident might affect 33.35 km2 at downstream area. The
lag time, inundated time and flood depth are also considerd in the study.

Keywords: Dambreak, reservoir, Mike flood, Thuan Ninh, flood disaster.

1 Introduction

In recent years, the weather condition has unusual occurrences, floods occur more fre-
quently and serevely [1]. The phenomena of erosion, landslides are complicated, espe-
cially in the hilly and mountainous areas; the lakes conserve water resource which is
dangerous for dam safety. In Vietnam, there are also phenomena of dyke breakage and
dam failure due to many different reasons such as the phenomenon of valve of Song
Bung 2 hydropower reservoir in September 2016, causing serious consequences, dam
failure at Dakrong 3 hydropower in October 12/2012, swept away tons of newly har-
vested crops, causing great damage to the residents and the dam break at Ia Krel 2
hydropower in August 1/2014, killing two people, breaking Khe Mo earth dam in Ha
Tinh province in the morning of October 16/2010 caused great damage to people and
facilities.
With more than 160 reservoirs in the area, Binhdinh is one of the localities, which
possesses the largest number of reservoirs in the country. This system has been contrib-
uting positively to the local socio-economic development. However, this reservoir sys-
tem after exploitation time is now seriously degraded. Moreover, similar to other cen-
tral provinces in the region, Binhdinhis located in the tropical monsoon climate, the

© Springer Nature Singapore Pte Ltd. 2020 1007

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_161
1008 N. D. Vo et al.

average annual rainfall is large, and ranging from 1,750 to 2,400 mm. Rainfall in the
area is unevenly distributed over time, concentrating from September to December. At
the same time, this area has relatively complicated terrain. Characteristic for the terrain
of the coastal plain of Central - Central Vietnam, the terrain of the area is short and
sloping, the West is bounded by high mountains, and the East is the narrow coastal
plain. The terrain affects short and steep rivers, increasing the flow rate, leading to rapid
time of concentration. With the hydrological climate and degradation status, the reser-
voir system in Binhdinhprovince has certain risks of dam safety, directly threatening
people who live in the downstream area of the dam. In order to proactively deal with
natural disasters caused by dam failure, the study of the construction of flooding maps
for reservoirs in Binhdinh in the event of a dam failure is an essential requirement,
especially for reservoirs with high age.
Thuan Ninh reservoir was built in the 1970s, which is one of the large reservoirs
with high age in Binhdinh. After nearly 50 years of exploitation, along with unusual
weather conditions that leading to Thuan Ninh reservoir now faces the high risk of
unsafe dams. In order to proactively respond to unusual conditions, unexpected circum-
stances and bad situations may need to be expected and build a detailed plan to prevent
bad situations from occurrence and maximize damage when incidents occur in both
construction sites and downstream, this study shall build flood maps for downstream
area of Thuan Ninh reservoir in the situation of dam failure. The research results hoped
to provide local authorities and disaster management agencies in the area with the nec-
essary information to help proactively deal with and mitigate losses due to dam breach
situations of the Thuan Ninh reservoir.

2 Methodology

MIKE software family was built by Danish Hydraulic Institute (DHI) [2], [3]. With
many powerful tools integrated, the MIKE software family has been used to effectively
solve many basic problems in the field of water resources.

Fig. 1. Sketch the 1-D and 2-D flow patterns in the river.
Dambreak Simulation at Thuan Ninh Reservoir … 1009

In the MIKE software family, MIKE FLOOD is built with the main purpose of sim-
ulating the flooding process in different cases, especially due to river flow, urban water
drainage problems, or other incident work. MIKE FLOOD is a combination of MIKE
21 model - simulating two dimensional flows in coastal area and MIKE 11 model -
simulating one dimensional flow in riverbed as well as public structures on the river
(Fig. 1). In MIKE FLOOD, MIKE 21 and MIKE 11 models connect via lateral links or
standard links.
In this study, MIKE FLOOD model was selected to show in practice the process of
flood inundation due to dam failure. In particular, the mechanism of dam failure as well
as flow in the river is simulated through MIKE 11 model. The downstream flood area
is shown by model MIKE 21.

3 Study area.

Thuan Ninh reservoir is an artificial reservoir located in Binhdinhprovince - a coastal

area of Vietnam. With a length of 492m and height of 28.7m, this structure can hold up
to 35.36x106 m3 of water. It provides irrigation water for 2,700 hectares of rice and
vegetables / year; It was built from the 70s of the twentieth century, with the earth dam
structure, Thuan Ninh dam is now degraded after nearly 40 years of exploitation. There-
fore, the current security reservoir has a high risk of dam safety. The study of dam
failure options and the introduction of mitigation measures is essential and urgent.

4 Model set up and computational scenario

4.1 Model set up

Fig. 2. Diagram of setting up MIKE FLOOD model.

1010 N. D. Vo et al.

As shown in section 2.1, in order to demonstrate the measure of impact of a dam breach
in the most realistic way with practical region, a MIKEFLOOD model has been estab-
lished based on a combination of one-dimensional model MIKE 11 and two-dimen-
sional model MIKE 21 FM (Fig. 2). The connection between these two sections is
shown in the 73 left-side links and 74 right-side links.

MIKE 11 model.
The calculation area consists of a main river from the upstream of Thuan Ninh reservoir
(chainage 00) to the entrance to Kone River (chainage 21000m) at Binh Nghi hydro-
logical station. From the chainage 00 to the chainage 5400 (26 cross-sections): the lap
of Thuan Ninh reservoir, the cross-section data taken from the reservoir map is provided
by BinhdinhIrrigation Works Exploiting Company Limited. From the chainage 5400 to
chainage 21000 (68 cross-sections): River cross-section data provided by BinhdinhIr-
rigation Works Exploiting Company Limited and inherited from the built projects in
the area (Figure 3). Boundary conditions: Upstream boundary: The upstream boundary
is the process of flow (Q ~ t) measured at Thuan Ninh reservoir. Downstream boundary:
The downstream boundary is the observed water level measured at Binh Nghi hydro-
logical station (H ~ t). Structure and dam break parameters are declared through module
Structure in the Network part.

MIKE 21 FM model.
The MIKE 21 FM model is built on the foundation that is a Mesh built with the follow-
LQJEDVLFSDUDPHWHUV1RGHV(OHPHQWV6PDOOHVWDQJOHÛ(OHYDWLRQ
parameters are used from a digital elevation map with 25mx25m resolution that inherit
the inundated by super storm project of Binhdinhprovince.

4.2 Computational scenario

In the flood simulation of downstream reservoirs due to dam failure, the problem of
analyzing the selection of break is very important. The mechanism of forming the break
will leave greatly effect on the downstream flooding.

Table 1. Dam breach parameters of overtopping

Initial breach Final breach Side Time

Zi- slope to fail- Failure
Zfinal Bfi-
nitial Binitial(m) of ure mode
(m) nal(m)
(m) breach (hours)
71.2 1 42.44 300 1 2 Overtopping

In this study, the mechanism of breakage formation is built on the basis of technical
standards TCKT 03:2015/TCTL Irrigation Constructions - Guidelines for inundation
mapping in cases of emergency flood release and dam failure. The break development
process for Thuan Ninh is shown in Table 1.
Dambreak Simulation at Thuan Ninh Reservoir … 1011

5 Results and conclusions

The simulation results of flooding due to dam failure are shown in Fig. 3 and Fig. 4.
Accordingly, in the 2016 flood event, if the dam failure occurs, it will cause a corre-
sponding flooding area of 6.823 hectares. The maximum flood water level is shown in
Fig. 4, whereby the maximum flood depth occurs at kilometres 17 downstream of the
dam, with the maximum flood water level estimated at 8.7 m. The flood depth is express
in Table 2.

Fig. 3. Graph of water level in the river basin in the case of simulating the December 2016
flood event (Scenario 1) before dam failure.

Fig. 4. Flood depth map of dam failure simulation with actual flood event in 2016 overtopping
type

Besides, the simulation also determined the time of flood transmission from the foot
of the dam to the downstream areas as well as at the end of Queo stream, with approx-
imately 17 minutes duration. With this result, the damage is shown in Table 3. This
1012 N. D. Vo et al.

result is expected to provide useful information to help the disaster management agen-
cies as well as residents in the area minimize the impact caused by the collapse of Thuan
Ninh dam.

Table 2. Statistics of flooded area according to innundated level

Flood depth (m) <=0.5 0.5-1 1-2 2-5 >5

Area ( Hectare) 1750 627.94 2852.94 1542.56 50.44

Table 3. Statistics of damage according to each type of land use.

No Land cover Area ( Hectare)

1 Rice field 1818.25
2 Annual crops 228.56
3 Perennial crops 193.69
5 Water surface 68.00
6 Natural forest 28.75
7 Plantation forest 106.31
8 Residential land 555.56
10 Unused land 324.75

6 Conclusion

The assessment of dam safety as well as the construction of scenarios to deal with dam
failure is an urgent requirement, especially, in the context of irregular weather happen-
ing regularly, the flow discharge to the lake change from the original design. In order
to assess the impact of the dam failure incident on the dam, through MIKE FLOOD
model, the study simulated the dam breach dam process in Thuan Ninh reservoir, one
of the large capacity reservoirs was build and exploit more than 40 years in Binh Dinh.
The study has shown that the failure of the Thuan Ninh reservoir dam will seriously
affect the downstream area, causing a devastating area of 6,823 hertares. The study also
showed that the time of flood transmission in the incident of breaking the Thuan Ninh
reservoir is about 17 minutes. This result is expected to contribute an important part in
timely response and mitigation of flood damage caused to the downstream area of
Thuan Ninh reservoir in the mentioned above situations.

References
1. & & 0R15( ³6HD /HYHO 5LVH 6FHQDULRV IRU 9LHWQDP´ 0LQLVW 1DW 5HVRXU (QYLURQ
Hanoi, Vietnam, 2009.
2. '+,0,.(=(528VHU¶V0DQXDO
3. -/DQGUHLQ³,QWURGXFWLRQWR0,.()/22'´LQ+\GURHXURSH
 Flood Mapping For Downstream of Ben Hai River,
Vietnam.

Ngoc Duong Vo1and Truong Son Ho2

1
The University of Danang ± University of Science and Technology
2
The Quang Tri Division of Hydraulic and natural disaster prevention and control.
vo.ngoc.duong@dut.udn.vn

Abstract. Flood map nowadays is seen as an indispensable tool in urbanism,

flood prevention and mitigation. For this reason, establishing flood map is mighty
necessary for developing the socio economy of a river catchment. Ben Hai is the
second largest river at Quang Tri province ± Viet Nam central region. Besides
the positive role towards the local socio ecenomic development, the Ben Hai river
in resent years has caused lots of disasters, specially on inundation. With the aim
of providing an overview for mitigating the natural disaster consequences, the
study is used the MIKE FLOOD-DHI to repersent the Ben Hai river flooding.
The model is set up for an area up to 100 km2. The flood maps are established
with five historical flood events and several assumptions. The result is expected
to provide the basic secientific evident for local authority to redude the impact of
natural disaster, also for sustainable development.

Keywords: River flooding, Natural disaster, Vietnam central.

1 Introduction

Floods have always been a leading threat and have caused great damage to people and
material possessions [1]. Especially in the context of unpredictable climate change, the
impact of natural disasters and floods on socio-economic development is getting more
and more intense. Therefore, in order to develop sustainably and have a plan to adapt
actively to climate change, besides the socio-economic development objectives, the
prevention of natural disasters, especially floods is also very important.
The central region in general and Quang Tri province in particular have a tropical
monsoon climate with a large average annual rainfall ranging from 1,750mm -
2,400mm. However, rainfall is unevenly distributed over time, concentrating from Sep-
tember to December. With characteristics mentioned above, Quang Tri is one of the
provinces that often suffer from various types of natural disasters with a high frequency
and intensity compared to nationwide. In particular, most are storms, floods, inunda-
tion, drought, saline intrusion, tornadoes and landslides. Natural disasters have caused
damage to people's lives, social properties and people. Due to the serious nature of
flooding in the river basins of Quang Tri province in general and the Ben Hai river
basin in particular, at the same time, flood prevention plan for Ben Hai river basin has

© Springer Nature Singapore Pte Ltd. 2020 1013

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_162
1014 N. D. Vo and T. S. Ho

not been built up so it is necessary to build a scientific and practical basis to devise
solutions to cope and rescue people in the rainy season. For the above reason, this study
was conducted to provide local authorities and disaster management agencies in the
province the necessary information to actively respond to and mitigate the damage
caused by the flood at Ben Hai River downstream.

2 Study domain

As the second largest river of Quang Tri province, behind Thach Han River, Ben Hai
River originates from the Dong Chau area with a height of 1,257 m and a length of 65
km. The average annual flow is 43.4 m3/s. Ben Hai River has 5 level I tributaries (di-
rectly discharging into the main river) and 06 tributaries of grade II (directly discharg-
ing into level I tributaries), the relatively large level I tributaries are Khe Khi, Khe Muoi
and Sa Lung, etc. The total catchment area is 809 km2, with the two main tributaries
are Sa Lung and Ben Hai, the average height of the basin is 115m, the average slope of
the basin is 8.60%. The river flows into the sea at Cua Tung. The flow of Ben Hai River
plays an important role in socio-economic activities of Vinh Linh, Gio Linh districts in
particular and Quang Tri province in general [2].

3 Methodology

Flood mapping for downstream area of Ben Hai River is proposed with 2 principal
steps:

x First step is hydrological modelling. In order to represent accurately the flood situa-
tion in downstream of Ben Hai River, a fully distributed hydrological model ± Mike
She has been constructed for whole catchment of Quang Tri province.
x The second step is for flood mapping. Due to catchment characteristics, the inunda-
tion frequently attacks at the downstream part of Ben Hai river system. Accordingly,
hydraulic models are only set up for downstream area of Ben Hai which is around
55.1 km2 at downstream on 809 km2 in total. The hydraulic model chosen for this
study is MIKE FLOOD which is also a product of DHI and developed on the cou-
pling between 1D model to 2D model.

Fig. 1. Proposed methodology for flood mapping in downstream of Ben Hai River [3].
Flood Mapping For Downstream of Ben Hai River … 1015

3.1 Hydrological model set up

For expressing the most detail hydrological process of Ben Hai catchment, the Mike
She model is established for 4.593 km2 of Quang Tri Catchment as follows:

x Topography: The elevation data using in the model is taken from STRM DEM
((http://srtm.csi.cgiar.org/) with the spatial resolution 30m.
x Precipitation: daily rainfall of 5 stations in Quang Tri Province from 1997 to 2013
is used.

Fig. 2. Scope and rivers network in the study area.

x Evapotranspiration: This data is inherited from the study of Vu et al. [4].

x Land use and Soil map: The original land use and soil map data from local authority.
x River and lakes: This model is developed on 44 big branches with the length varying
from 20 km to 202 km.

3.2 Hydraulic model set up.

Fig. 3. Hydraulic model set up for downstream of Ben Hai River.

1016 N. D. Vo and T. S. Ho

The MIKE FLOOD model is set up for an area of 55.1 km2. This model involves two
main components:

x 2D model ± MIKE 21: the mesh is created based on the DEM 30m. This mesh covers
an area 55.1 km2 and involves 3.012 elements.
x 1D model ± MIKE 11: River network: The model is developed on 4 big branches at
the downstream (Fig. 2). The geometry of each river branch is specified via cross
section from the measurements.

4 Results and discussions

4.1 Calibration and validation

Hydrological model.
The model is calibrated and validated against the average daily flow at Gia Vong station
for a period of 15 years, from 1997 to 2013. The comparison results show that the sim-
ulated flow is quite similar to the observed flow (Table 1)

Table 1. Statistical indices of MIKE SHE model in Gia Voong station.

Calibration Validation
Parameter Station
R NASH R NASH
Q (m3/s) Gia Vong 0.867 0.76 0.828 0.672

Hydraulic model.
In order to determine the parameters for hydraulic models such as river bed roughness,
it is necessary to calibrate and verify the calculation model for the observed flood event
at Hien Luong station. Through practice and analysis, the author selected to simulate
the 2006 flood event (from September 21, 2006, to September 30, 2006) and the 2009
flood event (from September 21 to October 9, 2009) to test the model. The results show
that there are insignificant differences between simulations and observations (Fig.4,
Fig.5).

Fig. 4. Calibrated hydrograph at Hien Luong station

Flood Mapping For Downstream of Ben Hai River … 1017

Fig. 5. Validated hydrograph at Hien Luong station.

Table 2. Statistical indices of MIKE FLOOD model in Hien Luong station.

Calibration Validation
R NASH R NASH
Water level (m) 0.807 0.52 0.914 0.798

4.2 Flood mapping

Using the set of parameters defined in the previous section to conduct simulation and
calculations of flooding scenarios for downstream Ben Hai River.

Fig. 6. Flood map for 1999 historical event.

In this study, flood maps were developed in accordance with 5 scenarios corresponding
to flood frequency of 0.5%, 1%, 2%, 5%, 10% and historic flood event in 1999. The
results are shown in the Table 3.and Fig 6.
1018 N. D. Vo and T. S. Ho

Table 3. Inundation area (hectare) according to 1999 flood event and frequency
H (m) 1999 10% 5% 2% 1% 0,5%
<0,5 1,441 1,375 1,402 1,437 1,408 1,499
0.5 - 1.0 845 776 836 915 955 1,003
1.0-2.0 688 594 672 792 888 965
2.0-5.0 116 90 99 121 139 155
>5 28 28 28 28 28 29
Total 3,117 2,862 3,037 3,292 3,418 3,651

5 Conclusions

For the purpose of providing an overview of the downstream flooding process of the
Ben Hai River, a hydrological model and a hydraulic model were built. The hydrolog-
ical model (MIKE SHE) is established with all components. The model was calibrated
and verified with the observed flow at Gia Vong station for 15 years from 1997 to 2013
with high reliability. The flooding process is expressed through MIKE FLOOD model.
This model has been calibrated and validated with two historical flood events 2006 and
2009 with high confidence.
Application of MIKE SHE hydrological modelling results and MIKE FLOOD hy-
draulic model, authors conducted flood simulation for typical floods that took place in
the basin in 1999, as well as flood simulation with different frequencies of 0.5%, 1%,
2%, 5%, 10%. The results of the project are expected to provide a basic view of the
flooding process downstream of Ben Hai River, serving as a basis for local authorities
to develop scenarios to deal with natural disasters.

References

[1] ' =RFFDWHOOL ) 0DUUD DQG ( 1LNRORSRXORV ³,PSDFW RI UDLQIDOO VSDWLDO
GLVWULEXWLRQ DQG UHVROXWLRQ RQ IODVK IORRGV UHVSRQVH´ LQ EGU General
Assembly Conference Abstracts, 2013, vol. 15, p. 12828.
[2] T. T. Nguyen, A. Verdoodt, T. Van Y, N. Delbecque, T. C. Tran, and E. Van
5DQVW³'HVLJQRID*,6DQGPXOWL-criteria based land evaluation procedure for
sustainable land-XVHSODQQLQJDWWKHUHJLRQDOOHYHO´Agric. Ecosyst. Environ.,
vol. 200, pp. 1±11, 2015.
[3] N. D. Vo, P. Gourbesville, M. T. Vu, S. V Raghavan, and S.-< /LRQJ ³$
deterministic hydrological approach to estimate climate change impact on river
flow: Vu Gia±7KX%RQFDWFKPHQW9LHWQDP´J. Hydro-environment Res., vol.
11, pp. 59±74, 2015.
[4] V. 19X'''RDQG7/'DQJ³3RWHQWLDOHYDSRWUDQVSLUDWLRQHVWLPDWLRQ
DQGLWVHIIHFWRQK\GURORJLFDOPRGHOUHVSRQVHDWWKH1RQJ6RQ%DVLQ´
 A Framework for Last Planner System Implementation
in Egypt

Eyad Aboseif1 [0000-0003-3097-7712] and Rana Khallaf2

1MSc. Student, Faculty of Engineering, Cairo University, Giza, Egypt.
eyad.hamid95@eng-st.cu.edu.eg
2Assistant Professor, Future University in Egypt, Cairo, Egypt

rana.khallaf@fue.edu.eg

Abstract. The Last Planner System (LPS) has been the focus of several studies
in the Lean Construction community. Previous research has focused on defining
lean in construction, its implementation in various projects, and the challenges
facing the implementation. However, no research has addressed lean implemen-
tation in Egypt. Therefore, this paper addresses the obstacles facing LPS adop-
tion in the Egyptian construction industry and proposes a framework for LPS
implementation in Egypt. A questionnaire survey is carried out involving a num-
ber of construction industry stakeholders to identify the current market
knowledge on LPS and the challenges expected from its implementation in
Egypt. The proposed framework details the techniques to be followed during the
project life cycle and how to measure the effect of LPS on the overall project
performance. The framework also tackles the expected challenges and proposes
key solutions for each one. This paper aims to push the boundaries of the current
state of the construction industry in Egypt by proposing the implementation of
LPS and the steps needed for its adoption in Egypt.

Keywords: Construction Industry in Egypt, Last Planner System, People and

Culture, Construction Planning, Framework and Theory.

1 Introduction

The Last Planner System was developed as a production planning and control system
for smoothing variations in construction workflow originally at the weekly work plan
level. This expanded to cover the full planning and schedule development process from
master scheduling to phase scheduling through Look-ahead Planning (LAP) and
Weekly Work Planning. The Egyptian construction market has been on the rise due to
political and economic changes but is also suffering in terms of time as most of the
government projects need to be fast-tracked, which prompts increased productivity.
LPS comes in handy to aid in such transition by ensuring projects finish on time or even
earlier through vigorous scheduling techniques that can decrease project time (Issa,
2013). This paper proposes a framework for LPS implementation in Egypt as an initia-
tive to further improve and induce the needed changes in the Egyptian construction
industry.

© Springer Nature Singapore Pte Ltd. 2020 1019

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_163
1020 E. Aboseif and R. Khallaf

2 Last Planner System

Alsehaimi et al. (2009) implemented LPS practices on four phases halfway through
two projects in Saudi Arabia; the insights from this case study revealed the unique na-
ture of Middle Eastern countries in terms of LPS implementation, which shows the need
for a framework tailored for the Egyptian market. Issa (2013) implemented LPS sched-
ules and metrics such as Percent Plan Complete (PPC) and Percent Expected Time-
overrun (PET) on an Egyptian milling factory project, however, it was applied on a
small scale project with a duration of only 12 weeks. His focus was on risk factors
effect on project performance which are not affected by lean tools and ignored any
expected LPS adoption challenges in Egypt and how to address them. Swefie (2013)
proposed the first framework to implement Lean Construction in the Egyptian market
focusing on techniques as process map and value stream mapping, neglecting LPS from
the framework. Additionally, Swefie (2013) included a questionnaire regarding lean
concepts in Egypt, but failed to include LPS again. Hence, it is apparent there is a need
for a framework for LPS implementation on larger projects, specifically tailored for the
Egyptian market and its expected adoption challenges.

3 Framework for LPS implementation in the Egyptian Market

The proposed framework is divided into two phases with respect to the period in which
each phase will occur to aid in introducing LPS gradually in Egypt. The first phase will
include the adoption of LPS tools on a small scale project halfway through the project
life cycle, while the second phase will implement LPS and BIM tools on a more so-
phisticated project with a higher budget from the start of construction.

3.1 Phase (1)

The first phase includes the implementation of basic LPS tools such as master/mile-
stone, look ahead and pull schedules with weekly work plans on a small-scale project
with a contract value less than 90 million EGP (about 5 million dollars). The owner of
such prototype project shall be an educational institution or real estate developer as they
are considered the most welcoming stakeholders towards new construction manage-
ment approaches in the construction industry in Egypt.
The project nature is most preferably to be of repetitive nature such as housing or
institutional units. A project will be divided into two stages and an LPS facilitator will
observe and track the construction progress/process in the first stage without any LPS
implementation followed by the second stage which will include the application of var-
ious LPS scheduling tools previously mentioned. This approach is similar to that used
by (Baladrón & Alarcón, 2017) in LPS implementation in a Chilean mining project.
PPC and Tasks Made Ready (TMR) values will be calculated and collected during the
first and second phases to show the change in the construction progress. Phase 1 will
provide results comparing project performance with and without LPS implementation,
which can be used for preparation of a marketing presentation to be used in Phase 2.
A Framework for Last Planner System Implementation … 1021

3.2 Phase (2)

The second phase will include the integration of LPS traditional scheduling tools
with 4D simulation on a larger project scope and value (more than 5 million dollars)
with another prominent general contractor, described as follows:
1. Step (1A) - Before Construction starts: Marketing presentation (including phase 1
results to convince 2nd phase project participants of LPS importance) shall be com-
bined with two workshops to define PPC, TMR calculation and types of LPS sched-
ules and Building Information Modeling (BIM) tools to be used.
2. Step (1B) - Before construction starts: Master/Milestones Schedule with BIM 3D
model of Level of Development (LOD) 300 for the project shall be prepared; (it is
uncommon to find LOD 400 or 500 BIM models in Egyptian projects). 4D schedul-
ing based on the Master Schedule to identify construction sequence and changes in
schedule shall be combined with BIM co-ordination meetings of 4D simulation fol-
lowed by a Pull planning session to identify the best construction sequence with the
Phases/Pull Schedule (Bhatla and Leite, 2012).
3. Step (2) ± During Construction: LAPs on two-ZHHNV¶ basis shall be extracted weekly
by project team identifying weekly work plans tasks, to be discussed in weekly meet-
ings on frontline level with site supervisors. The weekly meetings will identify tasks
made ready after removal of constraints and resource availability. PPC and TMR
will be calculated weekly to record project performance.
4. Step (3) ± After Construction Finish: Results (PPC & TMR) will be collected and
analyzed on a weekly basis. Surveys and interviews will be carried out with a sample
of project stakeholders for evaluation and feedback.

4 LPS status and adoption challenges in Egypt

4.1 Questionnaire on LPS in Egyptian Market

Interviews and close-ended questionnaires were carried out and distributed among
22 participants who represent different stakeholders involved in the Egyptian construc-
tion industry. The participants were identified based on their employment entity target-
ing one of the most active contractors and real estate developers. The first section in
the survey included general data on the participants such as their job title and whether
they have prior knowledge about LPS or not. After passing the first section, a concise
explanation related to the LPS schedules (as previously discussed in this paper) was
provided to the participants in two languages (Arabic and English). Later on, the par-
ticipants were asked to rate the difficulty of the implementation of such a system in an
Egyptian based project and identify possible adoption challenges.
Figure (1) shows number of participants working for owners, contractors or consult-
ants and describes the job title for each participant. The participants in the survey were
distributed along different job titles, to capture various perspectives on this issue, from
supervisors to site engineers, technical office engineers and project managers. Most of
the participants worked on the contractor or owner side as illustrated in Figure (1).
1022 E. Aboseif and R. Khallaf

Other 2 Other 1
Owner or Developer 4 Project Manager 2
Consultancy office 3 Technical Office Engineer 4
Sub-contractor 2 Site Engineer 9
Contractor 11 Site Foreman 6
0 5 10 15 0 5 10
Number of Participants Number of Participants

Fig. 1. Questionnaire Participants Stakeholder Classification and Job title

Figure (2) shows the diversity of the previous knowledge of the stakeholders in the
Egyptian construction market with regards to LPS. 91% of the participants indicated
that they have not heard of the LPS before. This lack of awareness in the industry proves
to be one of the toughest challenges to implement LPS in Egypt. In addition, around 18
of the 22 participants revealed how difficult or even impossible they think the imple-
mentation of LPS in the Egyptian market would be.
LPS previous knowledge LPS Implementation Difficulty
15
Number of Participants

9% 10
10 8

5 3
0 1
91% 0
1 (least 2 3 4 5 (most
Yes No difficult) difficult)

Fig. 2. LPS previous knowledge and implementation difficulty according to Questionnaire

Fig. 3. LPS Adoption main challenges in Egyptian Market

Participants were asked to describe the expected challenges during LPS adoption in
Egypt and the answers were grouped under six titles to broadly represent the challenges
A Framework for Last Planner System Implementation … 1023

as shown in Figure (3). The highest LPS adoption challenges according to the question-
naire were resistance to change in Egyptian culture, complexity of LPS on workers and
site personnel, and project time schedules¶ poor quality as represented in Figure (3).

4.2 Dealing with Egyptian Market Challenges to LPS adoption

Challenges to implementation were collected from the questionnaire survey and strate-
gies were proposed to manage them as discussed in Table (1) below:

Table 1. Challenges to LPS adoption in Egypt and Proposed Strategies

Challenge De-
# Proposed Strategy
scription
x Use of Training workshops and marketing presenta-
tions for LPS effectiveness at the project start will al-
Adaptation to ways help in breaking down the resistance to change
1 New Culture as done in Step (1A) in Phase (2) of this framework.
and Resistance x Phase (1) prototype project LPS results will be pub-
to Change. lished for future use as a tool to convince any upcom-
ing project stakeholders of LPS effectiveness.

Lack of com- x Adopt system of hierarchy meetings similar to that

munication and followed by (Reed, et al., 2017) .
information in x LAP meetings attended by discipline team leaders
Egyptian sites; ³&DSWDLQV´ and they then deliver the LAP tasks with
2/3 Time spent on the constraints to the discipline team.
planning issues x Weekly meetings with site supervisors attended by the
seen as a waste remaining discipline teams.
of time

x A system of Competitive Partnering ensuring com-

Lack of com- mitment to the adoption of LPS in return for allocation
4 mitment from of future projects and shared cost savings can be used.
different stake- x Use of project PPC/TMR performance-based bonus
holders schemes for workers. (Nani & Apraku, 2016)

x The adoption of strict scheduling project specifica-

Poor quality of tions since project tendering or initiation.
5 Time Schedules x Review process for time schedules by experienced
in Egyptian schedulers on prepared schedules by the contractor.
Projects
Project or Competitive Partnering allows for agreeing on flexible project objectives to
be included in the contract (Stehbens, et al., 1999). One of them can be LPS adoption,
which may prove worthy to ensure commitment of project stakeholders to enforce LPS.
The use of bonuses based on better PPC or TMR weekly performance values is to deal
with the workers in Middle East or Africa who are largely paid weak salaries compared
 1024 E. Aboseif and R. Khallaf

to their counterparts in other countries. Nani & Apraku (2016) found that workers in
Ghana (who represent African ZRUNHUV¶ conditions) ranked bonus schemes as their fa-
vorite financial reward system and said their work performance was significantly
boosted when financial incentives existed. Reed et al. (2017) approach to continuous
LPSURYHPHQWPHHWLQJVLVWREHDGRSWHGZLWKVRPHFKDQJHVDOORZLQJWKH³&DSWDLQ´RI
each discipline team to pass LAP cleared tasks to rest of team members, minimizing
time spent on planning issues. This approach also aids in efficient and organized com-
munication between project members to tackle lack of communication in Egyptian site.

5 Conclusion

Through a detailed literature review of LPS implementation on various case study

projects in Egypt and the Middle East, this paper presents a framework for LPS imple-
mentation in Egyptian market. The framework adopts a strategy divided on 2 phases,
the strategy will integrate BIM tools and concepts with LPS system. in addition to that,
the framework anticipates the challenges for LPS adoption in Egypt offering strategies
to deal with each challenge. The next step is the application of the proposed framework
on a real-life case study to understand the effect of LPS implementation based on the
framework. The case study will offer further insights on how to further refine the frame-
work and to what extent the LPS adoption resulted in better project performance.

6 References

1. AlSehaimi, A., Tzortzopoulos, P., & Koskela, L. (2009). Last Planner System: Experiences
from Pilot Implementation in the Middle East. 17th Annual Conference of the International
Group for Lean Construction, (pp. 53-66). Taipei, Taiwan.
2. Baladrón, C., & Alarcón, L. F. (2017). Assessing the Impact of Lean Methods in Mining De-
velopment Projects. 25th Annual Conference of the International Group, (pp.137-144). Hera-
klion, Greece. doi: https://doi.org/10.24928/2017/0272
3. Bhatla, A., & Leite, F. (2012). Integration Framework of BIM with the Last Planner System.
20th Annual Conference of the International Group for Lean Construction. San Diego.
4. Issa, U. H. (2013). Implementation of Lean Construction Techniques for Minimizing the Risks
Effect on Project Construction Time. Alexandria Engineering Journal, 697-704.
5. Nani, G., & Apraku, K. (2016). Effects of Reward Systems on Performance of Construction
Workers in Ghana. Civil and Environmental Research, 8(7).
6. Reed, D., Ashcraft, H., Khanzode, A., Fischer, M., Rischmoller, L., & Berg, P. (2017).
Integrating Delivery Of A Large Hospital Complex. Proceedings of the 25th Annual
Conference of the International Group for Lean Construction , (pp. 201-208). Herklion, Greece.
7. Stehbens, K., Wilson, O., & Skitmore, M. (1999). Construction Project Partnering: Two Case
Studies. Proceedings The International Council for Building Research Studies and
Documentation (CIB) W55 and W65 Joint Triennial Symposium, "Customer Satisfaction: a
Focus for Research and Practice", (pp. 229-237). Cape Town.
8. Swefie, M. G. (2013, 11 26). Improving Project Perfromance Using Lean Construction in
Egypt: A Proposed Framework. Retrieved from Digital Archive and Research Repository in
American University in Cairo: http://dar.aucegypt.edu/handle/10526/3728
 Climate Change Impact on Urban Flooding in Quynhon
City, Vietnam

Duc Phuoc Vo [1-2], Muhammed A Bhuiyan [2] and Ngoc Duong Vo[1*]

(1) University of Science and Technology, The University of Danang, Danang, Vietnam.
(2) School of Engineering, RMIT University, Melbourne, Australia

Abstract. Flooding is one of the major problems in many cities. Adequate devel-
opment of a drainage system is a primary measure the city used to deal with the
problem. However, the risk of exceedance of design flood is more likely to in-
crease due to the impact of urbanization as well as climate change. Changes in
rainfall pattern and sea level rise increase the risks of inundation of city dwellings
and amenities. This paper investigates the changing of flooding performances
using the Stormwater Management Model (SWMM) to describe the impact of
climate change on drainage in Quy Nhon city, Vietnam. Climate scenarios are
developed for future 2050 and 2100 by following the emissions proposed in IPCC
representative concentration pathways (RCP) of greenhouse gases. The results
examined the future flooding areas and their durations. It also highlights the in-
appropriateness in drainage design which could contribute to flooding increase
in the city.

Keywords: Urban flooding, Climate change, Extreme rainfall, Sea level rise,
Stormwater Management Model (SWMM)

1 Introduction

Urban flooding is one of the severe increasing problems which challenges the sustain-
able development of many cities in developing countries. In terms of climate change,
the issue tends to become more significant in both floods frequency and magnitude, [1,
2]. In 2013 the Intergovernmental Panel on Climate Change (IPCC) reported that pre-
cipitation and sea level have been projected to increase in many coastal regions [3]. As
a result, these coastal cities are likely to become more vulnerable to flooding, which
could influence the achievement of sustainable development goals. In order to deal with
the problem, developing of drainage systems is a major applied approach in reducing
risks of urban flooding in many regions [4]. In many developing countries, however,
pressures from economic growth and population explosion lead to the fact that the
drainage system investment is unable to keep up with the urbanization rate which may
lead to more serious inundation
Similarly, urban flooding is a great challenge for the sustainable development of
many cities in Vietnam. The country has an annual rainfall amount reaching 2500 mm
and averages 10 tropical storms landing every year along with high-intensity rainfall
[9].

© Springer Nature Singapore Pte Ltd. 2020 1025

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_164
1026 D. P. Vo et al.

Some research has been implemented to investigate increasing in the vulnerability

of urban flood due to climate change in Vietnam. One of the trends is climate change
impacts are analyzed together with urbanization and urban head island in cities [5]. In
another way, flood risks would be assessed narrowly in sea level rise caused by climate
change and storm which figures out the hazard area within low elevation [6]. In those
studies, some have simulated the change in climate condition to urban drainage by de-
velopment model in research areas such as SWMM [1] or Mouse [7].
However, these studies still show a lack of field of views in urban drainage problem.
Most of these studies focus on low elevation areas in Red River Delta [6] and Mekong
Delta[1, 5] which is not fully representative for the whole of Vietnam. In addition, some
studies tend to evaluate urban flooding accompanied by fluvial flooding. Meanwhile,
extreme rainfall and sea level rise, in contrast, play a minor role in this event [1, 7].
Therefore, evaluating assessment climate change impact on urban flooding in the
central region with hourly precipitation data would cover gaps of knowledge.

2 Methodology

In order to examine the potential impacts of climate change on urban flooding, urban
drainage models are implemented with different scenarios representing historical and
future climate in the research area. In which, short-duration and extreme rainfall data
are required for the region of interest. On the one hand, historical data is obtained by
statistics and analyzing recently history flooding at gauging station in the research area.
In contrast, climate models would give prediction change in climate conditions based
on greenhouse gases emission scenarios [2]. These data would be integrated into an
urban drainage model for simulation which is simply illustrated in Fig. 1.

Fig. 1. Urban hydrology impact analysis methodology

2.1 Climate scenarios

Future climate scenarios were developed based on RCMs results available in the region
of interest and delta change method. This is because integrating GCMs results for urban
catchment scale is not convincing. GCMs models are not able to provide accurate re-
sults of raining process for high regional and temporal resolutions requirement of a city
Climate Change Impact on Urban Flooding in Quynhon … 1027

[8]. In the research, national climate change scenarios give projections for future cli-
mate for all of Vietnam which will be used to develop scenarios [9].
There are three main scenarios being proposed depending on flooding types and cli-
mate change scenarios as well as the weather situation of the city: based cases, mid-
century (2050) and end-century (2100) scenarios. These scenarios were simulated
based on current drainage network which was upgraded in 2009 and 2010. For future,
scenarios would be developed based on high greenhouse gases emission RCP 8.5 and
medium emission scenarios RCP 4.5. Figure 4 illustrated rainfall hydrograph for each
scenario. Secondly, investigation of flooding risks with historical rainfall in different
return periods (2 years. 5 years and 10 years) and sea level in future are expected to
provide an understanding increase in vulnerability of inundation in the city.

Fig. 2. Rainfall pattern of simulated scenarios

2.2 Urban drainage model

Storm Water Management Model (SWMM) is a 1-D model which is widely used and
accepted by engineers and researchers. Recently, the model shows its capacities in eval-
uating the impact of climate change [1], investigating low impact development or as-
sessing stormwater quality. Besides that, conveniences in coupling and integration with
other models and software giving great change for further development. Therefore,
SWMM will be used to evaluate the impacts of climate change on the research area.

2.3 Case study

This study was conducted in Quy Nhon ± the capital city of Binhdinh province, Vi-
etnam. The city has been suffered from many flooding recently. The drainage system
of Quynhon City is a complex system which discharges stormwater to three major des-
tinations which are Phuhoa lake, Thinai lagoon, and Quynhon bay. In which, Phuhoa
lake is the most sensitive of fluvial flooding that may impact on stormwater flow. In
terms of meteorology, the majority of rainfall is concentrated in rainy seasons from
September to December, accounting for 80% of annually. Map of the drainage system
on SWMM for the case study is shown on Fig. 3.
1028 D. P. Vo et al.

Fig. 3. Urban drainage model in SWMM for the city of Quynhon

3 Result and discussion

The simulation gives different results for urban flooding under alternative climate sce-
narios in the city of Quy Nhon. They confirm an increasing in happening probability of
urban flooding and the impact of climate change on the stormwater discharge of the
city.

3.1 Flood risks under current condition

Simulation results for the2009 and the 2010 flood events find an agreement with reality
circumstance in the region of interest. On the one hand, flood 2009 was less significant
in interest area, with 16 flooded nodes appearing in the area. On the other hand, the
2010 event is more likely to impact wider with the number of flooded nodes being
roughly four times higher than the 2009 event. The simulation also illustrates that many
higher positions in city centres such as in Le Hong Phong, Hai Ba Trung or Tran Hung
Dao stress, which is mainly because of exceeding of conveying capacity of the drainage
system. In addition, the number of flooded nodes in 2010 flooding event is nearly four
times larger than 2009 figure, although the total amount of rainfall in 2010 is smaller
than 2009 even (with 348.5 and 367.6 mm respectively). In terms of flooding duration
and volume, an inundated node in the 2009 event is more significant than 2010 figures.
In detail, while average flooded time is simulated is approximately two hours, is two
and a haft times smaller than 2009 with over five hours.

3.2 Flood risks under climate change condition

The increase in the number of flooded nodes, duration, flood magnitude as well as fre-
quency is consistent with observation data. In addition, climate change projection with
Climate Change Impact on Urban Flooding in Quynhon … 1029

high greenhouse gases emission scenarios (RCP8.5) are the worst cases for flooding in
the city.
Due to climate change, more areas in the city are projected to be impacted by flood-
ing. If climate change follows RCP 4.5 scenarios, the number of inundated nodes in the
city in 2100 is predicted to reach nearly double the figure for 2010. Inundation is more
likely to be serious in RCP 8.5 scenario than RCP 4.5 scenario. In 2050, flooding is
projected to impact 53 nodes when the simulation follows RCP 4.5 scenarios, the figure
for the RCP8.5 scenario is predicted at 64 nodes, which 11 nodes more than the 2010
figure. Similarly, in RCP 8.5 scenario the number of flooding nodes in 2100 is projected
to account for 33.6 percent of simulated nodes, which 5.1 percent more than RCP 4.5
figure of 28.5 percent.

Table 1. Node flooding of simulation scenarios

Hours flooded Volume Flood flow

Flooded
Scenario (hours) (106 l) (m3/s)
nodes
Total Maximum Total Maximum Maximum Average
2009 16 45.93 10.94 61.71 23.46 1.17 0.60
2010 34 66.10 7.02 155.48 18.96 3.09 0.89
2050 rcp 4.5 53 123.2 10.94 354.10 32.53 4.20 1.18
2050 rcp 8.5 64 165.28 12.32 534.59 64.54 7.55 1.36
2100 rcp 4.5 60 145.55 11.54 435.77 35.77 4.58 1.26
2100 rcp 8.5 73 215.35 13.85 757.15 93.02 13.85 2.95
Furthermore, the challenges from increased flooding area, increase in flood duration
and magnitude threaten the city. By 2050, the average flooding time of the city in-
creases from 1.94 hours in 2010 to 2.33 hours and 2.58 hours following RCP 4.5 and
RCP 8.5 scenarios respectively. Notably, at peak times for some nodes, simulated re-
sults project inundated durations of over 10 hours. The projection for 2100 is more
drastic, the city would deal with nearly an average 3 hours flood. In the longest inun-
dation period, the figure reaches 13.85 hours. Similarly, flood magnitude would in-
crease with the same trend as flood duration. Specifically, the projected total flood vol-
ume has an increase in range from 127 percent in 2050 RCP 4.5 scenario to 397 percent
in 2100 RCP 8.5 scenario compared to 2010 flooding.

Fig. 4. Simulation result of different return periods rainfall

1030 D. P. Vo et al.

Simulating urban drainage systems with different rainfall amounts over periods of 2 to
10 ten years confirm increases in the frequency of flooding. In 2010, with 2 years return
periods rainfall, all of the parameters such as flooded nodes, flooding hours or flooded
volume were essentially trivial. However, the same rainfall pattern in future sea level
conditions leads to more significant inundation. More than one-third of flood critical
points which were safe in the 2010 simulation would be flooded with sea level in 2050
and 2100, see Fig. 4.

4 Conclusion

This study has shown that the city is becoming more vulnerable to flooding due to
climate change. It also is significant factors accounting for resulting in expanding area,
increasing duration time and enlarged magnitude of future flooding in the area. In the
case where greenhouse gases emission follows the RCP 8.5 scenario, the flooding prob-
lem becomes the worst with nearly one-third of the city being impacted. This new un-
derstanding should help improve predictions of the impact of urban flooding in coastal
cities separately from fluvial flooding.

References
1. Huong, H.T.L. and A. Pathirana, Urbanization and climate change impacts on future urban
flooding in Can Tho city, Vietnam. Hydrology and Earth System Sciences, 2013. 17(1): p.
379-394.
2. Willems, P., et al., Climate change impact assessment on urban rainfall extremes and urban
drainage: Methods and shortcomings. Atmospheric Research, 2012. 103: p. 106-118.
3. IPCC, Climate Change 2013: The Physical Science Basis. Contribution of Working Group
I to the Fifth Assessment Report of the Intergovernmental Panel on Climate Change, ed.
T.F. Stocker, et al. 2013: Cambridge University Press, Cambridge, United Kingdom, and
New York, NY, USA, 1535 pp. 1535.
4. Pazwash, H., Urban stormwater management. 2016: CRC Press.
5. Storch, H., and N.K. Downes, A scenario-EDVHG DSSURDFKWR DVVHVV +R &KL0LQK &LW\¶V
urban development strategies against the impact of climate change. Cities, 2011. 28(6): p.
517-526.
6. Neumann, J.E., et al., Risks of coastal storm surge and the effect of sea level rise in the Red
River Delta, Vietnam. Sustainability, 2015. 7(6): p. 6553-6572.
7. DiGregorio, M., Learning from Typhoon Mirinae: Urbanization and climate change in Quy
Nhon City, Vietnam. Institute for Social and Environmental Transition. Hanoi, Vietnam:
ISET-Vietnam, 2013.
8. Willems, P. and M. Vrac, Statistical precipitation downscaling for small-scale hydrological
impact investigations of climate change. Journal of Hydrology, 2011. 402(3-4): p. 193-205.
9. Tran, T., et al., Climate Change and Sea Level Rise Scenarios for Viet Nam - Summary for
Policymakers. 2016.
 Health Impacts of Traffic-related Air Pollution:
Cause-effect Relationships and Mitigating Measures

Huong Le Thi1,2
1
Technische Universität Darmstadt, Otto-Berndt-Str. 2, 64287 Darmstadt, Germany
2 Vietnamese-German University, Le Lai Street, Hoa Phu Ward, Thu Dau Mot City, Binh
Duong Province, Vietnam
huong.lt@vgtrc.vgu.edu.vn

Abstract. The health impacts of traffic-related air pollution (TRAP) are one of
the major concerns regarding the sustainability of the transport sector. Particu-
larly in urban areas, transport is a main contributor to air pollution and affects a
large part of the population due to the high number of motor vehicles and urban
population. Recently, the evidence from epidemiological studies on the health
impacts of TRAP has increased significantly. The major evidence indicates that
TRAP increases the risk of mortality and morbidity. Despite the significant im-
provements in reducing vehicle emissions, TRAP is still increasing in many areas
worldwide and becoming a global threat to human health. In order to solve the
problem, a clear understanding of the existing situation is necessary. Therefore,
a comprehensive cause-effect relationship between TRAP and its health effects
is developed, using the Driving Force-Pressure-State-Exposure-Effect-Action
(DPSEEA) framework. Then, some measures for mitigating the health impacts
of TRAP are presented.

Keywords: Traffic-related air pollution, Health impacts, DPSEEA framework.

1 Introduction

Clean air is considered as a basic requirement for human health and well-being.
However, air pollution currently is posing a significant threat to health worldwide. Ex-
posure to ambient air pollution is a leading contributor to the global burden of disease,
which increases the risks of morbidity and mortality. Particulate matter (PM), ozone
(O3), and nitrogen dioxide (NO2) are considered as most harmful TRAPs to human
health. Among these pollutants, particulate matter (PM) has the greatest impact on hu-
man health. The long-term exposure to PM2.5 caused 4.2 million deaths and 103.1 mil-
lion lost years of healthy life in 2015, representing 7.6 % of total global mortality, mak-
ing it the fifth-ranked global risk factors in 2015, and around 254,000 deaths were
caused by exposure to ozone additionally [1].
The awareness of the health effects of TRAP has grown, and many countries have
implemented stricter emission control and made significant progress in reducing the

© Springer Nature Singapore Pte Ltd. 2020 1031

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_165
1032 H. L. Thi

emissions from motor vehicles and in improving air quality. However, the number of
ZRUOG¶V PRWRULVHG YHKLFOHV DQG WUDYHO GHPDQG VWLOO LQFUHDVH VLJQLILFDQWO\ ZKLFK are
offset the benefit of improvement [2]. Thus, a clear understanding of the cause-effect
relationships between TRAP and its health impacts is required, to provide guidance for
policy makers, authorities and public when developing the mitigation solutions.

2 Methods

The DPSEEA framework was developed by WHO in 1995 to describe the cause-
effect linkages between human health and environment. Different driving forces gen-
erate the pressures that affect the state of the environment and ultimately human health,
through the various exposure pathways [3]. The developed framework for this topic is
presented in figure 1.

3 Results and discussions

3.1 Driving force

Driving forces are factors on the macro level and affect at large scale, presenting the
root causes of the problems. In the transport sector, the main driving forces are eco-
nomic development, population growth, urbanisation, policies, and technology devel-
opment. These factors highly affect the transport demand. Historical trends have shown
that there is a close statistical correlation between the growth of Gross Domestic Prod-
uct (GDP) and income level with the growth of passenger and freight transport demand
[4]. 7KHZRUOG¶VSRSXODWLRQLVSUHGLFWHGWRFRQWLQXHWRJURZZLWKSHU\HDUUH
sulting in additional 83 million people annually and reaching 9.8 billion in 2050; the
largest growth is expected in Africa and Asia [5]. The same trend can be seen for ur-
EDQLVDWLRQ ZRUOGZLGH DURXQG  RI ZRUOG¶V SRSXODWLRQ LV OLYing in urban areas in
2018, it is projected to grow to 68% by 2050; and almost 90% of this growth is hap-
pening in Asia and Africa [6]. As a result, the greatest growth in transport demand can
be predicted in Asia and Africa. The policies such as transport policies, urban policies,
environmental policies and economic policies can also be driving factors for influenc-
ing mobility. For example, pricing policies may lead to a reduction of demand for pri-
vate motorised transport and an increase in demand for active transport and public
transport. Finally, the development of technology such as autonomous driving, provid-
ing a travelling opportunity for non-drivers may have a huge impact on future mobility.
Health Impacts of Traffic-related Air Pollution … 1033

Fig. 1. The cause-effect relationships between TRAP and health effects

3.2 Pressure

Increasing travel demand creates pressure to increase the number of private vehicles,
energy consumption, climate change and transport infrastructure. It also stimulates
noise pollution and air pollution and land-use consumption. 7KHZRUOG¶VPRWRUYHKLFOH
population is growing continuously, with the largest rates in Asia and Africa. For ex-
ample, the number of motorcycles in Vietnam increased from 4 million vehicles in 1995
to 52 million in 2016, with 565 motorcycles/1000 inhabitants; the number of cars grew
from 460,000 vehicles in 1995 to 3.25 million vehicles in 2016, with 35 cars/1000 in-
habitants [7]. A large part of the final energy consumption is caused by transport, the
transport sector (in Germany) is responsible for nearly 30% of the final energy con-
sumption [8]. Petrol and diesel vehicles are still dominant in the vehicle fleet, and they
are one of the main causes of greenhouse gas emission, which is the leading source of
climate change. The greatest growth in travel demand is predicted to happen in Asia
and Africa. Thus, the pressures created by transport in these areas on current infrastruc-
ture and environment will be much higher than in developed countries. These countries
are also suffered from high air pollution and noise pollution levels. The increase in
transport demand leads to the expansion of roads and parking spaces for vehicles, which
reduces the land for green spaces, public spaces, parks and other social purposes.

3.3 State
Motor vehicles emit a wide-range of air pollutants and can be divided into three
groups: (1) Exhaust emissions as results of the combustion process, (2) Non-exhaust
emissions produced by mechanical abrasion and corrosion of vehicle parts such as
YHKLFOH¶VW\UHV and road wear, and (3) Evaporative emissions resulting from vapours
escaping IURPYHKLFOH¶VIXHOV\VWHP [9]. Exhaust emissions have been regulated widely
and reduced significantly by implementing stricter vehicle emission regulations and
innovative pollution control technologies worldwide. However, the non-exhaust emis-
sions currently are neglected and unregulated, and it will be a major source of TRAP
when the emission control for combustion becomes more effective. Recently, the non-
1034 H. L. Thi

exhaust emissions contribute up to 60 % and 42% of primary PM10 and PM2.5 of the
total road transport emissions of these pollutants respectively [10].
There is a number of factors, which are influencing vehicle emissions, and can be
grouped into four main categories: (1) Vehicle characteristics: Vehicle type, fuel used
and fuel quality, operation and maintenance condition of the vehicle, vehicle age and
vehicle kilometre travelled, vehicle size, loading factor/occupancy rate, emission stand-
ard, and emission control technology. (2) Traffic conditions: Traffic volume, the av-
erage speed of traffic flow, traffic signal control, speed limit. (3) Weather conditions:
Wind speed, wind direction, air temperature, snow, rain, and presence of sunlight. (4)
Others: Road conditions, and driving behaviours. These factors affect differently to
the amount of vehicle emissions and require a comprehensive investigation.
After emitting from vehicles, some secondary pollutants are formed in the atmos-
phere, then these pollutants are dispersed into the air depending on the type of pollu-
tants, weather conditions and surrounding environments.

3.4 Exposure

+XPDQH[SRVXUHLVSHRSOH¶VFRQWDFWZLWKDSROOXWDQWZKHUHLVDQHFHVVDU\FRQGLWLRQ
for the occurrence of a health effect [11]. The number of exposed people, the duration
of contact, and the concentration of the pollutant in the place where the contact takes
place define the magnitude of the exposure and the extent of health effects in a popula-
tion. 7KHSHRSOH¶VDFWLYLWLHVDQGWKHLUEHKDYLRXUSDWWHUQVhighly influence their expo-
sure level to TRAP. For measuring the level of exposure to TRAP, the direct indicator
can be used, which is the amount of air pollutants that person inhaled, but the study on
this is still limited. The other indicators including time spent in traffic, proportion of
population living in proximity of sources of air pollution, and proportion of population
exposed to air pollutants in excess of standards or guidelines are more common used
for estimating the exposure of an individual and population to air pollutants [2]. HEI
identified an exposure zone within a range of up to 300 to 500m from a highway or a
main road as the area most affected by TRAP.

3.5 Effect

Traffic emits a wide range of air pollutants, but the strongest evidence for public
health concern are PM, NOx, and O3, and exposure to these pollutants are associated
with a range of health effect. For example, PM with diameter less than 10 microns
(PM10) and less than 2.5 microns (PM2.5) are the most harmful to human health; these
particles can cause premature death, harmful effects on cardiovascular system and res-
piratory system, reduce lung function, increase frequency of respiratory diseases, and
reduce life expectancy [12]. The air pollution from vehicles caused 184,000 deaths
globally, including 91,000 deaths from ischemic heart disease, 59,000 deaths from
strokes, and 34,000 deaths from lower respiratory infections, chronic obstructive pul-
monary disease, and lung cancer [13]. The studies on health impacts of TRAP are
mostly conducted in developed countries, while in developing countries, where the
problems would expect to be more severe, but the study about this topic is still limited.
Health Impacts of Traffic-related Air Pollution … 1035

Recently, the other health outcomes caused by TRAP such as reduce the cognitive func-
tion, low birth weight, and preterm birth are investigated, but the results are not con-
sistent and insufficient, therefore, required for further investigation. In addition, TRAP
may result to the other indirect health impacts because it creates the possibility of re-
ducing the level of outdoor exercises of people or choosing not to use active transport
[14]. The other indirect effect may come from climate change since road transport is
one of the main contributors to CO2 emission influencing climate change, which con-
sequence in severe extreme weather events and affected million people.

3.6 Action

The action can be taken at each step in the framework, different interventions can be
developed for different levels. The measures for reducing the health impacts of TRAP
can be classified into four groups including (1) Avoiding measures aiming to reduce
the travel demand both in the volume and travel distanced by motorised vehicles, (2)
Shifting measures aiming to shift the private motorised travel demand to cleaner and
healthier transport modes, changing the time-activity patterns of people to reduce the
exposure to TRAP, (3) Improving measures targeting to improve the vehicle energy
efficiency and traffic flow, and (4) Measures influencing the dispersion of TRAP.
There is a number of measures which have proved to be effective in avoiding and
reducing travel demand, such as mixed land use, congestion charge, low emission zone,
vehicle sharing, and parking control. For example, the introduction of low emission
zone in Germany since 2008 has brought positive effects on reducing the air pollution
concentrations of PM10, PM2.5 and NOx [15].
Currently, there is an increasing trend in using electric vehicles globally. The electric
vehicles may be better in terms of air pollution reduction, but the other health impacts
such as safety remain unclear. The shifting from motorised vehicles to public transport
and active transport seems to bring better health benefit for people not only in terms of
air pollution but also physical activities and other health aspects. The changing the time-
activity patterns of people requires a good information system providing them infor-
mation about air quality, then they can change their travel time or travel route to reduce
the exposure to TRAP.
The measures for improving vehicle energy efficiency should focus on both, new
registered vehicle and in-use vehicles through the application of stringent emission
standards and inspection/maintenance program. Measures to improve traffic flow such
as traffic signal control and speed limit could play an important role in reducing the
TRAP. Finally, the measures influencing the dispersion of TRAP such as road design,
planting trees along the roads, and the sound wall should also be considered.

4 Conclusions

Despite the huge efforts have been conducted to reduce vehicle emissions, the ben-
efits are offset by the increase in the travel demand. Urban areas, particularly in devel-
oping cities, are expected to suffer the greatest pressures from the rapid growth of travel
1036 H. L. Thi

demand. In these cities, the existing situation about health effects of TRAP is estimated
much higher than in developed cities, but the research on this issue is still limited due
to the insufficient data, resulting in the poor awareness of governments and public on
this issue. Fortunately, the health impacts of TRAPs are preventable; the wise develop-
ment can help to leapfrog over the disaster of the past. To solve the problem, collabo-
ration and coordination actions are required at local, national, and international levels.
The solutions involving technological development, structural changes and behavioural
changes are needed, together with an integrated multidisciplinary approach.

Reference
1. &RKHQ $ - %UDXHU 0 %XUQHWW 5 $QGHUVRQ + 5 )URVWDG - (VWHS . «
Forouzanfar, M. H. (2017). Estimates and 25-year trends of the global burden of disease
attributable to ambient air pollution: an analysis of data from the Global Burden of
Diseases Study 2015. The Lancet, 389(10082), 1907±1918.
https://doi.org/10.1016/S0140-6736(17)30505-6
2. HEI. (2010). Traffic-Related Air Pollution: A Critical Review of the Literature on
Emissions, Exposure, and Health Effects. Health Effects Institute.
3. WHO. (1997). Health and Environment in Sustainable Development: Five Years after
the Earth Summit. World Health Organisation.
4. OECD/ITF. (2017). ITF Transport Outlook 2017. OECD. Paris.
https://doi.org/10.1787/9789282108000-en
5. UN. (2017). World Population Prospects The 2017 Revision, Key Findings and Advance
Tables.
6. UN. (2018). World Urbanization Prospects: The 2018 Revision. UN DESA.
https://population.un.org/wup/Publications/Files/WUP2018-KeyFacts.pdf
7. VGTRC. (2018). 9DLWUzFͯD[HPi\WURQJKL͏QW̩LYjW˱˯QJODLW̩LYL͏WQDP.
8. BMU. (2018). Climate Action in Figures Facts, Trends and Incentives for German
Climate Policy 2018 edition.
9. EEA. (2016). Explaining road transport emissions - A non-technical guide. European
Environment Agency. https://doi.org/10.2800/71804
10. EEA. (2018). Emissions of air pollutants from transport. European Environment Agency.
https://doi.org/10.1088/0004-637X/705/2/L195
11. WHO. (2005). Health Effects of Transport-related Air Pollution: Summary for policy-
makers. World Health Organization.
12. WHO. (2013). Health Effects of Particulate Matter: Policy implications for countries in
eastern Europe, Caucasus and central
13. Bhalla, K., Shotten, M., CRKHQ$%UDXHU06KDKUD]6%XUQHWW5«0XUUD\&
J. L. (2014). Transport for Health: The Global Burden of Disease from Motorized Road
Transport. https://doi.org/10.1016/B978-012373960-5.00335-X
14. Geurs, K. T., Boon, W., & Wee, B. Van. (2009). SRFLDO,PSDFWVRI7UDQVSRUWࣟ/LWHUDWXUH
Review and the State of the Practice of Transport Appraisal in the Netherlands and the
United Kingdom. Transport Reviews, 29. https://doi.org/10.1080/01441640802130490
15. Jiang, W., Boltze, M., Groer, S., & Scheuvens, D. (2017). Impacts of low emission zones
in Germany on air pollution levels. Transportation Research Procedia, 25(July 2016),
3370±3382. https://doi.org/10.1016/j.trpro.2017.05.217
 Evaluation of the Effects of Traffic Management at
School Areas

An Minh Ngoc1 and Khuat Viet Hung2

1
University of Transport and Communications, 3 Cau Giay, Hanoi, Vietnam
2 National Traffic Safety Committee, 80 Tran Hung Dao, Hanoi, Vietnam

anminhngoc@utc.edu.vn, kviethung@ntsc.gov.vn

Abstract. Traffic accidents are known as one of the leading causes of mortality
among teenagers and young adults in Vietnam. Specifically, the risky road be-
havior lead to 6.2 per cent of killed and injured persons are youth aged between
0 and 18 years. While children injured or killed when walking or travelling as
passengers in motorcycles are a serious concern in Vietnam, it is obviously that
road users perceive on traffic safety for children is one of the most significant
factors affecting accident reduction. The proper application of traffic manage-
ment measures at school areas needs to be recognized as one of the most com-
mon and vital ways to improve traffic safety for children.

Keywords: Children, traffic accident, school zones.

1 Introduction

Road crashes are recognized as one of the top two causes of mortality among young
people. Children account for approximately 21% of injury-related mortalities world-
wide and two thirds of child road traffic injury deaths occur in the South-East Asia
[8]. Among risk factors, road behaviors are one of the most important predictors [3].
Children and young people are at risk in road environments around schools and col-
leges. Their behaviors can be less predictable than adults and they may have errors of
judgement when dealing with traffic. Therefore, managing traffic and road safety at
schools become a concern in worldwide.
There are different solutions are applied to the same safety issue. However, the re-
viewed literature [3,4,6] shows that the most appropriate solution for reducing traffic
accident may involve the four Es approach (Engineering, Education, Enforcement and
Encouragement). In which, Engineering focuses on implementing facility improve-
ments; Education focuses on taking the normal road safety education curriculum;
Enforcement focuses on campaigns by the police to reduce illegal actions by road
users such as speeding, red light violation; and Encouragement aims to change the
behavior of road users.
According to [1], one of the most important goals for a suitable city is ensuring
traffic safety of all traffic movements, including children and elderly people. This

© Springer Nature Singapore Pte Ltd. 2020 1037

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_166
1038 A. M. Ngoc and K. V. Hung

goal should be divided into four objectives (i) safe road infrastructure, (ii) safe road
user behavior, (iii) traffic safety management, and (iv) safety environment. The effec-
tiveness of traffic management measures is obtained by the achievement of these
goals and objectives.

Table 1. Categorization of traffic management measures

Goup Measures Group Measures

ROW on unsignalized inter- Provision of maps and
section guidebook
Improvement of road surface Education Production of textbooks
Installation of safe facilities measures Production of materials
Education by accident
Effective pedestrian signal
Engineering scenes
measures Installation of crosswalk Reinforcement of regulation
Installation of pedestrian Enforcement Reinforcement of traffic
priority zone crackdown
Active travel to and from
Provision of ITS system
Encourage- school
ment Establishment of school
zone

2 Traffic Safety Issues at Schools in Vietnam

In Vietnam, where this study is carried out, road crashes account for approximately
6.2 per cent of fatalities among teenagers. As examined by statistics, children injured
or killed while travelling as passengers constitute the largest category of children
involved in road traffic crashes. However, the trend of children involved accident
while driving is appallingly rising.

Table 2. Distribution of serious accidents for children

Children 2014 2015 2016

Passenger 157 141 263
Driver 146 198 219
Pedestrian 159 210 196
Source: Department of traffic police, 2017
To determine the contributing factors influencing the occurrence of each accident,
80 road traffic accidents involving children were analyzed in detail. Of the 80 acci-
dents, human factor (91.4%) had contributed to majority of fatal/serious injuries, fol-
lowed by the infrastructure and environment factor (7.4%) and vehicle factor ac-
counted for the smallest proportion.
+RZHYHUIURPURDGXVHU¶VSHUVSHFWLYHKXPDQHUURUVDUHDIIHFWHGE\PDQ\IDFWRUV
According to School Traffic Safety Survey (SUD, 2017), it follows that road users
Evaluation of the Effects of Traffic Management … 1039

consider unsafe road facilities, and unsafe road environment as the contributing fac-
tors for risky behavior.

Table 3. Contributing factors influencing the Table 4. 5RDGXVHU¶VSHUFHSWLRQRQWKH

occurrence of 80 accidents causes for human errors

Accident causes Percent Causes for human errors Percent

Human error 91.4% Illegal occupancy of sidewalk 27.5%
Infrastructure and environ- Lack of sidewalk or ped. road 26%
7.4%
ment error Under-standard road design 18%
Vehicle error 1.2% Damage road pavement 14.7%
Source: Department of traffic police, Lack of crosswalk or
2017 4.8%
ineffective crosswalk
Others 10.5 %
Source: SUD, 2017
Combination the causes recorded from both sides, it can be concluded that the traf-
fic safety issues raised may be caused by road users or by problems in the road envi-
ronment. Maximizing safety for children depends on both addressing road infrastruc-
ture as well as the behavior of the people who use it.
This study had two main objectives: First, to identify the applicable measures to
ensure traffic safety for children; and second, to assess the effect of these measures on
WKHURDGXVHUV¶EHKDYLRU

3 Sample and Methods

3.1. Sample

To adopt with the objectives of this study, a total sample of n = 300 public citizens
(170 males, being 56.67% of the sample, and 130 females, representing 43.33% of the
sample) was used, all coming from Hanoi, Vietnam. The main age of the full sample
wasࡄ . 31.4% of participants were students; 26.7% were self-employed; 26.1% of them
were employees, and 7.3% of them were government.
The reason for choosing the citizens derived from the fact that they are main bene-
ficiaries from the outcome of traffic management measures and under the position of
consumers, they easily assess whether the solution are appropriate and how help them
to change their behaviors.

3.2. Methodology

Firstly, the weights of effectiveness criteria groups are estimated through the Ana-
lyti-cal Hierarchy Process (AHP). Seven-point scale was assigned to calculating the
weights of criteria and sub-criteria. In which, the first level indicates the equal im-
portance and the seventh level shows the extreme importance between one criteria
1040 A. M. Ngoc and K. V. Hung

over another. The comparison matrices of each respondent was individually analyzed
to examine the consistent ratios and to calculate the weights. And because they are
JHQHUDO SXEOLF DQ H[SHULHQFH¶V UDWH ZDV QRW FRQVLGHUHG LQ WKLV VWXG\ 7KH JURXS¶V
weights were calculated by simple geometric means of the individual weights.
In the next step, effectiveness of measures is assessed by estimating the impacts on
the transport system toward the traffic safety at school zones. Levels of impact are
qualitatively scaled (High, Medium, Low, and Non) and corresponding points (from 3
to 0 scale) are given for further evaluation.
)LQDOO\ EDVHG RQ WKH JLYHQ UDWHV RI PHDVXUH¶V LPSDFWV DQG WKH JLYHQ ZHLJKWV RI
goals and objectives, one can simply calculate the Effective Score (ESij) of measure
by below formula. Measure would be more effective if the Effectiveness Score is
higher, vice versa.
݆
‫ ݅ܵܧ‬ൌ σͶ݉ൌͳ ܹ‫ כ ݉ܥ‬ൣσ݇݊ൌͳ ܹܵ݉݊ ‫ ݊݉ܲܧ כ‬൧ (1)

Where
ESi: Effectiveness score of measure I; ESi •0HDVXUHLKDVKLJKOHYHORIHI-
fectiveness; 1.5 < ESi < 2.4: Measure j has medium level of effectiveness; and
ESi d 1.5: Measure j has low level of effectiveness
WCm: Weight of criterion m (m = 1 to 4)
WSmn: Weight of sub-criterion n, under criterion m
EPmnj: Effective point of measure j, on objective n, under goal m

4 Results
Table 5 shows aggregate opinions of participants that ensuring safe road infrastruc-
ture is the most important traffic management measure with a weight of 45.5%, safe
road user behavior ranked second (29.7%), while traffic safety management (S3) and
safety environment (S4) ranked third and fourth with 12.8% and 12% weight, respec-
tively.

Table 5. Weights of the assessment criteria

Weigh of criteria Criteria

45.5% Safe road infrastructure (S1)
29.75% Safe road user behavior (S2)
12.8% Traffic safety management (S3)
12% Safety environment (S4)

In accordance with the definition in above section, fifteen candidate measures are
selected for further analysis. After getting through the assessment, three traffic man-
agement measures have been selected to be in the first priority group, which includes
two engineering measures, and one encouragement measures. The other five measures
are assigned in the second, which includes four engineering measure, and one educa-
tion measure.
Evaluation of the Effects of Traffic Management … 1041

The assessment results indicate that the most recommended traffic management
improvements on engineering are facility installation, and crosswalk establishment.
Implementation of these two measures will help to prevent illegal movement. Look-
ing at engineering measures in the second priority group, one can recognize that the
next recommended improvement of infrastructure, the pedestrian oriented signal will
provide better physical accessibility, high priority right of way for pedestrian. The
second rank of ROW on unsignalized intersection and improving road surface provide
the good condition for travelling.

Table 6. Assessment of traffic management measures at school zones

No Measures Effectiveness ESi

S1 S2 S3 S4
1 Engineering
1.1 ROW on unsignalized intersection 1.75
1.2 Improvement of road surface 1.75
1.3 Installation of safe facilities 2.48
1.4 Effective pedestrian signal 1.85
1.5 Installation of crosswalk 2.41
1.6 Installation of ped. priority zone 1.59
1.7 Provision of ITS system 1.28
2 Education
2.1 Provision of maps and guidebook 1.19
2.2 Production of textbooks 1.10
2.3 Production of materials 1.54
2.4 Education by accident scenes 0.74
3 Enforcement
3.1 Reinforcement of regulation 1.24
3.2 Reinforcement of traffic crackdown 1.11
4 Encouragement
4.1 Active travel to and from school 1.46
4.2 Establishment of school zone 2.60

High Medium Low Non

One encouragement measure is selected in the first priority group and it is the most
important solution among measures. In practical, the school zone is popular applied
nationwide and is the effective way to reduce traffic accidents of children. The basis
of school zone comes from engineering because this measure also focusses on in-
stalling traffic safety facilities around the school zone, such as installation of traffic
signal and safety sign, installation of carriageway and road appurtenance for leading
speed reduction.
1042 A. M. Ngoc and K. V. Hung

5 Conclusion and Discussion

The study aims to assess the effect of traffic management in ensuring traffic safety for
young people and children. To obtain this objective, a multi-criteria assessment model
is applied to recommend the most suitable traffic management measures. In this mod-
el, four goals and ten objectives are selected for assessment. In which, the safe road
infrastructure was rated as the most expected improvement of traffic safety in Vi-
etnam (45.5% weight). The followings are safe road user behavior (29.7%), safe traf-
fic management (12.8%) and safe road environment (12%). It is necessary to empha-
size here that designated users of this model are not the traffic managers or a decision
PDNHUVZKRKDYHWRPDNHD³<HV´RU³1R´GHFLVLRQIRUDVSHFLILFVFKRRO7KHPRG-
HO¶VXVHUVVWD\LQD neutral position to assess and recommend traffic management
measures based on the specific conditions of Vietnam.
Based on the Effectiveness Score, the priority of traffic management measures are
recommended, three traffic management measures have been selected to be in the first
priority group, which include two engineering measures, and one encouragement
measure. The other five measures are assigned in the second; four are at engineering
measure, and one is at education. In general, these groups were built from a good mix
of engineering, education, and encouragement, which represent the traffic manage-
ment activities in all forms. Although the levels of impacts may be varied according
to specific conditions of schools, each measure of the list is highly recommended to
apply in every school in Vietnam. It is necessary to emphasize that the recommended
measures are qualitatively assessed and selected according to the existing transport
conditions of Vietnam. Therefore, the results of this study may be different from oth-
ers. The final emphasis here is that, the implementation of any traffic management
measure needs sufficient protection of traffic laws, effective and promptly enforce-
ment service, and good awareness of all stakeholders.

References

1. Albert Speer and Partner: Frankfurt Urban Integrated Traffic Management, Dezernat Bau and
Straßenverkehrsbehörde, Stadt Frankfurt am Main (1993)
2. Bowen, W. M.: AHP: Multiple Criteria Evaluation. Klosterman R. et al, Spreadsheet mod-els
for urban andregional analysis, New Brunswick: center for urban policy Research (1993).
3. Ibrahim, J.M.; Day, H.; El-Setouhy, M.: Road risk-perception and pedestrian injuruies
among students at Ain Shams University, Cairo, Egypt. J.Inj.Violence Res. (2012).
4. Juillard, C., Labinjo, M., Kobusingye, O. & Hyder, A.A.: Socioeconomic impact of road
traffic injuries in West Africa: exploratory data from Nigeria. Injury Prevention, BMJ, 16, pp
389-392 (2010).
5. Lassarre, S.: Analysis of progress in road safety in ten European countries. Accident anal-
ysis and prevention, 33, pp. 743-751 (2001)
6. Natulya, V.M. & Reich, M.R.: The neglected epidemic: road traffic injuries in developing
countries. BMJ, 324, pp. 1139-1141 (2002).
7. School Traffic Safety Survey. SUD (2016).
8. World report on child injury prevention,
http://www.who.int/violence_injury_prevention/child/en, last accessed 2019/03/21.
 Policy Implications from Traffic Accident Analysis: A
Study Case from Vietnam

An Minh Ngoc1 and Truong Thi My Thanh 2

1
University of Transport and Communications, 3 Lang Thuong, Dong Da, Hanoi, Vietnam
2University of Transport Technology, 54 TriӅu Khúc, Thanh Xuan, Ha Noi, Vietnam
anminhngoc@utc.edu.vn

Abstract. According to the statistics of Vietnam, the impact of traffic accidents

cost accounts for 2.5% of GDP. Thus, evaluating and analyzing the damages
caused by traffic accidents has become more significant during the last decades.
It is necessary to find out the effective approaches to identify the influencing
factors of traffic accidents. Among many regression accident-frequency models,
Classification and Regression Tree Analysis (CART), one of the most popular
data mining techniques and has been applied in many areas of business admin-
istration, medicine, industry, and engineering fields, etc., is invited in analyzing
traffic safety problems. The results showed that when there are motorcyclists
involved in accident, the probability of fatality increases up to 61%. Therefore,
searching the solutions for ensuring motorcycle safety should be placed on the
first priority in Vietnam.

Keywords: Road traffic accidents, fatality, injury, model

1 Introduction

Traffic safety policy aims to towards ensuring traffic safety of all traffic movements,
which are performed by the concerned transport facilities and services. Over the dec-
ades, countries focus on measures of reduction of possibility and occurrence of traffic
accidents as well as increasing survivability of people, vehicle, and reducing extent of
damage to facilities and environment in cases of accidents. While many countries gain
a lot of success by applying the appropriate strategies, for example Colombia reduced
the number of traffic death by 50% in ten years [14] by implementing an integrated
approach to road safety and mobility, or children deaths in Korea declined 90% when
school zones were built [14]. Sweden and the United Kingdom are also the typical
success cases over the world [12]. Other countries, especially low- and medium-
income countries, which have the mixed traffic flows, are facing with a great concerns
of traffic accidents. In these countries, the majority of deaths are among riders of
motorized two- and three-wheelers, they are victims of being ignored in the planning,
design and operation of safe systems. Road safety policy is typically derived from the

© Springer Nature Singapore Pte Ltd. 2020 1043

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_167
1044 A. M. Ngoc and T. M. Thanh

domain of many different disciplines, from traffic engineers, economists, psycholo-

gists, statisticians, and more recently urban planners.
An important issue for a better understanding of how to improve road safety is be-
longing to the choice of metric. The scale of measurement, whether by mode, road
type, or area can influence policy. For example, highlights from accident analysis
with a real-life data create various safety policies, from engineering and design poli-
cies, policies aimed at reducing crashes among young drivers, policies to reduce
drunk driving [12]. However, a placement of road safety policy within mixed traffic
flow, especially motorcycle-dominated traffic flow tends to be lacking.
This paper examines several issues of road traffic safety from a historic data per-
spective, explicitly considering how to improve road safety. This begins with a dis-
cussion of theoretical and practice models for understanding literature and the formu-
lation of a proposed model that unifies the previous researches. This is followed by an
experiential study associated with empirical estimations. Use of model results for
policy implication is then discussed. Conclusions examine how to improve the pro-
cess of analyzing safety policies with the hope that improvements in knowledge and
actual reductions in crash and severity outcomes can be achieved.

2 A Review of Practice Models

By utilizing real-life data, a various aspects of driver injuries and fatalities have been
explored through various models. For example, Regression, Logit, and Probit models
have been applied to forecast accident frequencies over specific factors, offering a
significant improvement to safety programming approaches [7,8]. These models re-
quire the assumptions among the variables. If these assumptions are violated, or ho-
moscedasticity of the residuals is violated, the erroneous analysis results would be
generated [9]. Thus, data mining is commonly applied to traffic accidents databases to
extract information and knowledge [9]. Data mining techniques are nonparametric
and do not require any previous assumption about the problem.
In the field of data mining, several techniques have been applied to accident analy-
sis. For example, neural networks were use d to analyze the severity of an accident
and the dependency relationships between the variables related to the accident [1].
Bayesian networks were adopted to research the prediction of the severity of acci-
dents [11]. Decision trees have been applied as well to analyze the severity of acci-
dents [2]. Decision tree provide a very useful model to identify the causes of accidents
because they can be easily interpreted and, most importantly, decision rules can be
easily extracted from them. These rules can be used by road safety analysts to identify
the main causes of accidents.
To build decision tree, a CART (Classification and Regression Tree) method is
used. The development of a CART model generally consists of three steps. The first
step is tree growing. The principle behind tree growing is to recursively partition the
WDUJHW YDULDEOH WR PD[LPL]H ³SXULW\´ LQ WKH WZR FKLOG QRGHV 7KHQ WKH GDWD LQ HDFK
child node are more homogeneous that those in the upper parent node. In the tree-
growing stage, splits are generated at each internal node of the tree. The Gini index
will be used to measure the contribution of each split in maximizing homogeneity.
Policy Implications from Traffic Accident Analysis … 1045

when GI is equal to zero, the node is pure (all the cases in the node have the same
class). The second step is tree pruning. Pruning is the mechanism of producing a se-
ries of simple trees by removing the important nodes. During the whole process of
pruning, the smaller trees are created gradually, forming into a pruned tree series. The
third step is to select the tree of an appropriate size from the pruned ones. The Chi-
square tests were used to establish whether an observed frequency distribution differs
from the theoretical distribution
More details about CART analysis and application can be found in [2,3].

3 Experimental Study

The dataset for this study contains traffic accident records from 2014 to 2017, a total
number of 1,337 cases. These data were collected from Hanoi Traffic Police Bureau.
It consists of information variables that are to be used as inputs: GULYHUV¶DJHJHQGHU
vehicle type, road segment, time, and cause, and the accident result variable that is to
be used as an output: severity level, three levels represent the severity: accident free,
injury, and fatality. The sample characteristics are represented in Table 1.
Table 1. Sample characteristics (N = 1,337)
Characteristics Statistics Characteristics Statistics
Section: 88.63% Male: 85.19%
Road segment Gender
Intersection: 11.37% Female: 14.81%)
6h00-17h59: 48.62% Violation: 84.29%
Time Cause
18h00 ± 5h59: 51.38% Non-violation: 15.63%
Motorcycle: 53.1% < 18 years old: 1.4%
Commercial vehicle: 36.9% 18-30 years old: 33.36%
Vehicle Age
Passenger car: 7.3% 30-45 years old: 44.05%
Other: 2.8% Above 45: 21.17%

For the tree growing, some indicators were grouped into smaller categories to
achieve a sufficient representation. Group 1 is road segment, including segment and
intersection. Group 2 relates to time features of which daytime is considered as one
variable. Variables in Group 3 are divided into two important categories (motorcycle,
FDU RWKHUV7KHYDULDEOH³DJH´LVUHGXFHGLQWRWZRLQ*URXS*URXSLVJHQGHU
and Group 6 describes traffic violation.
The first step of CART model is to determine the variables highly important to
predict the dependent. This is achieved by using the importance index.
Using this index, five variables having the greatest influence on accident severity
are detected (see Table 2). The most important variable causing fatality is vehicle type
(57.64% importance). The violation variable has 52.39% importance in the model.
The next important variable is age (42.08%). Gender variables are less important with
percentage of 39 %.
The splitting procedure is explained as follows. The initial splitting of node 1 is
based on the vehicle type. If the involved vehicle is motorcycle, CART puts it on the
left side, forming terminal node 1, otherwise it puts it on the right side as node 2. For
terminal node 2, CART predicts that 85.2 % accident free occurs. It means that the
1046 A. M. Ngoc and T. M. Thanh

probability of fatality when accidents occurred is very small when using car. Howev-
er, it is still possible to cause severity when people using motorcycles in node 1.
Hence, the variable of splitting accident frequency selects the location. If accident
occurs on road section, node 3 is produced on the left side, while node 4 is on the right
side to indicate the accident at intersection. Because the time impacts to severity of
accidents, node 3 and 4 are divided into node 7 and 8, node 9 and 10 respectively.
Considering the influence of gender on traffic accidents, node 9 and 10 are divided
step by step until the terminal nodes. Following this procedure, the whole tree of acci-
dent frequency prediction can be obtained.
Following the paths created between root node and each terminal node at the mod-
HOV EXLOW LQIRUPDWLYH ³LI-WKHQ´ UXOHV DUH H[WUDFWHG ,QWHUHVWLQJ UHODWLRQVKLSV RI YDULD
bles could be identified to explain the complication situations in traffic accidents. For
example, specific traffic accidents face the following rules:
x NODE 2: IF (Vehicle using is AUTOMOBILE) THEN (accident frequency =
ACCIDENT FREE).
x NODE 3: IF (Vehicle using is MOTORCYCLE and location is at SECTION)
THEN (accident frequency = FATALITY
x NODE 4: IF (Vehicle using is MOTORCYCLE and location is at
INTERSECTION) THEN (accident frequency = INJURE).
x NODE 7: IF (Vehicle using is MOTORCYCLE and location is at
INTERSECTION and time is NIGHT) THEN (accident frequency =
FATALITY).

4 Use of Results for Policy Implication

Results from experimental analysis show that riders meet much dangerous in accident
while using motorcycle. A probability of being dead reaches 61.1%. The probability
of getting dead even increases to 83% if riders use motorcycles at night. These results
show the fact that although motorcyclists are vulnerable as a result of being less pro-
tected than car occupants, infrastructure and vehicle design do not prioritize them.
Motorcyclists cannot afford to the safest vehicles and infrastructure, they are still
largely ignored in the planning, design and operation of roads. Similarly, while vehi-
cle standards and safety regulations have been developed for car occupants, they seem
missing for motorcycles.
Enacting and enforcing legislation on key risk factors including speed, drink-
driving, motorcycle helmet use, are already done in Vietnam. These greatly contribute
on preventing road traffic deaths recently. However, looking on the gaps of legislation
for protecting vulnerable road users, it is necessary to supplement more key risk fac-
tors to bring them in line with best practice.
Solution to eliminate human factor for motorcycle accident might be built from
frequently reviewing the driving ability and qualification of drivers to improve driver
safety. Policy maker should request a typical driver's license system for drivers who
are more likely to have traffic accident as well as a special inspection for drivers who
are in high-risk group.
Policy Implications from Traffic Accident Analysis … 1047

Table 2. Proposed strategies and detailed measures

Key Strategies Description
Relief of human factor for
motorcycle accidents
Motorcycle Safety
Relief of vehicle factor for
motorcycle accident
ITS based traffic safety
Safe Road Infra- enhancement
structure
Road safety facility im-
provement
Actions/Measures
Reinforcement of aptitude test and
safe driving education for motorcy-
clists in high risk group
Enhancement of helmet quality
Life-cycle management of motorcycle
Systemic management of motorcycle
performance
Traffic violation notification system
Improvement of automatic traffic
violation enforcement system
Road safety facilities installation
Visibility improvement under poor
condition (night)

Enforce people wearing quality helmets is urgent regulation. Currently, 80% of the
helmets used in Vietnam failed to meet the quality standard, particularly in impact
absorption performance (according to 2012 WHO investigation).
Vietnam is also lacking the legal performance inspection of motorcycle; motorcy-
cle usage only depends on rider's discretion. If a systematic performance management
procedure is introduced, the traffic safety level hopes to be significantly improved.
Furthermore, there is a lack of enforcement personnel and a strong opposition to
the crackdown, which has not led to substantial improvements in traffic safety aware-
ness. Therefore, it is necessary to strengthen the 'guidance' along with the crackdown.
In other words, it is required to utilize road facility with high technology to detect
traffic violation activities. Then, it alarms to violator and raise awareness about traffic
violation activities to citizens.
Install safety facilities on the roads is necessary to prevent traffic accidents and to
eliminate defects in road environmental factors.
To make safer roads, protective fence is needed to prevent the vehicle departing
from the normal driving route during driving to move to the opposite vehicle or the
sidewalk. Delineation facilities need to be installed at the sides of the roads to specify
the edge and the shape of the roads, guiding GULYHU¶VVLJKWduring day and night time.

5 Conclusion

Based on CART method, the key factors influencing traffic accidents were identified.
The most important variables for the accident evaluation are vehicle type, cause,
age, location, and time. Several interesting findings can be summed up. More than
50% of traffic accident has motorcycles involved. When there are motorcyclists in-
volved in accident, the probability of fatality increases up to 61%. Accidents with the
involvement of cars tend to have minor consequences, accident free. There is a higher
probability of fatality in accidents that caused by violated male drivers of motorcycle
in road segment are identified. The policy implication including enforcement of risky
driving behaviors, installation of safety facilities, and introduction of a motorcycle
1048 A. M. Ngoc and T. M. Thanh

safety system is recommended. Through multiple policy implications, road traffic

crashes in Vietnam are expected to reduce in a relatively short time period.

References
1. Abdel-Aty, M.A., Chen, C.L., and Radwan, E., (1999). Using conditional probability to
find driver age effect in crashes. Journal of Transportation Engineering, 125, 502-507.
2. Breiman L, Friedman JH, Olshen RA, Stone CJ, (1998). Classification and regression
trees. London: Chapman & Hall/CRC.
3. Chang Ly, Chen WC (2005). Data mining of tree-based models to analyze freeway acci-
dent frequency. J Safety Res. 2005; 36(4): 365-375
4. Evans, L., (1991). Older-GULYHUV¶ULVNVWRWKHPVHOYHVDQGWRRWKHUURDGXVHUV7UDQVSRUWD
tion Research Record: Journal of the Transportation Research Board, 1325, 34-41.
5. Kashani, A. and Moheymany, A. (2011). Analysis of the traffic injury severity on two-
lane, two-way rural roads based on classification tree models. Safety Science 49, pp. 1314-
1320
6. Kashani, A.T.; Shariat-Mohaymany, A.; Ranjbari, A. (2012). A Data Mining Approach to
Identify Key Factors of Traffic Injury Severity. PROMET Traffic Transp. 2012, 23, 11±17.
7. Kuhnert, P.M.; Do, K.A.; McClure, R. (2000). Combining non-parametric models with lo-
gistic regression: An application to motor vehicle injury data. Comput. Stat. Data Anal.
2000, 34, 371±386
8. Mujalli, R.O.; de Oña, J. (2013). Injury severity models for motor vehicle accidents: A re-
view. Proc. Inst. Civ. Eng. Transp. 2013, 166, 255±270.
9. Pakgohar, A.; Tabrizi, R.S.; Khalili, M.; Esmaeili, A. (2011). The role of human factor in
incidence and severity of road crashes based on the CART and LR regression: A data min-
ing approach. Procedia Comput. Sci. 2011, 3, 764±769.
10. Savolainen, P.T.; Mannering, F.L.; Lord, D.; Quddus, M.A.(2011). The statistical analysis
of highway crash-injury severities: A review and assessment of methodological alterna-
tives. Accid. Anal. Prev. 2011, 43, 1666±1676.
11. The Korea Transport Institute KOTI (2013). Korea¶s best practices in the transport sector.
Transport safety policy in Korea. ISBN 978-89-5503-559-9
12. Washington SP, Karlaftis MG, Mannering Fl (2003). Statistical and econometric methods
for transportation data analysis, NewYork: Chapman and Hall/CRC; 2003.
13. Williams, A.F., and Shabanova, V.I., (2003). Responsibility of drivers by age and gender,
for motor-vehicle crash deaths. Journal of Safety Research, 34, 527-531.
14. World Health Organization WHO (2018). Global status report on road safety 2018. Li-
cence: CC BY-NC-SA 3.0 IGO).
 Analysis of Tourist Travel Behavior and
Recommendation for Active Transport Encouragement
Strategies, the Case of Hue City

Truong Thi My Thanh1, Dong Van Dao1, Hai-Bang Ly1

1
University of Transport Technology, 54 Trieu Khuc, Thanh Xuan, Hanoi, Vietnam
thanhttm@utt.edu.vn

Abstract. Over the past years, several transport encouragement strategies have
been implemented to stimulate the mode shift from motorized transport to walk-
ing and cycling from international studies. However, information concerning
the effectiveness of such strategies in tourism cities in Vietnam is still limited.
This study aims at recommending strategies to encourage active transport usage
of tourists through the analysis of travel behavior. The study firstly reviewed
literature to propose travel modal shift measures that specified for local travel
characteristics. Secondly, travel interview survey was conducted in Hue city to
understand the current travel status of 259 tourists and the possible encourage-
ment measures on mode shifting to walking and cycling. Finally, the study
showed the core encouragement measures concerning the possibility of mode
shift under tourist perspective. Combination of more than one measures re-
vealed the most possible effectiveness and applicability to implement mode-
shift strategies.

Keywords: Tourist Travel Behavior, Active Transport, Mode Shift, Transport

Demand Management, Tourism Cities.

1 Introduction

Locating at the center of Vietnam with a population of 450.000 inhabitants, Hue at-
tract 3.800.000 tourists in 2017 which increasing 17% after one year. Because of the
high speed increasing of tourists, traffic situation in Hue city is more congested dur-
ing recent years. In many occasions, this city is fully covered by tourist buses, taxi
and mini-vans. High level of motorization has been bringing negative impacts on
transport system and quality of life.
$FUXFLDOTXHVWLRQLV³+RZWRPDQDJHWKHWUDIILFVLWXDWLRQLQsuch typical tourism
FLW\DV+XHWRDFKLHYHVXVWDLQDEOHGHYHORSPHQW"´DQGDFWLYHWUDQVSRUWKDVEHHQ con-
sidered as a potential solution.
Active transport includes all forms of travel that do not rely on an engine or motor
for movement [1]. This includes walking, cycling, small-wheeled transport (skates,
skateboards, push scooters, cyclo, pedicab) and wheelchair. These modes of transport
can provide both recreation and mobility.

© Springer Nature Singapore Pte Ltd. 2020 1049

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_168
1050 T. T. M. Thanh et al.

The study aims at investigating the travel behavior of tourists and the important
degree of encouragement measures to shift from motorized to active transport in the
case of tourism city in Vietnam. The study firstly reviewed international studies to
recommend a list of encouragement measures for travel mode-shift that specified for
local travel characteristics. Secondly, the travel interview survey was conducted in
Hue city to understand the current travel status of 259 tourists and the possible en-
couragement measures on mode-shifting from motorized vehicles to walking and
cycling. The important degree of encouragement measures was highlighted from tour-
LVW¶V SHUVSHFWLYH 5HFRPPHQGDWLRQV IRU DFWLYH WUDQVSRUW HQFRXUDJHPHQW PHDVXUHV
were finally proposed.

2 Literature Review

From literature review, encouragement measures are mapped aiming to induce a

mode shift from private modes to active modes of travel (walking or cycling).
Measures were categorized in work-place-based, architectural and urbanistic adjust-
ments, population-wide, and bicycle-renting-system groups.
Many studies showed positive effects concerning a mode shift from car use to ac-
tive transport. The tools used were either legal [2], economic (subsidy and reward
system [3], penalty [4]), communicative (written materials, behavioral tools) [5] or
physical tools (providing bicycles, providing better bicycle facilities at work, adjust-
ment of the environment) [1]. In many cases, a combination of measures was used.
Based on the reviewed mode-shift interventions, this research proposes measures
that specified for local travel characteristics which categorized into (i) Economic
tools, including subsidy (free bicycle parking), reward system (discounted vouchers),
and penalty (increasing parking charges, pay a toll fee); (ii) Communicative tools,
including mass media campaigns (increase awareness and knowledge), written mate-
rials (regulations from local authorities); and (iii) Physical tools, including better bi-
cycle facilities (separated lanes for bicycles), adjustment of the environment (develop
a new trail, bicycle parking locates at good access area). These encouragement
measures were included in the travel interview survey to understand the degree of
importance under tourist perception.

3 Data Collection

$ VXUYH\ RQ WRXULVW¶V travel behavior was conducted in September 2018. A total of
259 people was randomly selected at different tourist attraction areas, including 109
foreign and 150 domestic tourists. They were directly interviewed about their mobili-
ty characteristics (transport mode usage, walking distance, cycling distance) and the
willingness to shift from motorized transport (motorcycle, taxi, and car) to active
transport mode (walking and cycling). They were also asked for the evaluation of the
important level of walking and cycling encouragement measures.
Mobility data are significant to understand the specific characteristics of tourists
during their visiting at a tourism city. The encouragement measures for travel mode-
Analysis of Tourist Travel Behavior … 1051

shift, therefore, should be proposed to fit with such features to ensure the effective-
ness and applicability of measures.
The demographic of interviewed tourists is presented in Table 1.

Table 1. The demographic of interviewed tourists

Variables Freq. (%) Variables Freq. (%)

Tourists Visiting Time

Domestic 150 57.9 First time 193 74.5
Foreign 109 42.1 2-3 times 38 14.7
Gender 4-5 times 24 9.3
Female 137 52.9 >5 times 4 1.5
Male 122 47.1
Age Night of Stay
<18 4 1.5 a night 130 50.2
18-24 79 30.5 2-3 nights 103 39.8
25-35 149 57.5 4-5 nights 20 7.7
36-50 22 8.5 > 5 nights 6 2.3
51-60+ 5 1.9

4 Results

4.1 Transport Mode Usage

To understand the current status of transport mode usage of tourists, nine options of
travel vehicles were proposed for tourists to select that best fit with their visiting (see
Fig.1). Based on the travel interview survey, it is important to highlight that motor-
ized vehicles (taxi, car, and motorcycle) are still popularly used as tourism transport,
whereas walking and cycling are seldomly used. For instance, 65% of domestic tour-
ists selected taxi comparing with 38% of foreign tourists did. It is also interesting to
see that foreign tourists more frequently used motorcycle (24%) than domestic tour-
ists (12%). Foreign tourists answered that driving motorcycles provided them better
opportunities to explore local culture.
Importantly, foreign and domestic tourists are dissimilar in transport mode usage
during the time they stay in Hue city. Local travelers were more dependent on motor-
ized vehicles (taxi and car) while foreign visitors utilized walking and cycling more
frequently (with 12% and 19% respectively). A major concern of tourists was the
convenience of bike-services, for instance, the available of bicycle renting stations or
bike-sharing system at visiting areas. This finding is meaningful for transport man-
agement authorities to plan for green transport system in Hue, especially serving for
tourists.
Moreover, understanding the travel behavior of tourists is essential to propose
mode-shift measures applicable for their specific mobility. Based on the data analysis,
walking and cycling encouragement might have higher impact on foreign tourists.
1052 T. T. M. Thanh et al.

Therefore, the willingness of walking and cycling distance of each tourist group was
investigated in the next section.

Transport Mode Usage of Tourists

(9) Walking
(8) Dragon boat
(7) Taxi
(6) Car as passenger
(5) Car as driver
(4) Motorcycle taxi
(3) Motorcycle as passenger
(2) Motorcycle as driver
(1) Bicycle
0% 10% 20% 30% 40% 50% 60% 70%
(n=150) (n=109)
Domestic Tourists Foreign Tourists

Fig.1. Transport mode usage of tourists

4.2 Walking and Cycling Distance
The willingness to walk and go cycling was significantly different between foreign
and domestic tourists. Among 109 foreign tourists interviewed, 77% of them were
willing to walk more than 1.000m, whereas only 51% of domestic tourists under this
survey might walk in the same distance (see Fig.2).

Willingness of Walking Distance

> 1000m
500-1000m
200-500m
100-200m
0 10 20 30 40 50 60 70 80 90

Foreign Tourists Domestic Tourists

(n=109) (n=150)

Fig.2. Willingness of walking distance of tourists

They are also different at the willingness to go cycling. Based on the travel behav-
ior survey, 82% of foreign tourists were willing to cycle more than 10.000m for sight-
seeing trip compared with only 27% of domestic tourists. The most convenient cycle
distance of domestic tourists is 5.000-8.000m (52%).
The willingness to cycle is an important evidence to plan for the bicycle-supply
services to encourage bicycle shifting. The location arrangement of renting stations,
Analysis of Tourist Travel Behavior … 1053

for instance, should serve both domestic and foreign tourists and best fit with the will-
ingness to cycle. There is a big advantage to encourage walking and cycling in Hue
city due to a good provision of green spaces and harmony traffic density.

Willingness of Cycling Distance

> 10.000m
8.000-10.000m
5.000-8.000m
2.000-5.000m
0 10 20 30 40 50 60 70 80 90
(n=109) (n=150)
Foreign Tourists Domestic Tourists

Fig.3. Willingness of cycling distance of tourists

4.3 Tourist Perspective on Walking and Cycling Encouragement Measures
Walking and cycling are part of the tourist experience at visiting areas and also signif-
icant elements of sustainable mobility. To encourage tourists to walk and cycle more
frequently, measures are recommended based on international review. Questionnaires
are utilized to ask tourists for their assessment of walking and cycling encouragement
measures.
Results showed that safety and securities of the sidewalk, the quality of walking
and cycling infrastructure were the most important indicators (76% and 69% tourists
selected respectively). It is also interesting that limitation of shopping on the sidewalk
did not receive strong supports from tourists, especially the foreign ones (see Fig.4).
They prefer observing the real life of local people which can be seen as the authentic
culture of Hue province.

Important Level of Walking Encouragement Measures

6. Provision of benches and trees along the…

4. Limitation of parking on the sidewalk

3. Limitation of shopping on the sidewalk
2. Expansion of sidewalks, improvement of…
1. Construction of separated lane for walking
0 10 20 30 40 50 60 70 80 90 100

Definitely NOT important NOT important Moderate Important Very Important

Fig.4. The important level of walking encouragement measures

1054 T. T. M. Thanh et al.

To encourage cycling, having separated lanes for bicycles and no parking fee were
evaluated as the most important indicators to promote cycling (see Fig.5). Besides, the
provision of bicycle renting services gets strong support from tourists in order to en-
courage them to bike more frequently.

Important Level of Cycling Encouragement Measures

(5) Provision of bicycle renting services

(4) Bicycle parking is priority in shopping areas…
(3) Bicycles are freely to access the parks…
(2) Parking is free for bicycles
(1) Having separated lanes for bicycles on the…
0 10 20 30 40 50 60 70 80 90 100

Definitely NOT important NOT important Moderate Important Very Important

Fig.5. The important level of cycling encouragement measures

5 Conclusions
The study provided empirical evidences for effective strategies to encourage tourist
shifting from motorized to active transport, based on the travel interview survey of
259 foreign and domestic tourists in Hue city. Even the active transport is not fre-
quently used by the tourists at current status, the potential to promote such kind of
sustainable transport is especially high. Sufficient encouragement strategies are highly
important in contributing to green and sustainable cities and promoting tourism-based
economics.
The evidences reveal that improvement of infrastructure quality, safety and securi-
ties, and the development of bicycle renting service are the keys to promote active
transport in such tourism city as Hue. In combination with the green environment
along Huong River, walking and cycling have high potential to be the dominant
modes in Hue city. The outputs of this empirical research are also useful to other tour-
ism cities in Vietnam, for instance, Da Nang, Hoi An, Nha Trang, Can Tho.

References

[1] C. E. Scheepers, G. C. W. Wendel-Vos, J. M. den Broeder, E. E. M. M. van Kempen,

3 - 9 YDQ :HVHPDHO DQG $ - 6FKXLW ³6KLIWLQJ IURP FDU WR DFWLYH WUDQVSRUW $
V\VWHPDWLF UHYLHZ RI WKH HIIHFWLYHQHVV RI LQWHUYHQWLRQV´ Transp. Res. Part A Policy
Pract., 2014.
[2] 60HODQG³*HQHUDOLVHGDQd advanced urban debiting innovations: the Gaudi project.
2YHUYLHZ´Traffic Eng. Control, 1995.
[3] / 0 :HQ 1 2UU - %LQGRQ DQG & 5LVVHO ³3URPRWLQJ DFWLYH WUDQVSRUW LQ D
ZRUNSODFHVHWWLQJ(YDOXDWLRQRIDSLORWVWXG\LQ$XVWUDOLD´ Health Promot. Int., vol.
20, no. 2, pp. 123±133, 2005.
Analysis of Tourist Travel Behavior … 1055

[4] 5 %URFNPDQ DQG . 5 )R[ ³3K\VLFDO DFWLYLW\ E\ VWHDOWK" 7KH SRWHQWLDO KHDOWK
EHQHILWVRIDZRUNSODFHWUDQVSRUWSODQ´ Public Health, vol. 125, no. 4, pp. 210±216,
2011.
[5] 5 $OFRWW DQG 0 0 'H&LQGLV ³&OHDQ $LU )RUFH &DPSDLJQ -1990: Programs,
$WWLWXGHVDQG&RPPXWH%HKDYLRU&KDQJHV´Transp. Res. Rec., no. 1321, pp. 34±44,
1991.


 
 






 




  
  


 
 



   ! " !
#
    $    !
!    ! $%
      &# '  !! (&

  (& )
   $
  !  % ! %
    "      $ &#
 
 $     &       $    * 
    #
     !   %
    !$! !    !   %
 + ! " ! #

 " %  #

!



             $   !
 " ,
-#'$$! !    ! ! 
$  )../$0.1! !   $23#/1#4! %
" !    ! !
     5  &# 6 !   "    $%
   $ #7 !
     ! 
  #    %8
9+!$:7    ; $ < #

   &   $           &  !  
 #:  !   ).
3   %
     ;     ,)-        
   *           % 8  9+!$
=   !  
 > $  # :  ?  ! 
 "  @   !  %
!  &#
  "%
!   $ = A 
 ##>   $ B
7  %
#

© Springer Nature Singapore Pte Ltd. 2020 1057

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_169
1058 T. N. To et al.

"
#



"
#

:  !
%)     ('<9 CDD $     
  + !#'('<9

 
$!  B
7 4  ?$
 !$  ('<9)
  &  %
     ('<9 CDD#
      $  ! 
 B'      !#' %
!  $  $  $#'
   ! !!=).
3>!  
 $$ *   !    #
'  !  ).
3)../!= 
*  >#

""



B%
7   !$ & 
$  #'  
         ! ! ! !  
    $#


$ #

)
 ! !$  
   '(,0-! *#
 ! !$ %
    '(! *   !#
   % 8  +%
!$:7  ; % <   #

 +!    !         #
 
  =9(>! ).
3$$!
 A #
9(  !$   
  E      =  >  !   

F)...G$ ! 

FH.. 
 
F
....#
The impacts of urbanization on urban flooding … 1059

%
&
 

%

'


 $  $ *  )..I
)../!#
! $    $ %
  <:B 
#
   %
=4>% =>   !&=1>
$ !#
!   %
&    " = 
>#

   J
 #

  =)..I>  =)../>

     
 4  4
!&=1> !&=1>
 < .#I/ .#/K
 K#H21 .#2K0 .#/2) 0#HI1
: .#II .#/K2 )#./1 .#/K3 .#/3) )#KH1
B 
 .#IH .#/I0 .#/
1 .#22. .#/IH )#HK1


$"D $=> < G= >: G
=>B 
 #

C)  $ !   $#'

     !  $      $  # '%
!  !& $
  #

%"
!
 
  
 

 ! $)../).
3 $0#
'        =
%8+!$:7  ; % <  >
  #
!$  &!   ).
3
!  )../#'  ! !%
 !! $  
 !   #
1060 T. N. To et al.


$%
 $ !  9()../).
3#

"
   $)../).

3
C ! :=>

  !E 
.J

I2H

J) 0/H/
)J0 K3K
L0
.2)

 I)/.

 $ ! )/
0

 )     !# ! )..3

     $   !   0  
.2)=>#

$!   I)/.=>).
3$
$    !   )/
0=>#
     " 
5 5 !#

%%
!
 (
 

   ! * $ &).
3
 $K#'    
     $  $     *   #
 ! $  %
 & !  *  
 !$  #
+ 
 ! 
 0#
The impacts of urbanization on urban flooding … 1061


$ )
 $ !  $
$ *  #

%
   $$ *  #

' =>
C !=>
  
 %
!  !
.J
 0
32 333

J) 3K3
K0
)J0

 K


C   0       !   $

 !         $    !#

   !&       # '    %
!!   %
     & #
 $%
 !     ; 
  ).
3,)-#
CH=> $  ! $$ 8
%$#
8%$$ $
 #
&$   $#:
 !    $    $  $  
 .#2$ $ .#H#' 
    !     # '  
      $           
! $             $#
     .#K  
 
!   #
C :7     !   =C
H >#
 $              $ 
1062 T. N. To et al.

 #
! 
#H!  #'
 $$  ! 
 .#H#
$ $$   
#H#

           ; %   <      
=CH>#6 $  ! .#K#


$*
  ! $$  #=>8J
+!$= >:7 G=> ; J <  #

)
 

'    (& C        "   $
 !!           #

  ! $     "      !!%
  $ #
!! $  
A  $ !#
 !
   0  !        !
.2) #

     * !  
!  #' " :7  %
$  $ 
#H#C %8+%
! ; J <  .#2.#H!#

 +


     C    
    !  
 ! * 7).
I%M.)%)/#

&  

#

# #   4# $  <#
 N:    +!     &
 O


   #)/ #0
!!#
)0J
0K).
0#
)#
N4!   C          ;  
 O).
3#
0#
'( N         !   ( # 4
!  O).
)#
 Factors influencing the choice of Bike Sharing System:
An investigation of visitors and local people in Vietnam

NGUYEN Thanh Tu1 and LE Thu Huyen2

1,2
Faculty of Transport Economics,
University of Transport and Communications, Hanoi, Vietnam.
1
Telephone +84 989-194-794, Fax. +84 243-766-4053
1
thanhtunguyen.utc@gmail.com

Abstract. Bike Sharing System (BSS) has become a popular travel mode to sat-
isfy urban residents travel in many cities in the world. This paper examines the
influencing factors of BSS choice of 2 group: visitors and local people. For tour-
ists, they are the person who has used the bicycle in the past; have travel duration
from 3 to 7 days and most of them come from the US (for foreign tourists). Be-
sides, the visitors who use the car/coach during their trip are negative for BSS.
The residents who are students or individual business, have habit to use bike,
undertake 3-4 trips per day with the short travel distance and have lower income,
are more likely to use the BSS. The identification of these influencing factors
plays an important role in the deployment of bike-sharing systems in the big cities
in Vietnam. Several policy recommendations are proposed based on these results
for government and BSS investor in order to have the effective BSS in Vietnam.

Keywords: Bike-sharing system, Factors, Travel behavior, Mode choice,

1 Introduction

Bike sharing system (BSS) is a service in which bicycles are provided for individual
using in a short distance or in a period of time. Many bike share systems allow people
to rent a bike from a "dock" and return it at another dock belonging to the same system.
Docks are special bike racks that lock the bike, and only release it by computer control.
The user enters payment information, and the computer unlocks a bike. In many sys-
tems, smartphone mapping apps show nearby available bikes and open docks. BSS is
popular in Europe, USA and China. Until now, more than 800 cities around the world
operate a Bike Sharing System (BSS).

© Springer Nature Singapore Pte Ltd. 2020 1063

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_170
1064 T. T. Nguyen and T. H. Le

Nowadays, many cities in Vietnam have been deeply troubled with traffic congestion
as well as environment pollution issues. The city authority has been currently trying to
solve these issues with a large public transportation policy. However, the outcomes
may not be obtained soon and the city should think of other ways to complete its
transport policy. Thus, the main questions to be answered here are: Could the benefits
of a Bike Sharing System help the Vietnamese big cities addressing its mobility issues
and how to implement it? The purpose of this study is to understand the factors influ-
encing the possibility to choose BSS of two main groups: visitors and local people. The
results of this study could be useful not only for the businesses but also for the local
government to plan their services more effectively.
Exploring the bike-sharing user characteristics is necessary to understand their travel
demands and improve their loyalty. The socio-economics and travel behavior are im-
portant factors related to the choice of BSS. In term of the socio-economics factors,
gender is one of the important one, the majority of the literature has confirmed that
male is more prevalent than female users. For instance, the proportion of male bike
sharing users in Melbourne is 76.6%, in Brisbane is 59.8% [2], in Montreal is 58% [3]
and in Washington is 57% [4]. In terms of age and education level, young and educated
person are more likely to use BSS than other groups. The authors of [5] found that
86.5% of users were 20±50 years old and that 91,2% of BSS users at a greater level
than higher school in a bike sharing study in Beijing. The personal income is also an
impact factor. Dill [8] found that the people with a higher income were more willing to
ride bicycles. This finding was confirmed by [1] in their research from England and
Wales. In term of travel behavior, travel distance is the one of the important factor
LQIOXHQFLQJWKH%66¶FKRLFH7KHVWXG\RI>6] have shown that the trip distance from 1-
4km is more suitable to accept the BSS. Many people will choose other public transport
if their travel distance is more than 3.5 km [7]. However, when the travel purposes are
for entertainment or sport, the travel distance by BSS will be longer, research confirmed
by [8@%HVLGHVVHYHUDOVWXGLHVVKRZHGWKDWWKHWUDYHOSXUSRVHV³JRIRUZRUN´DQG³JR
WRVFKRRO´DUHRIWHQFKRVHQE\ELNHVKDULQJXVHUV%XWLQWKHtourist city, bike sharing
is the choice of a lot of visitors. Another factors influence the bike sharing is the user
preferences. Among preference factors, convenience is considered to be the main cause
of the use of public bicycles [9]. The study of [3] showed that people tend to use bike
if they live within a 500m from a bike station or near the commercial center.
From the above review, there is abundant researches on the travel behaviors of public
bicycle users from over the world. In Vietnam currently, there are no BSS in any city,
except for some small system running as pilot projects inside some beach cities. From
our knowledge, there is little research available regarding BSS in Vietnam. In order to
provide users with the good services and with great user satisfaction, it is necessary to
understand the travel characteristics and influential factors affecting travel patterns in
BSS systems.
Factors influencing the choice of Bike Sharing System … 1065

2 Data collection and descriptive

The survey was conducted with 3 types of respondents: foreign visitors, domestic visi-
tors and local people. In which, the visitor interview was conducted in Hanoi ± the
tourist city and the last was conducted in Haiphong, one of the big cities in the North
of Vietnam. There were 3 parts in the questionnaire: (1) Personal information; (2)
Travel information; (3) Bike user preferences. In the questionnaire, tKHTXHVWLRQ³:LOO
\RXXVHELNHVKDULQJV\VWHP´ZDVFKRVHQDVDGHSHQGHQWYDULDEOHRIWKHORJLWPRGHO
which analyses the IDFWRUVLQIOXHQFLQJWKH%66¶VFKRLFH

2.1 Characteristics of visitors

The data showed that WKHPDOHLQWHUYLHZHGUDWHVDUHDOZD\VKLJKHUWKDQZRPHQ¶VRQH
With foreign visitors, there are more than 86% of visitors from Europe and America.
The foreign visitors stay longer than domestic visitors, up to 49% of foreigners stay in
Hanoi from 3-7 days, while most of domestic visitors only stay under 3 days. During
the stay, foreign visitors use taxi or walk (totaling more than 60%) for their trips, while
domestic visitors choose buses or motorcycles (respectively 34,6% and 20.1%). Re-
garding the use of bicycles in the past, according to many studies, the possibility of bike
sharing use is higher for those who have biking habit. So in this study we have asked
this question. Accordingly, foreigners use bicycles more usually than domestic visitors.
The question about visitor income is not designed for this survey because the income
levels vary widely among countries and continents.

2.2 Characteristics of local people

There were 362 people participated in the survey. The majority of respondents are male
(accounting for 55.8%), at the age of 18 to 35 years old. In terms of occupation, the
percentage of individual business is quite high (nearly 27%). Regarding personal in-
come, most of them are on average about 5 million VND per month.
For travel characteristics, nearly 60% of respondents said that they used to ride bi-
cycles (and now have changed to motorbikes or cars) and over 22% said that they have
never used bicycle before. More than 30% of respondents have a total daily travel dis-
tance between 12 and 20 km for 5-6 trips per day. It means that the average trip distance
is short, about 3.7km and the total daily distance is 19.1km.

3 Model Results and Discussion

3.1 Logit model of visitors

Based on the logit model results, for the foreign visitors, the gender, stay duration,
travel purposes, travelling frequency and mode of transport do not affect the choice of
BSS. However, the results show that the frequency of bicycle use in the past has a
significant impact: those who have used bicycles regularly in the past will be easier to
1066 T. T. Nguyen and T. H. Le

choose BSS than the others. The rate of BSS choice of regular bike users is 4.7 times
higher than those who use only a few times a year. In addition, passengers from Latin
America are more significant to use BSS than people from other continents.

Table 1. Result of logit model for foreign visitors

Foreign visitors Domestic visitors

Variable Description Estimate Variable Description Estimate

Daily** 0.553 < 3 days 0.012

Fre- Duration 3-7 days** 0.325
quency Few 0.102
time/week of stay 7 ± 30 days* 0.256
of bike More than 30 days** -0.544
using Few time/ 0.346
month** Bus** 0.626
Europe -0.195 Bike 0.166
Residen- Mode of
Moto 0.360
tial loca- America** 0.587 transport
Car/Coach* -0.532
tion Other -0.174 Taxi 0.338
Walk 0.145
Frequency Daily** 0.598
of bike us- Few times per week* 0.381
ing in the Few times per month -0.214
past Few time per year 0.189
**: high significant; *: significant; variable without (*) is not significant

For domestic visitors, the following factors affect the possibility to choose BSS: the
frequency of bike using in the past, duration of stay and mode of transport during their
stay. Accordingly, those who have a habit of using bicycles regularly in the past will
have a higher probability of BSS choice than others, the choice of these people is 4.8
times higher than those who have never used a bicycle. In addition, those with a stay of
3-7 days more significant of BSS choice than those with a short stay or stay longer than
1 week. Those who use cars or shuttles during the trip will not choose BSS.

3.2 Logit model for local people

As the same as results above, the probability of choosing BSS of local people is
related to the habit of bicycle using: those often use bike have a higher probability of
BSS choice than the others. In particular, the probability of these people is 22 times
higher in those who have never used a bicycle.
In term of personal occupation, students and individual business are more significant
to choose BSS. Regarding income: The high the income (10 to 15 million Vietnam
dong per month) is negative to BSS using with a high significant level. Concerning the
travel distance, the longer the distance traveled, the less of BSS choice and those have
Factors influencing the choice of Bike Sharing System … 1067

3-4 trips per day is positive to use BSS with high significant level. This result is rea-
sonable because public bicycle is suitable for short distance trips. The result is presented
in the table 2.

Table 2. Result of logit model for local people

Parameter Description Estimate
Individual business** 0.752
Occupation
Student/pupil* 1.479
No income** 2.336
Income (VND/month)
10-15 mil.* -14.179
Still to use bike** 17.543
Frequency of bike using in the past
Used to use bike* -0.419
Distance travelled per day 20-30km** -0.719
Number of trip travelled per day 3-4 trip** 0.705
**: high significant; *: significant, variables non-significant were deleted.

4 Conclusion

The objective of this study is to explore the factors that affect the choice of bike sharing
in Hanoi and Haiphong in Vietnam. The interview surveys were conducted with 1549
visitors in Hanoi- the capital, which is also a major tourist city - and 362 people in
Haiphong - a growing urban area with high travel demand. This study was carried out
within the project of Public Bike Feasibility Study in two cities. However, the goals of
BSS in these cities are different. In Hanoi, the project goal is to develop the public
bicycle system for visitors and the one in Haiphong is to support the development of
the urban public transport system.
A logit model was established to find the factors influencing the choice of BSS. The
results show that the habit of using bicycles, the preferable transport mode during the
traveling trip, the travelling duration and residential location (of foreign visitors) are
significant to the choice. The result implies that the visitors who travel in group - will
be less likely to use public bikes. This will be a suggestion to investors to set up the
bike stations in areas where there are many free (non-group) visitors and at the main
tourist attraction places.
Among the local people, students and individual business are positive to use bike at
the high significant level. In additions, public bicycles will be suitable for people with
short distance trips and undertakes 3-4 trips per day. This finding is similar to the study
by [7], where users were found to be more likely to use a public bike when the travel
distance was within 1.5±3.5 km and the majority of travelers would choose other public
transports if the travel distance was more than 3.5 km. This finding is suitable for short-
distance trip. due to limitations of time and effort [10].
1068 T. T. Nguyen and T. H. Le

This will be a suggestion for local authorities to start implementing public bicycles
in the city center area where there is a high density of land use.
In the study, we also interviewed respondents about their expectations of public bike.
Accordingly, the visitors interested in convenience, which means that bicycle stations
are close to each other for easy rental. The domestic tourists would like to have a rea-
sonable rental fees. Meanwhile, the local people hope that the bike shared systems will
be easy to access. Other factors such as vehicle types, bike color of bike accessories
are not important. Therefore, the investors should consider these factors when imple-
menting the project to mitigate risks.
In additions, the feasibility of a BSS in Vietnamese cities should be more studied
because now the motorcycles account for more than 80% of mode share and motorcy-
cles are more flexible and faster than bikes. Therefore, it is difficult to convince people
to change the mode of transport from motorbikes to BSS. Moreover, the current
transport infrastructure in Vietnam urban is not really safe for cyclists.
It is also recommended to conduct further research. Firstly, the BSS should be stud-
ied relating to weather and climate variables, such as rainy days, intense-sun ray days,
summer and winter. Secondly, the health condition of users (BMI indicators, etc.) and
transport infrastructure for cyclist could influence on the BSS choice but have not been
considered in this paper.

References
1. Parkin. J.; Mark. W.; Matthew. P. Estimation of the Determinants of Bicycle Mode Share
for the Journey to Work Using Census Data. Transportation. 35. 93±109 (2008).
2. Fishman. E.; Simon. W.; Narelle. H.; Armando. M. Barriers to Bikesharing: An Analysis
from Melbourne and Brisbane. J. Transp. Geogr. 41. 325±337 (2014).
3. Bachand-Marleau. J.; Lee. B.H.Y.; El-Geneidy. A.M. Better Understanding of Factors In-
fluencing Likelihood of Using Shared Bicycle Systems and Frequency of Use. Transp. Res.
Rec. J. Transp. Res. Board. 2314. 66±71 (2012).
4. Pucher. J.; Buehler. R. City Cycling; MIT Press: Cambridge. MA. USA (2012).
5. Zhang. Y.; Liu. X.; Zhang. H. Survey on the Reorganization and Using Status of Public
Bicycle System in Urban Fringe Areas: Taking Tongzhou and Daxing Districts of Beijing
for Example. Urban Probl. 3. 42±46 (2015).
6. Rahul. T.M.; Verma. A. A Study of Acceptable Trip Distances Using Walking and Cycling
in Bangalore. Transp. Geogr. 38. 106±113 (2014).
7. Keijer. M.J.N.; Rietveld. P. How Do People Get to the Railway Station? The Dutch Experi-
ence. Transp. Plan. Technol. 23. 215±235 (2000).
8. Dill. J.; Gliebe. J. Understanding and Measuring Bicycling Behavior: A Focus on Travel
Time and Route Choice. Bicycl. Lanes 29 (2008).
9. Olson. K. A Review of Recent Literature. Glob. Bus. Organ. Excell. 1. 129±132 (2010).
10. Li. X.. Y. Zhang. M. Du. and J. Yang. ³6RFLDO)DFWRUV,QIOXHQFLQJWKH&KRLFHRI%LF\FOH
Difference Analysis among Private Bike. Public Bike Sharing and Free-Floating Bike Shar-
ing in Kunming. &KLQD´.6&(-RXUQDOof Civil Engineering (2019).

 
  
  
 


 




 
 



        
   


 
 


 
   !"! #
$ %! &   ' ''  
% (      )           ! 
'   "    #
 %
' '"     )   
"    ' %
) !    !     
"   " *       %+ ""
!            ,   "   '     
%

! $  #   %

!  

! $ - ""  "      
,  "  %
      '  #
 "    !  -%+  
 "   '#   ') ./0 1 2
  '"   %
  '    $ "  
' !      )         !      
 )    "         
%+    '!    ' #
)     % 3 ! . ,  "  0 
 '     4 #
 3      "   )  
5 2%  '   6   .     0       '   
47
3 !    )  
2%
) 
      -   8 '
 *   ! 
2% 9 "             #
)       "  %

      !  "    " #
       "    !        % :    
         '  "!        !
"%

© Springer Nature Singapore Pte Ltd. 2020 1069

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_171
1070 T. N. To et al.

! &   ' ''   ' 

" !;
<7;=<%
!    "  ! & 
-  )       %+  " " " #
             !   '       
)     %
""       '
!  ,  "  %
 $   " 
'   )     %

"
#




"
$



   !  , "  .>

0%+  ,  !"! $ %
 

 '  4#
3 %+ ?  
      '   '! 
     ) %



     @A"%

""
! 



+  ' !  
B   %

  '  !           
  %
 


"  !      /           !   !#
    A+B@ A%
 @A        !   " '
      C  "

D1 8  '! " 
DE     
 
D
%9 "
#1   A+B@>::'
           ?       "    %

  '!"   4+  !!   "
Flood risk assessment in the planning of new urban … 1071

"%
$


  '!   -"   )  E2
 
2%>  "  ' !  '
! ) %   '  !
%

&
 > -  %

   > ># @A"

) ./0
B3

2      ! 
B31
2 +    ! 
B35 E2      ! 
B3= E2 +    ! 

%
'
 

%
(

 


 

    ' !      '  *    . 
FFF
 1 G0%
  !   " 
    .(0 #  .0   
 "-.
H20%> !1 (   ! 
    "   %GE %IE   %GI %II "     
  "  % + 
FFF   1 G      "-  .
H !0 
1%
G2  %JF2"%
!    
"  !   -     #
! $ %

&
"   
B %
! 
FFF   1
J
 
 (   (

B %GE %IE  %GI %II

&
%
  "-   
FFF 1 G%
  K H !.0 H.0
H !.20

FFF 5%JF 5%J
 1%
G2

-
1 G 1%F 1%F1 %JF2

!     
B #
    "    >1% +   !     "  '  
  ' !%

1072 T. N. To et al.

+
,

+,

+,



":! ' *   .0C
FFF .!0C1 G%

%"
&  (

) * 

 

                '   '     
   ' "       )    '    >% 5 =% + 
 " "'!-  " 
        " ! 
  %

?     "! 
!=%> #
 "       )  E2 

2%
  "    )  E2 

2"#
 "     89" "  
   "     !  # %E   /LE2   # %1E   /L
2% > 
  )  E2  "    " 5#5%E#
!I%
1-1  "    " =#=%E #
!I%GE-1%>  " 5   =%E 
   %>  )  
2  "   
 " 1#1%E 5#5%E!=%JF-1 
%=1-1"%

  "    " 5%E#=% !
E%I
-1%
>  " 1%E   =%E    
 %
Flood risk assessment in the planning of new urban … 1073



 %
    ' "    ' 
'    ! .> )  
20%



 -
    ' "    ' 
'    ! .> )  E20%
1074 T. N. To et al.

&
-
        ) ./LE2 /L
20%

+   .-10 +   .-10

> " > "
B35 B3= B3
 B31
% # %E %1F %1I % # %E
%GJ
%I

%E7
% 
%=I
%E1 %E7
% 
J%G

G%
I

% #
%E
J%FE
J%FJ
% #
%E I%1
 G%GI

%E71% 
%FI
%F5
%E71%  5%E= 5%EI
1% #1%E J%
 J% 5 1% #1%E
I% F
I% 5
1%E75%  1 %I  1 %JI 1%E75% 
G%
=
1%=E
5% #5%E
F%JJ

%GI 5% #5%E
G%EF G%
G
5%E7=%  1
%E1 1
% I 5%E7=%  5%IE
F%JJ
=% #=%E %

 I%IJ =% #=%E %=  %= 
=%E7E%  %5  %5  =%E7E%  %EJ %EJ
ME
%JJ
%JG ME %  % 

 FF%IJ
% F
 IG%IE II%J


-
(  

   '  -                  

    "      )   %
 
  '    "  "       )   E2  
2
" "     %
 '  
 '! -    "  '#
 )  ! %

  

     ! >       
    "   
     " * !31
G#N 1#1F%

' 

%
 %
%B%A%CA #     $  

3 "      $   
%9""%4 %5
.50
I17
F5.1

0%
1%
 %%  %C @  " "   "     ?
   ,   %%$GE.10
G=G7
GJE.1
E0%
5%
 %%    O%  %
  A%C   4+   " " " 
 ?   %9""%4 %=
JE7G=.1
50%
=%
($ ! 3%%% B (% 9 A% +  %C 9 $    -   
"   !          % A% 9"% % 4 !%  %

F.10
GG7
FI.1
=0%

 Low Impact Development, a novel technique for cutting
down urban flooding in Quynhon city, Vietnam

Nguyen Duc Phuoc Vo[1,2], Muhammed A. Bhuiyan [2] and Ngoc Duong Vo[1,*]
1
University of Science and Technology, The University of Danang
2
School of Engineering, RMIT University, Melbourne, Australia

Abstract. Many coastal cities are becoming more vulnerable to flooding due to
increases in extreme precipitation as a consequence of climate change and reduc-
tion of infiltration due to urbanization. Application of Near-natural stormwater
management measures such as Low Impact Development techniques (LIDs) is
considered as an alternative option to reduce the risks of urban flooding caused
by climate change and urbanization. This technique reduces inundation possibil-
ity by using natural or man-made systems to filter and recharge stormwater into
the groundwater. In this research, applicability of two LIDs, infiltration trend and
swales, was evaluated with case study in Quynhon, Binhdinh province, Vietnam.
The results show that, the LIDs work effectively to reduce peak flood and en-
hance recharged amount groundwater in short duration rainfall. Their effects,
however, depend on reaching saturation points of applied measures.

Keywords: Stormwater management; urban flooding; low impact develop-

ment; climate change; extreme precipitation.

1 Introduction

Urban flooding is one of severe problems that many cities are currently facing. Devel-
oping and upgrading drainage systems are common approach to tackle the risks of in-
undation in many areas [1, 2]. However, the systems may lead to unexpected influences
in cases of exceeding flooding as compared to the designed capacities.
Urban flooding problem is more likely to increase under pressures of urbanization
and climate change [4]. AR5 report from Intergovernmental Panel on Climate Change
in 2013 stated that, the increase in extreme precipitation and sea level rise in many
coastal urban areas is subject to occur over the world [3]. Moreover, urbanization asso-
ciates with an issue that many pervious land covers have been replaced by impervious
areas, which reduce initial abstraction ratio in urban areas. As a consequence, storm-
water runoff significantly increases and critically threaten many urban areas.
During the last decades, applications of near-natural based solutions to reduce
stresses on stormwater discharge have been widely used as a sustainable option. In
which, Low Impact Developments (LIDs) is proven as a potential alternative for storm-
water management [9]. The technique improves returning stormwater runoff to natural
hydrology through infiltration and evaporation, which contributes to reduce

© Springer Nature Singapore Pte Ltd. 2020 1075

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_172
1076 N. D. P. Vo et al.

concentration time and peaks of flooding flow [1]. In addition, LIDs further decrease
contaminants in stormwater and provides green places and habitat for many animal
species. Therefore, LIDs are applicable to support for sustainable development of urban
areas.
Applications of LIDs in Vietnam however remains limited. For instance, stormwater
discharge is still mainly relying on sewage systems. LIDs effects have not been fully
evaluated for field applications into sewage networks. Consequently, the local authori-
ties are less likely to implement the technology into the field or even in integration into
plans for brand-new systems. Thus, the main objective of this research is to intensively
study LIDs performances, evaluate applicability of LIDs in condition of in local scale.

2 Methodology

To achieve research objective, several stormwater drainage models were implemented

in different scenarios, with LIDs integration and without LIDs (based cases). The
Stormwater Management Model (United State EPA) was used to set up the models. The
simulation outputs were then analyzed to obtain the characteristics of the selected LIDs.
This enables some suggestion for the most suitable practices to apply in the future.

2.1 Case study

The project was implemented in Quynhon, the capital city of Binhdinh Province, Vi-
etnam. Only the central area which mainly comprised of residential and public areas
with the impervious rate ranging from 50% to 80% [7] was studied in this research. The
drainage system in Quynhon was developed in the 1980s and upgraded from 2007 to
2010 [6]. According to Sewage Drainage Planning of Quynhon, no upgradation or re-
construction will be made from 2020 to 2050. Instead, advanced stormwater manage-
ment such as LIDs will be applied to cope with climate change [6]. Therefore, applica-
tion of LIDs matches the sustainable development strategy of the city.
The study area has the typical characteristics of climate in South Central Coast of
Vietnam. Rainy season usually occurs from September to December, accounting for
80% of annual rainfall [7]. Specifically, most of heavy rainfalls happen in October and
November along with monsoon and tropical storms and become the main reason for
flooding in the city [6].

2.2 Low Impact Development technology

To investigate the applicability of LIDs, infiltration trenches and swales were selected
in this research. The selection of those practices was made due to the ability to fit suit-
ably and conveniently in the center area of roads, gardens, and sidewalks in developing
cities. In our case, these LIDs have accommodated with the current drainage system in
Nguyen Tat Thanh Street, placing LIDs along the street. The proposed scenarios of
/,'V¶FULWHULDDUHUHSUHVented in Table 1.
Low Impact Development, a novel technique … 1077

Table 1. Design parameters of selected LID techniques

Indicator Infiltration Trenches Swale
Bern high (m) 0.5 1
Storage (m) 2000 N. A
With (m) 5 30
Slope 0.1 0.1
Sweep rate 0.8 N. A

2.3 Rainfall simulation

According to observation data from Flood and Storm Committee of Binhding province,
there were two severe flooding events in 2009 and 2010 [10]. The flood in 2010 high-
lighted the impacts of high-intensity rainfall (56.1 mm/hour) on the drainage system
while the 2009 event emphasized the system sensibility under long duration rainfall.
Figure 2 shows the historical rainfalls record at Quynhon meteorology station. To ex-
amine the potential effects of LIDs, under variation of rainfall amount suitable in dif-
ferent return periods (Table 2).

Table 2. Amount of rainfall in different return periods.

Frequency (%) 2% 5% 10% 20% 50%
Return period (years) 50 20 10 5 2
Rainfall (mm) 334.14 297.36 266.78 232.30 173.83

2.4 Urban drainage model

Stormwater drainage scenarios were simulated for sewage system at Nguyen Tat Thanh
street by using Stormwater Management Models (SWMM) (United States of America
EPA). The SWMMs model is widely used by researchers and engineers to investigate
the efficiency of LIDs [1]. The map of SWMM model is described in Fig. 1.

Fig. 1. Study area drainage system in SWMM

1078 N. D. P. Vo et al.

Fig. 2. The simulation rainfall for these 2009 and 2010 events.

3 Result and discussion

3.1 Effect of simulated LIDs

Simulation results for three scenarios with rainfalls in 2009 and 2010 are shown in Fig.
3. Infiltration trenches and swales contribute to reducing total runoff in outfall. Under
the infiltration trenches scenario, flood volume decreases 13.03% and 12.96% for 2010
and 2009 events, respectively. Meanwhile, flood reduction in swales scenarios reaches
to 11.01 % and 10.83 % in the flooding event.
These LIDs, infiltration trenches and swales have reduced flooding in different ways.
Swales cut off flood more effectively in the first period of the flood events (before the
17th hour). In 2009, simulation with swales application reduced the total flow reaching
the outfall 80% as compared to the based cases. This is 60% for infiltration trenches.
However, after the moment of 17th hour, infiltration trenches work more significant in
cutting down runoff generated in the drainage system.
Simulated results also indicate that, infiltration trenches are more likely to reach sat-
uration points later than swales performance. In 2009, simulation with swales applica-
tion shows that the technique reaches saturated at around the 17th hour while the figure
for infiltration trenches simulation is at 18th hour. Similarly, in simulation for the 2010
event, saturation point is seen at 18th hour for swales computation, and it is 19th hour
for infiltration trenches.
These simulations further expose limitation that infiltration trenches and swales ef-
fects decrease when they close to saturation. In other words, LIDs work less effectively
after reaching saturated points. For example, in the 2009 event, total flow at the outfall
reach to a peak of 1 m3/s according to the base case at 10-hour and 35-hour points.
However, the reaction of LIDs contradicts at the moments, with cutting down over 60%
and 10% total flow at former and later points respectively.
Low Impact Development, a novel technique … 1079

Fig. 3. Simulated flow at outfall in three scenarios for flood events in 2010 (a) and 2009 (b).

3.2 Effect of different rainfall patterns

Figure 4 represents the simulated results of scenarios in different rainfalls. LIDs flood
reduction capacities consistently depend on rainfall characteristics. Both LIDs work
more efficiently in small rainfall. Furthermore, their capacities decrease gradually fol-
lowing increase of rainfall amount. Regarding reduction of flood flow, swales are more
practical than infiltration trenches only for events which have less than 220 mm of rain-
fall. This is however vice versa for the ones greater than 220 mm.

Fig. 4. Flow reduction rate of LIDs in different rainfall.

1080 N. D. P. Vo et al.

Figure 4 further indicates the variation of flow reduction of swales and infiltration
trench under different durations of rainfall. Their performances of LIDs are impacted
negligibly by the duration of rainfall which total amount is in a range from 200 to 350
mm. However, in lower rainfall, the impact mechanism is different between these two
LIDs. Both LID practices work more effectively in the 2010 event and less in the 2009
event. This finding shows that, the recovery ability after saturating of these methods is
much better to the rainfalls having similar pattern as the 2010 event.

4 Conclusion

Application of Low Impact Development is considered as a potential measure contrib-

uting to reducing flood impact in urban areas. Infiltration trenches and swales are sen-
sitive in different intensity and duration of rainfall. Therefore, integrating multiple op-
tions of LIDs can promote better solution for stormwater management and flood reduc-
tion in urban areas.

5 Reference
1. Qin, H.P., Z.X. Li, and G. Fu, The effects of low impact development on urban flooding
under different rainfall characteristics. J Environ Manage, 2013. 129: p. 577-85.
2. Willems, P., Revision of urban drainage design rules after assessment of climate change
impacts on precipitation extremes at Uccle, Belgium. Journal of Hydrology, 2013. 496: p.
166-177.
3. IPCC, Climate Change 2013: The Physical Science Basis. Contribution of Working Group
I to the Fifth Assessment Report of the Intergovernmental Panel on Climate Change, ed. T.F.
Stocker, et al. 2013: Cambridge University Press, Cambridge, United Kingdom and New
York, NY, USA, 1535 pp. 1535.
4. Miller, J.D. and M. Hutchins, The impacts of urbanisation and climate change on urban
flooding and urban water quality: A review of the evidence concerning the United Kingdom.
Journal of Hydrology: Regional Studies, 2017. 12: p. 345-362.
5. IGIP and VIWASE, Sewage drainage planing of Quynhon to 2020 and the vision to 2050
and integrate climate change scenarios. 2015.
6. BD-PPC, Sewage drainage planing of Quynhon to 2020 and the vision to 2050 and integrate
climate change scenarios, B.D.P. Commitee, Editor. 2016.
7. Phung, T.T., Vietnam - Coastal Cities Sustainable Environment Project : environmental
assessment : Environmental and social impact assessment for Quy Nhon city subproject
(English). 2016. p. 32.
8. DiGregorio, M., Learning from Typhoon Mirinae: Urbanization and climate change in
Quynhon City, Vietnam. Institute for Social and Environmental Transition. Hanoi, Vietnam:
ISET-Vietnam, 2013.
9. Eckart, K., Z. McPhee, and T. Bolisetti, Performance and implementation of low impact
development - A review. Sci Total Environ, 2017. 607-608: p. 413-432.
10. Binhdinh Flood and Storm committee (2010), Flood and storm committed report, Binhdinh
province 2010
 Vehicle usage/ownership control for a Sustainable
Transport system in the Motorcycle Dependent Cities
Dr.-Ing. Le Thu Huyen 1 and Dr. Nguyen Thanh Tu 2
1, 2
University of Transport and Communications, Vietnam
huyen318@gmail.com

Abstract. There is a distinguishing characteristic between traffic flow in Vi-

etnam (and in many other Asian countries) and those in developed countries:
two-wheeled vehicles (so-called motorcycles) consist a high percentage in the
road traffic system. Motorcycles, for their special advantages, play a dominant
role in the traffic flow in Vietnam cities, leading to the context of traffic de-
pending on motorcycle (so-called motorcycle dependent cities- MDCs). The
transportation system in MDCs is facing with a lot of chaotics and problems,
such as congestion, pollution and traffic accidents. Recently, there have been a
lot of traffic control measurements conducted in the effort of solving the prob-
lems. Unfortunately, there is still very little that cities know about managing
this mode. This article describes work undertaken to provide a better under-
standing of motos and what might be done to better position their role in
providing urban mobility. Lessons learned and potential solutions for Vietnam
could be widely informative across East Asia and other regions where develop-
ing cities are facing similar challenges.

Keywords: transportation management, motorcycle dependent cities, motorcy-

cle usage control, sustainable development.

1 Current situation of motorcycle dependent cities

The rapid increasing rate of motorcyclerization in Asian cities is significantly impact-
ing the environment and living standards as well. The rapid growth in private car
ownership has led to excessive road constructions that are still insufficient and induce,
for instance, traffic congestion, road accidents.
The combination of low transit mode share and high motorcycle ownership makes
Hanoi relatively different. Unlike most major cities in developing countries where
public transport plays an important role, Hanoi is a city with a public transport mode
share that is much lower than those in similar size cities in developed countries and
high private vehicle ownership, albeit motorcycles rather than cars.
Motorcycles in Hanoi, and Vietnam in general, are now almost exclusively pow-
ered by four-stroke internal combustion engines, with fuel injection accounting for
only a relatively small, albeit rising, share of the fleet. Engine size is low, with virtu-
ally all motorcycles and scooters in-use having engine displacement of less than 150
cc and with the most common size being 125 cc.
From statistics data1, there were 610 motorcycles and 36 cars per 1,000 people in
Hanoi in 2011 (with 700 motorcycles and 35 cars per 1,000 people in Ho Chi Minh
City in the same year). The number has increased to 736 motorcycles and 86 cars per
1,000 people in Hanoi in 2017. In comparision, the total number of registered motor-
ized vehicles (including buses, trucks and other commercial vehicles) in developed
countries typically rank at the rate of 500-600 per 1,000 people in Europe and 700-
800 per 1,000 people in countries such as Australia and the United States of America.

Vietnam Registration Annual Report (2018)

© Springer
1
Nature Singapore Pte Ltd. 2020 1081
C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_173
1082 L. T. Huyen and N. T. Tu

Currently, motorcycles cover more than 75% travel demand in the whole country.
With the outstanding advantages of flexibility, maneuverability, low cost,... in the
context of under-developped public transportation, in the near future, motorcycles
will remain to be the major transportation mode of Vietnamese in general and
Hanoian in particular.
However, motorcycles are those with the highest risk of traffic accidents among
road vehicles. Crash rates are rather high, with motorcycle riders extremely vulnera-
ble in the lack of physical protection when travelling at a relatively high speed. While
motorcycles dominate the traffic flow, the different operation characteristics of mo-
torcycles, bicycles, cars, buses and trucks makes the flow uneven and reduces the
effective road capacity as well as heightening the risk of accidents. The rapid growth
in vehicles has led to a substantial increase in air pollution, with many motorcyclists
wearing masks to reduce inhalation of pollution and dust. Vehicle users treat foot-
paths as an extension of the roadway, with pedestrians required to manoeuvre be-
tween vehicles to use footpaths. Also with their flexibility, motorcycles cause a lot of
issues in urban chaotics in parking and moving activities (eg. occupying the
toothpaths, in the illegal places,...).
This typical context presents a huge challenge for government when trying to de-
velop a sustainable urban transport system, in order not only to ensure traffic safety
and traffic order in motorcycle dependent cities, but also to take their full advantages.

2 Method

The study was conducted using a questionaire based survey in Hanoi in 2017 in order
to understand behaviours of road users and explore the ability of optimizing their
vehicle usage. There are 2,000 questionaires launched and 1,853 collected. The sur-
vey has used the best practice method for such surveys, including an indirect means
for ascertaining the views of respondents and the collection of data for a large sample
that covered a balanced representation of the community. Sampling also tried to en-
sure that a minimum number of some smaller socio-economic groups were surveyed
in order to ensure identifying the effect of various socio-economic factors.
The Survey has used the Stated Preference methodology to evaluate quantitative
factors influencing the use of various transportation modes, including travel time,
time approaching to public transport stops and terminals, fuel and maintenance cost,
parking cost and public transport fares for motorcycle, car, motorcycle taxis (³xe
ôm´), street bus and rapid transit. The second part of the questionaire has asked for
the relative importance of the vehicles attributes and the quality of transport services
to travellers, comparing both private and public transport modes.

3 Results

3.1 Motorcycle Advantages and Disadvantages

The survey has proved the fact that motorcycles have many advantages over cars and
bicycles. They are (i) the motorcycle advantages of flexibility, maneuverability, pow-
er, convenience and relatively low cost of purchasing, operation and maintenance; (ii)
the lack of convenient and safe infrastructure for bicycle using; (iii) the uncomforta-
ble climate for bicycling of hot and high humidity; (iv) the fact that motorcycles make
more efficient use of road space than cars in terms of numbers of people travelling on
any width of road.
The survey reveals the fact that it is not vehicle ownership increasing but the usage
of these additional cars and motorcycles causing problems and issues in Hanoi traffic
flow. However, many of the negative aspects of motorcylce usage can be ameliorated
at relatively low cost through education and training measures, minor traffic man-
Vehicle usage/ownership control for a Sustainable … 1083

agement works, and improving traffic management together with better regulation and
enforcement of the regulations.

3.2 Road UVHUV¶Attitudes Toward Different Transport Modes

The Survey provides the evidence on how road users react to attributes of various
modes of transport. 5RDG XVHUV¶ sensitivity, with the purpose of their trip and their
income and gender, to various attributes of different transport modes has been suc-
cessfully estimated.
Though people do not have the same response to the fluctuation of these attributes,
the survey results provides generally reliable results that can be used to identify, eval-
uate and apply the best initiatives to achieve sustainable objectives.
The results indicate that attracting large numbers of people from motorcycle and
car to public transport will be challenging. People are generally not highly sensitive to
µVWLFN¶ LQLWLDWLYHV VXFK DV KLJKHU WD[HV DQG FKDUJHV for use of motorcycles and cars,
QRUDUHWKH\YHU\UHVSRQVLYHWRµFDUURW¶LQLWLDWLYHVVXFKDVLPSURYLQJSXEOLFWUDQVSRUW
This result should not be especially surprising. The mode share of public transport in
Hanoi at present is similar to that in many developed countries. Despite supportive
and financing policies, governments in developed countries have generally had lim-
ited success in securing a strategic shift in travel from private to public transport. This
reflects the challenge of getting people who prefer pULYDWHYHKLFOHV¶attributes to trans-
fer to public transport, which is compounded by the evolution of land use patterns that
are better matched to private transport.
This situation suggests two broad strategies. The first is to introduce measures that
apply much more pressure on people to change thir travel choices. The other is to use

Source: Survey data (2017)

Fig. 1. Relative Importance of Factors that Affect Vehicle Ownership

a mix of complementary initiatives that draw on the insights that can be gained from
the analysis presented in this section. This mix of initiatives will need to draw not just
1084 L. T. Huyen and N. T. Tu

of the characteristics of modes that it has been possible to examine quantitatively, but
also those that qualitative in nature.
The results of the survey of the relative importance of the qualitative factors on ve-
hicle use are shown in Fig.1. They indicate that:
x Road safety and person security are the most important matters. These matters
(items 3 and 4 in the Fig.1) have a strong influence on the choice to use a partic-
ular mode of travel. Motorcyclists are more concerned about road safety than car
and public transport users.
x Users rank safety as the most important facture when purchasing and using mo-
torcycles. However, a significant number of users still drive dangerously includ-
ing failing to observe traffic signals, entering already crowded intersections, ex-
cessive lane changing, and using a mobile phone while riding.
x Travel decisions by motorcycle users are heavily influenced by travel condi-
tions. These conditions include travel time, weather conditions, traffic condi-
tions, and the presence of dust and fumes (items 1, 11, 12 and 14).
x Bus users are concerned by a range of attributes associated with bus services.
Matters that influence decisions to use a bus that received an above average
score are travel time, safety, personal security, flexibility, personal space, relia-
bility, traffic conditions, the presence of air conditioning, the number of transfers
needed to complete a journey and the convenience of getting to a bus stop (items
1, 3, 4, 6, 8, 9, 12 and 15-17).
x Bus users are not overly concerned by fares. While motorcycle users are con-
cerned with costs (item 2), bus users indicate it is not a serious matter. The latter
may seem to be an anomaly given the common negative response of the com-
munity to rises in public transport fares.
x Motorcycle and bus users see the environmental benefit of their respective
modes as being moderately important. This provides a challenge in terms of us-
ing the environment as a means for encouraging people to shift from motorcycle
to public transport.
x People travelling with education and shopping purposes are more concerned
with travel costs that those travelling for other purposes. It is likely that this
finding reflects the effect of income rather than the trip purpose. Views on the
importance of the other factors shown in Fig. 1 do not vary substantially.
x The attributes cited in the surveys consist the flexibility regarding to travel loca-
tion, travel time and household travel arrangements, as well as moderate pur-
chase costs and low operating costs. Moreover, there are also factors that make
motorcycle use unattractive, including poor safety, air pollution and hot and wet
weather, which are especially unpleasant for longer distance journeys. Even so,
when surveyed, motorcycles users showed little interest in switching to other
modes in the near future. They felt that public transport, bus in particular, did
not match the positive attributes of motorcycles and was not sufficiently attrac-
tive to overcome the unappealing attributes of motorcycle use.
Along with vehicle ownership, opinions on vehicle use generally do not differ sub-
stantially with other socio-economic variables. The Survey also indicates that men
and women have very similar views, as do people of various ages and people coming
from households with or without children. The principal variations are that people
from households with higher income are less concerned with the cost of using their
vehicles and those from households with low income are more concerned with road
safety.
3.3 Road User Awareness and Behaviour of Using Motorcycle
The Survey also collected data on the awareness and using private vehicles in the
households of respondents. From the data, it can be seen that (i) Users misperceive
the use of vehicles and travel costs; (ii) Motorcycle usage is not directly proportional
Vehicle usage/ownership control for a Sustainable … 1085

with ownership; and (iii) Motorcycle usage has close relationship with the number of
people in a household.
Based on available data on the average distance travelled by cars and motorcycles
each year (15,000 km and 7,400 km respectively), car and motorcycle users both ap-
pear to under-estimate the quantity of travel that they undertake by around 40 percent.
,QVRPHFRQWUDVWGDWDIURPWKHVXUYH\RQUHVSRQGHQWV¶HVWLPDWHRIWKHWUDYHOGLVWDQFH
and the weekly cost of using their vehicles reported in the survey is equal to the esti-
mated average operating cost of VND 1,160 (US$0.055) per km for motorcycle users
and 55 percent of the estimated VND 7,500 (US$0.354) per km cost for car users.
Capital costs are incurred in addition to operating costs. Some capital costs are related
to time (e.g. a vehicle will still depreciate even if left unused), some capital costs are
related to the quantity of travel that is undertaken. This adds to the extent to which car
and motorcycle users under-estimate the cost of their vehicle use. The extent of mis-
perception of vehicle operating costs will be more severe than indicated if the weekly
perceived cost indicated by respondents was correct but it applied to the higher quan-
tity of kilometres of travel actually made.
A key matter is the potential role for controls on vehicle ownership to achieve de-
sired urban transport outcomes. In practice, the objectives of urban transport policy
focus on the outcomes that result from vehicle use, e.g. congestion, pollution and
crashes. While the availability of a vehicle is needed to allow it to be used, the ques-
tion remains as to the extent to which increased ownership of motorcycles results in
more vehicle-km of travel. Hence, it seems likely that a government regulation to
reduce motorcycle ownership would lead to a lesser reduction in motorcycle-km of
travel, with households re-arranging their affairs to make more use of remaining mo-
torcycles. It is thus concluded that the level of ownership of motorcycles does have an
impact on the quantity of travel undertaken, but the effect is modest and thus the more
important focus should be motorcycle use rather than ownership. This is reinforced by
the next item.
Motorcycle use is more closely linked to the number of people in the family. Drawing
on the same database described above, the coefficient of correlation between motor-
cycle-km of travel and the number of people in a household is estimated to be 0.58.
This indicates that the quantity of motorcycle use is strongly linked to the number of
people in a household, and that this is more than twice as important as per capita mo-
torcycle ownership.

3.4 Solutions for Optimizing Motorcycle Usage Behaviour

Understanding the attitudes toward different transportation modes as well as their
awareness and behaviour of motorcycle usage helps in proposing solutions and policy
for traffic management.
From the survey, it is suggested to focus on efficient solutions to change the vehi-
cle using behaviour of people rather than their ownership decisions. Such tools and
measurements are listed as follows.
x Parking policy can be used to improve travel decisions and to improve public
amenity. Some examples can be that developing regulations on parking fees of
motorcycles, including the parking time, especially parking in the toothpaths or
on road. Illegal parking should be fined at the high rate.
x Travel fees can be better exploited to achieve improved transport outcomes.
Transportation fees serve two roles: to generate revenue; and to provide signals
to the public that enable them to make better decisions on how much travel they
make, the locations to which they travel, the time of their trips and the mode
they use. Improving the acknowlegement of road users in the fact that vehicle
operation cost is bigger than the cost of fuel and parking only. Such understand-
ing can make the public transportation more attractive with their low cost.
1086 L. T. Huyen and N. T. Tu

x Air pollution from motos could be reduced remarkably through the faster im-
plementation of stricter emission standards.
x People with high income prefer owning and using cars, however, their VOT is
higher. Hence, they may be attractive with public transport at a high quality
and high speed.
x Actions that may make motorcycle usage more appealing need to be comple-
mented with additional efforts to maintain the relative attractiveness of public
transport.
x The capacity of signalised intersections can be increased by up to 25 percent
through minor adjustments to intersection layout and signal settings together
with higher levels of compliance with traffic regulations.
x Seperating motorcycles from other vehicles using a combination of fixed bar-
riers and overhead signs may reduce conflicts among them. It may also in-
crease the traffic flow capacity for each group and improve traffic safety.
x Last but not least, it is required a long-term strategy to optimize the role of
public transportation in a comprehensive manner from the master plan to con-
sistent and efficient implementation scheme.

4 Conclusion and Recommendation

In the near future, Hanoi and other Vietnam cities may still depend on motorcycles
because of nRWRQO\WKHLUDGYDQWDJHVEXWDOVRWKHFLWLHV¶LQIUDVWUXFWXUHZLWKPDQ\ at-
tributes convenient to two-wheelers. From the study, it can be seen that not ownership
but using motorcycles and/or cars is the direct cause of the traffic chaotics in the ur-
ban areas. Therefore, optimizing vehicle usage should be the target of traffic policy
and management in such motorcycle dependent cities as Hanoi.
The ideal model of transportation in the urban area still remain with the public
transit as the backbone of the network; non-motorized vehicles are encouraged to be
used as feeders to support the public transportation. Cars and/or motorcycles are man-
aged in such a way to provide the urban people other mobility options, and to encour-
age them to use in a smart way, aiming at improving the performance of transporta-
tion system, protecting the environment, reducing traffic congestion and accidents.

References
1. Kumares C.Sinha; Samuel Labi (2007), Transportation Decision Making, second edition,
John Wiley & Sons, Inc.
2. Le Thu Huyen (2009). Driver Behaviour, Risk Analysis and Traffic Safety in Motorcycle
Dependent Cities, PhD Thesis, TU Darmstadt, Germany
3. Litman, T. A. (2014). Transportation Elasticities: How Prices and Other Factors Affect
Travel Behavior. Victoria Transport Policy Institute, Canada.
4. Vu Anh Tuan (2011). Dynamic interactions between private passenger car and motorcycle
ownership in Asia: Across country analysis. Journal of the Eastern Asia Society for Trans-
portation Studies, 9:541-556.
5. Vietnam Registration Annual Report (2018)
 Typhoon Wave Modelling For Viet Nam Central Region.

Ngoc Duong Vo1, Quang Binh Nguyen1 and Cong Phong Nguyen2
1
The University of Danang ± University of Science and Technology, Viet Nam
2
Southern Institute of Water Resource Research
vo.ngoc.duong@dut.udn.vn

Abstract. Coastal zone plays a very important role in Vietnam socio-economic

development, especially with the central region. However, the coastal line in Vi-
etnam central region has been changed seriously. It becomes more complicately
in recent years when shoreline erosion has occurred in many localtities. The con-
sequence of this natural phenomenon has affected significantly to these area. The
sea level rise and typhoon wave are considered as the main causes of unexpected
phenomenon. With the aim of mitigating the impact of typhoon wave to coastal
line in cetral Vietnam, this study is realised. By using Mike softwave family of
DHI for coastal zone from Binh Dinh to Binh Thuan, Viet Nam, the result will
allow not only us to understand, predict and design mitigation measures to com-
bat the effects of these typhoons, it will also allow us to develop a methodology
to be used elsewhere on the Viet Nam coast. The model is calibrated and vali-
dated versus water leve data of Quy Nhon station and of two other temporary
stations in central area, also comparered to WAVEWATCH III model result with
high performance. Result from four different typhoon scenario (from level 6 to
level 12) are analysed to provide basic information about the wave situation then
estimating their impact to study area.

Keywords: Coastal area, typhoon wave, natural phenomenon, Vietnam central.

1 Introduction

The sea plays a very important role in our country's socio-economic development.
Therefore, the protection of the coastline and the marine environment needs special
attention. The central region has a long coastline, many rivers, a potential and a great
advantage for economic development. However, the problem of coastal erosion in re-
cent years has become a hot topic of many localities in the country [1]. Along the central
region, landslide hot spots continue to take place like Vinh Hai commune, Phu Loc
district, Thua Thien - Hue; landslide on Cua Dai beach, Hoi An city, Quang Nam prov-
ince; beach erosion through Xom Ro, Phu Dong ward and An Phu commune area, Tuy
Hoa city, Phu Yen province; Coastal erosion at localities in Tuy Phong district, Phan
Thiet city, La Gi town, Binh Thuan and Cua Dai provinces (Quang Ngai); Tam Quan
and Cua An Du doors (Binh Dinh), etc. According to statistics, along the Central Coast
strip, there are 88 landslides with a total length of more than 120 km. The erosion has
caused negative consequences to the local socio-economic development, causing great

© Springer Nature Singapore Pte Ltd. 2020 1087

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_174
1088 N. D. Vo et al.

losses of people and property [2]. Therefore, the research and finding solutions to min-
imize landslide and stabilization of the coast for socio-economic development are very
urgent and necessary.
Besides, the central provinces have focused on investing to exploit the strengths of
the marine economy. Many infrastructures are invested in construction, especially in-
frastructure for smokeless industry. Many resorts, resorts, residential areas and treat-
ment facilities are built more and more in coastal areas. These buildings are directly
affected by the waves and face serious problems of charging and discharging. Espe-
cially in the context of climate change and sea level rise, when extreme weather events
often occur with higher intensity and frequency.
In this context, the issue of studying the impact of waves to the coast and coastal
works is very urgent. Currently there are many solutions, but the digital model is con-
sidered one of the most effective and convenient solutions. In order to have a scientific
basis to propose solutions to minimize waves to coastal works, this study conducted
simulating the hydrodynamic phenomenon at Vietnam central coast. This research
hopes to provide the necessary information to serve in coastal reinforcement as well as
to plan coastal schemes in the central region, especially Khanh Hoa province.

2 Methodology

Simulation is done with the MIKE 21 FM model of DHI. MIKE 21 FM is a performance

tool for solving oceanographic problems, coastal and estuarine environments [3]. Due
to the peculiar nature of the waves, in order to properly and fully reflect the wave char-
acteristics of the study area, this study uses models with different ratios and levels of
detail. The whole East Sea model (model 1) is used to simulate tidal processes as well
as waves and storm surges transmitted from offshore into coastal areas. This model is
inherited from Study on changes in the hydrodynamic of coastal estuaries affected by
the Vung Tau - Go Con sea dyke project and adjusted for this case study.The results of
this model will be used to build the seaward-opening boundary for the detailed coastal
2D model of the study area (model 2).

Fig. 1. Method using multi-scale model- Model 1 (left) Model 2 (right).

Typhoon Wave Modelling For Viet Nam Central Region 1089

2.1 Model set up

Model 1 (East Sea model): Fig 1 shows the range of model and grid used. The main
boundaries of the model are the Taiwan Strait, Luzon, Mindoro, Babalac, and Malacca.
With the HD hydrodynamic module, these margins are the water level boundaries with
tidal data built from harmonic constants. For a module that calculates SW spectrum,
these margins are assumed to be lateral boundary. Because the study area is the coastal
area of Khanh Hoa province, the area from Binh Dinh to Binh Thuan is updated to
include a fine coastal net. In shallow coastal areas, the grid is also divided smaller.
Model 2: The scope and grid of the model of the coastal area in the detailed study
area are shown in Fig 1. The model range extends from Qui Nhon ±Binh Dinh to Binh
Thuan, with the open sea side about 200 km from the shore. The grid of the model is
an unstructured grid with a quadrilateral element used. In order to assess the impact of
the project on the hydrodynamic regime of the study area, the grid area of the construc-
tion area is about 10m ÷ 15m. With the total number of elements in the detailed study
area is 28496 (element).

2.2 Data and material

Topographic data are taken from various sources: Coastal area of Khanh Hoa province
and neighboring provinces using seabed topographic map of 1 / 50,000 scale provided
by the Ministry of Natural Resources and Environment. The topography of other study
areas of the East Sea is taken from the SRTM15_PLUS V1.0 data of Scripps Institution
of Oceanography at the University of California, USA. This is a data set with a resolu-
WLRQRIƎîƎDERXWP
Data of tidal water level in Qui Nhon station in the period of 1995 ± 2015; Cam
Ranh, Nha Trang in the period of 1998 - 2017 were collected and adjusted, referring to
the national elevation system.
The observational wave data from satellites of AVISO organization (France), and
wave simulation results by WAVEWATCH-III model of NCEP / NOAA organization
(USA), the wave measurements of Bach Ho station were also collected for this study.
Wind field data and background atmospheric pressure used in this study are ex-
tracted from the global climate model CFSR (Climate Forecast System Reanalysis) of
the National Centers for Environmental Prediction - National Oceanic and Atmospheric
Administration (NCEP/NOAA).

3 Calibration and validation

The model 1 is calibrated and verified according to the actual measured water level and
forecast water level of FES2014 model at 55 locations. The results are shown in the Fig
2.
To calibrate and test the model of MIKE21 SW calculating the East Sea wave, the
simulated wave results from the model are compared with the following wave data: (i)
Real wave data measured at Bach Ho station (Fig. 3), (ii) Number satellite monitoring
1090 N. D. Vo et al.

data from satellites of AVISO organization (France), and (iii) Wave simulation results
by WAVEWATCH-III model of NCEP / NOAA organization of the USA.

Fig. 2. Correlation of tide level simulated by MIKE21 model and predicted tide by global tide
model FES2014 at positions P14

Fig. 3. Comparison of wind data extracted from model 1 with wind data observed at Bach Ho
station in 1996.

Fig. 4. Comparison of simulated water level with MIKE21 model and actual water level meas-
ured at Qui Nhon station

The model 2 is calibrated versus the data at Quy Nhon Station (Fig. 4) and
WAVEWATCHIII model (Fig. 5). The results of calibration and verification of the
Typhoon Wave Modelling For Viet Nam Central Region 1091

model with the actual integrated water level at Qui Nhon station in November 2007 and
December 2009 are presented respectively. The results show a fairly high similarity
between the real measured data and simulation results.

Fig. 5. Comparison of significant wave height results of MIKE21 SW model with wave data of
WAVEWATCH-III model at position S1

4 Application.

For the purpose of simulating the impact of storm waves to the region, the study simu-
lated the impact of a 12-level storm based on LINGLING storm trajectory assuming
landings in Khanh Hoa area. Results is presented in Fig. 6.

Fig. 6. Peak height wave distribution without and with the impact of typhoon.

Fig. 7. The variation of peak wave under the impact of Typhoon.

Accordingly, under the impac of a typhoon level 12 hitting the province of Khanh
Hoa and neighboring area, the water level in the region could increase up to 1.67m and
the peak wave height could increase 6.8m compared to normal as Fig. 7 and in Table.1
1092 N. D. Vo et al.

Table 1. Water level and Peak wave height comparion in normal and typhoon condition.
Quy Nhon (Binh Dinh) Nha Trang (Khanh Hoa)
Normal Typhoon Normal Typhoon
Water Peak Water Peak Water Peak Water Peak
level wave level wave level wave level wave
0.13m 0.14m 1.31m 4.37m 0.14m 0.04m 1.81m 6.76m

5 Conclusion.

For the purpose of providing an overview of the impact of storm waves on the coast
and coastal structures in the central region, the study simulated a hypothetical storm
wave effect on this area. The simulation is performed on the modified MIKE 21 model
and tested by the measured data, satellite observation, forecast results from the global
FES2014 tide model and other models such as WAVEWATCH III. . Although there
are still some differences, from the test results can be concluded that the model simu-
lates the hydrodynamic mode of the study area with quite good accuracy. The simula-
tion results also show that the simultaneous incorporation of hydrodynamic-wave phy-
sical processes plays a very important role in the surges simulation. The simulation
results show that, with a typhoon level 12, the peak wave height can increase 6.8m
compared to normal. This shows the impact of storm waves on the area as well as coas-
tal structures. The results of the paper provide valuable consultation tools for urban
planning schemes and management, including climate change and disaster risk consi-
derations.

Acknowledgement

This research is support by the University of Danang ± University of Science and Te-
chnology in framework of 2019 project with number T2019-02-17.

References

[1] , %2$7(1* ³6SDWLDO 3ODQQLQJ DQG &OLPDWH &KDQJH Adaption in Coastal
Regions᩿7KH&DVHRI9LHWQDP´LQthe 7th FIG Regional Conference in Hanoi,
Vietnam, 2009, no. October 2009, pp. 19±22.
[2] & 'RUHO , 3ăFXUDU 6 52ù&$ ù %,/$ù&2 0 'Ì5-$ DQG $ , %2
³9XOQHUDELOLW\ $VVHVVPHQW RI /DQG IRU 6XUIDce Erosion using Usle Model.
Case Study: Someș 0DUH +LOOV´ ProEnvironment Promediu, vol. 9, no. 25,
2016.
[3] DHI, Mike 21 Flow Model-Hydrodynamic Module Scientific Documentation.
2012.
 Application of Win-TR55 for calculating designed flood
in small watersheds in mountainous region of Vietnam

(C=.<=)(

/B=J9FG=HMC:.F5BGDCFH5B8CAAIB=75H=CBG "5BC= 0=9HB5A
dtnoi@utc.edu.vn

Abstract.F5BGDCFH5H=CB=B:F5GHFI7HIF9D@5MG5G=;B=:=75BHFC@9=BH<989J9@CD

A9BHC:5B5H=CB9J9@CDA9BHC:H<=G=B:F5GHFI7HIF9=B0=9HB5A=G:57=B;8=::=

7I@H=9G 9GD97=5@@M=BACIBH5=BCIG5F95G8I9HC@57?C:F5=B:5@@5B8:@CK85H5:CF
75@7I@5H=CB C: 89G=;B98 :@CC8 88=H=CB5@@M  A5BM K5H9F 8F5=B5;9G <5J9 699B
85A5;986M<95JMF5=B5B8G9J9F9@M5::97H986M7@=A5H97<5B;9.<9DFC;F5A
1=B
.,65G98CB--A9H<C8=G5H97<B=75@DFC798IF9:CF75@7I@5H=B;GHCFA
FIBC:: JC@IA9  8=G7<5F;9  5B8<M8FC;F5D<M #B CF89FHC IG9 ., D5F5A9H9FG
:CF75@7I@5H=B;89G=;B98:@CC8=B7@I8=B;F5=B:5@@ GC=@HMD9 5B8@5B8IG9A5D5F9
F9EI=F98.<=GDFC;F5A=GK=89@MIG98=BA5BM7CIBHF=9G6IH<5GBCH699B5D

D@=98HCH<9GA5@@65G=BG=B0=9HB5A6975IG9C:85H5A=GG=B;#BH<=GGHI8M G7=

9BH=:=7 F9G95F7< CB F5=B:5@@ 5B8 89H5=@98 ;FCIB8 GIF:579 7CB8=H=CBG K5G 89J9@

CD98 IG=B; 1=B

., DFC;F5A :CF GA5@@ K5H9FG<98G  K5H9FG<98 7CJ9F=B;
5B<5H6F=8;9CB(5H=CB5@<=;<K5M=B<&5D8=GHF=7H &5B;-CBDFCJ=B79 
0=9HB5AK5GG9@97H985G575G9GHI8M.<9GHI8MG<CKG5;F95HDCH9BH=5@C:H<9
1=B
.,A9H<C8HC9GH=A5H9H<9<M8FC;F5D<:CFH<9GA5@@5B8A98=IAK5H9F

G<98G=B0=9HB5A

Keywords:1=B
., 9G=;B98:@CC8 ,5=B:5@@ -C=@HMD9 &5B8IG9 5B<5H
6F=8;9

1 Introduction

.<9G=N9C:8F5=B5;9KCF?G89D9B8GCBH<989G=;B<M8FC;F5D<HCA5?9GIF9H<9MK=@@
BCH 69 =BIB85H98 =B H<9 :IHIF9 '5BM A9H<C8G 89J9@CD98 HC 75@7I@5H9 H<9 89G=;B98
<M8FC@C;MGC:5F34 34#B;9B9F5@ H<975@7I@5H=B;C:89G=;B98<M8FC@C;M65G98CB
H<9 F5=B:5@@
FIBC:: F9@5H=CBG<=D #H75B 69 7@5GG=:=98 =BHC H<F99 ;FCIDG.<9HF58=H=CB5@
A9H<C8IG9GG=AD@99EI5H=CBGHC:CFAI@5H9H<9F5=B:5@@
FIBC::F9@5H=CBG<=D9; FIBC::
7C9::=7=9BH.<958J5BH5;9C:H<9G9A9H<C8G=GH<5HH<9M75BDFCJ=895EI=7?9GH=A5H9
C:H<989G=;B<M8FC;F5D<"CK9J9F H<9M8CBCH9LD@=7=H@M577CIBH:CFH<9=AD57HC:
K5H9FG<987<5F57H9F=GH=7G@=?9@5B8IG9 HCDC;F5D<M5B8GC=@HMD9CFK5H9FG<987CB8=

H=CBG GI7< 5G GC=@ AC=GHIF9 .<9 G97CB8 ;FCID C: A9H<C8G =G 65G98 CB D<MG=75@@M

65G98 <M8FC@C;=75@ AC89@G .<9G9 AC89@G 75B K9@@ G=AI@5H9 H<9 K5H9F 8MB5A=7G CB
H<9 K5H9FG<98 6IH F9EI=F9 5 @CH C: 85H5 H<5H A5M 69 BCH 5J5=@56@9 =B F9ACH9 F9;=CBG
.<9 H<=F8 ;FCID C: A9H<C8G IG9G 7CB79DHI5@ <M8FC@C;=75@ AC89@G .<9G9 A9H<C8G
DFCJ=895F9@5H=J9@M577IF5H99GH=A5H=CBC:F=J9F:@CK:FCAF5=B:5@@K=H<5B5::CF856@9

© Springer Nature Singapore Pte Ltd. 2020 1093

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_175
1094 N. T. Doan

F9EI=F9A9BH C: =BDIH 85H5 #B H<9G9 A9H<C8G  D5F5A9H9FG C: AC89@G 5F9 IGI5@@M 9GH=

A5H98 :FCA H<9 K5H9FG<987<5F57H9F=GH=7G IG=B; !#-HCC@G CF =HG577IF57M 5B87CB

J9B=9B79HC9GH=A5H9 H<=GGHI8MIG9GIG9CB9C:H<9A9H<C8G=BH<9H<=F8;FCID?BCKB
5G.,K<=7<K5G89J9@CD986MH<9/B=H98-H5H9G9D5FHA9BHC:;F=7I@HIF934#B
0=9HB5A A5BMGHI8=9G<5J9699B7CB8I7H98HC89J9@CDIB=H<M8FC;F5D<G CF9L5A

D@9 &9349GH=A5H98H<9DCFH56@9A5L=AIA:@CC8*' :FCAH<9DCFH56@9A5L=AIA

DF97=D=H5H=CB *'* &9 34 8C7IA9BH98 5B8 5DD@=98 8=::9F9BH A9H<C8G HC 75@7I@5H9
H<9 89G=;B <M9HC;F5D< 5B8 <M8FC;F5D< (;C 34 7CBG=89F98 H<9 =AD57H C: 7@=A5H9
7<5B;9 CB H<9 9GH=A5H=CB C: H<9 89G=;B <M8FC;F5D< "CK9J9F  GC :5F H<9F9 <5J9 BCH
699BA5BMGHI8=9GCBH<989J9@CDA9BHC:H<989G=;B<M8FC;F5D<:CFGA5@@K5H9FG<98G
K<9F9H<9H=A9C:7CB79BHF5H=CB=GIGI5@@M@9GGH<5B <CIFG6975IG9C:J5F=CIGF95

GCBG#BH<9G9K5H9FG<98G IG=B;85=@MF5=B:5@@HC7CBGHFI7HH<9<M8FC;F5D<=GBCHGI=H5

6@9'95BK<=@9 H<9G9GA5@@K5H9FG<98G577CIBH:CFIDHC C:8F5=B5;9GHFI7HIF9G

#BH<5H 7CBH9LH H<989J9@CDA9BH C:H<9 89G=;B<M8FC;F5D<:CF GA5@@ K5H9FG<98 =G5B
IF;9BH89A5B8
.<9GHI8MC:5=AG=GHC5DD@MH<9.,A9H<C8HC9GH=A5H9H<989G=;B<M8FC
;F5D<
C:5B<5HK5H9FG<98 5GA5@@ACIBH5=BCIGK5H9FG<98@C75H98=B&5B;-CBDFCJ=B79 
(CFH<9FB 0=9HB5A #B H<=G A9H<C8  H<9 -- 7IFJ9 BIA69F =G 9AD@CM98 HC 9GH=A5H9
FIBC::5B8H<9--IB=H<M8FC;F5D<=GIG98HC9GH=A5H9H<97CCF8=B5H9GC:<M8FC;F5D<
:FCAFIBC::

2 Methodology

.<=G GHI8M 9AD@CMG H<9 1=B.,

 GC:HK5F9  K<=7< =G 5 G=B;@9
9J9BH F5=B:5@@
FIBC:: 
GA5@@ K5H9FG<98 <M8FC@C;=7 AC89@ HC 75@7I@5H9 H<9 89G=;B <M8FC;F5D<G .<=G AC89@
K5G89J9@CD98=B6MH<9/B=H98-H5H9G9D5FHA9BHC:;F=7I@HIF9 CF957<GI6

5F95  5 <M8FC;F5D< =G ;9B9F5H98 65G98 CB @5B8 5B8 7@=A5H9 7<5F57H9F=GH=7G #B H<9
1=B.,
 A5GGF5=B:5@@=G7CBJ9FH98HCA5GGFIBC::6MIG=B;5FIBC::7IFJ9BIA69F
(K<=7<<5G5F5B;9C: HC ,IBC::=GH<9BHF5BG:CFA98=BHC5<M8FC;F5D<6M
IG=B;IB=H<M8FC;F5D<H<9CFM5B8FCIH=B;DFC798IF9GH<5H89D9B8CBFIBC::HF5J9@H=A9
H<FCI;<G9;A9BHGC:H<9K5H9FG<98/- .<9GHCFAFIBC::JC@IA9IG=B;H<9
--A9H<C85B87CBGHFI7HGH<9<M8FC;F5D<IG=B;H<9H56I@5F<M8FC;F5D<A9H<C8.<9
--A9H<C89GH=A5H986M9EI5H=CB69@CK

$%&.()+
𝑄= 
$,&.)

K<9F9 Q =G H<9 FIBC:: P =G H<9 F5=B:5@@ S =G H<9 DCH9BH=5@ A5L=AIA F9H9BH=CB 5:H9F
FIBC:: 69;=B S =G F9@5H98 HC H<9 GC=@ 5B8 7CJ9F 7CB8=H=CBG C: H<9 K5H9FG<98 H<FCI;<
7IFJ9BIA69FCND5F5A9H9F6M
/&&
S=  -10

(75B6989H9FA=B98:FCAGC=@HMD9 7CJ9FHMD9 HF95HA9BH <M8FC@C;=77CB8=H=CB5B8

5BH97989BHFIBC::7CB8=H=CB(5@GC89D9B8GCBK<9H<9F=AD9FJ=CIG5F95=G8=F97H@M
Application of Win-TR55 for calculating designed flood … 1095

7CBB97H98HCH<98F5=B5;9GMGH9ACFK<9H<9F:@CKCJ9FH<9DF9J=CIG5F9569:CF9F957<

=B;H<9 8F5=B5;9 GMGH9A (9LH  H<9 <M8FC;F5D< 75B 697CBGHFI7H98 :CF H<9 K5H9FG<98
IG=B; H56I@5F <M8FC;F5D< A9H<C8 5G 69@CK = =J=89 H<9 K5H9FG<98 =BHC GI65F95G
K=H< F9@5H=J9@M <CAC;9B9CIG K5H9FG<98 7<5F57H9F=GH=7G == .<9 <M8FC;F5D< 7CCF8=

B5H95HH=A9tC:5GI65F95E=G75@7I@5H986M

q=q6 A8 Q
=BK<=7<qt=GH<9H56I@5FIB=H8=G7<5F;9 K<=7<=G89H9FA=B98:FCAH=A9C:7CB79BHF5

H=CBTc5B8F5H=C69HK99B=B=H=5@56GHF57H=CB5B8F5=B:5@@Ia/PK=H<Ia SAm=GH<9
8F5=B5;95F95C:GI65F955B8Q=GH<9FIBC::CBH<9GI65F95=== =B5@@M <M8FC;F5D<
5HH<9CIH@9HC:H<9K5H9FG<98=GC6H5=B98:FCAH<9<M8FC;F5D<7CCF8=B5H9GC:GI65F95G
5B8 H=A9 C: HF5J9@ :FCA 957< GI65F95 HC H<9 K5H9FG<98 CIH@9H .<9 A9H<C8 65G98 CB
J9@C7=H=9G C: :@CK H<FCI;< G9;A9BHG C: H<9 K5H9FG<98 =G IG98 :CF 9GH=A5H=B; H<9 D5

F5A9H9FG IG98 HC 8=GHF=6IH9 H<9 FIBC:: =BHC 5 <M8FC;F5D< .KC A5>CF D5F5A9H9FG 5F9
H=A9 C: 7CB79BHF5H=CB .75B8 HF5J9@ H=A9 C: :@CKH<FCI;<H<9 G9;A9BHG .H.F5J9@
H=A9.H=GH<9F5H=CC::@CK@9B;H<HC:@CKJ9@C7=HM.H& 0K<9F9&=GH<9
:@CK@9B;H<:H0=G5J9F5;9J9@C7=HM:HG.7=GH<9GIAC:.HJ5@I9G:CFH<9J5F=CIG
7CBG97IH=J9:@CKG9;A9BHGBCH<9FA5=BD5F5A9H9F=BH<9.,
GC:HK5F9=G(#B
H<=GGHI8M H<9G9D5F5A9H9FGK9F99GH=A5H986MIG=B;!#-:FCAHCDC;F5D<M @5B8IG9
5B8GC=@HMD9A5DG

3 Data availability and Results

3.1 Study area

#B H<=G GHI8M  K9 7CBGHFI7H98 H<9 89G=;B <M8FC;F5D< :CF H<9 5B <5H K5H9FG<98  5
GA5@@ K5H9FG<98 =B &5B; -CB DFCJ=B79  (CFH<9FB 0=9HB5A  =; 5 .<9 K5H9FG<98
<5G5B5F95C:5FCIB8?A5B8575H7<A9BH@9B;H<C:?A&C75H98=B5ACIB

H5=BCIG F9;=CB  H<9 K5H9FG<98 HCDC;F5D<M =G EI=H9 GH99D K=H< 5B 5J9F5;9 G@CD9 C:
  K<=7< =B8=75H9GH<5HH<9 H=A9 C: 7CB79BHF5H=CB C: H<=G K5H9FG<98=G F9@5H=J9@M
GA5@@  ACF9 89H5=@98 89G7F=DH=CB C: H<9 K5H9FG<98 7<5F57H9F=GH=7G 75B 69 :CIB8 =B
.56@9

Table 1<5F57H9F=GH=7GC:5B7<5H6F=8;9K5H9FG<98

(C <5F57H9F=GH=7 -MA6C@ /B=H 05@I9

 9G=;BDFC656=@=HM *  
 5H7<A9BH5F95  ?A 
 '5=B:@CK@9B;H< & A 
 5H7<A9BH@9B;H< &@J ?A 
 5H7<A9BH5J9F5;98K=8H<  ?A?A 
 5H7<A9BHFCI;<B9GG    
 5H7<A9BH5J9F5;98G@CD9 -6   
 ,=J9F698G@CD9 -F   
1096 N. T. Doan

3.2 Determination of design hyetograph

.C89J9@CDH<989G=;B<M8FC;F5D<:CF5B<5HK5H9FG<98 K97C@@97H98<CIF@M85H5
CJ9FH<9D9F=C8:FCAHC 5H&5B;-CBF5=B;5I;9.<9BK989H9FA=B98H<9
A5=BF5=B:5@@9J9BHGC77IFF=B;CJ9FH<9
 D9F=C85B87@5GG=:=98H<9G99J9BHG
=BHCHKC;FCIDG9J9BHGK=H<HCH5@5ACIBHC:
<F5=B:5@@@CK9FH<5B AA.MD9#
5B8 9J9BHG K=H< 
< F5=B:5@@ <=;<9F H<5B CF 9EI5@ HC   AA .MD9 ## =;IF9G 5
5B8 5 G<CK H<9 
<CIF8=A9BG=CB@9GG 7IAI@5H=J9<M9HC;F5D<G C: F5=B 9J9BHG 
<
F5=B:5@@@CK9F5B8;F95H9FH<5B AA F9GD97H=J9@M



567

Fig. 1.IAI@5H=J98=A9BG=CB@9GG<M9HC;F5D<5B89G=;B<M9HC;F5D<=B&5B;-CB#

5G98 CB H<9G9 <M9HC;F5D<G  K9 G9@97H98 H<9 HMD=75@ <M9HC;F5D< C: 957< ;FCID 5G
G<CKB =B =; 65B8 6 =; 75B8 7 G<CKH<9 HMD=75@ =B7F9A9BH<M9HC;F5D<G C:
957<;FCID#H=G7@95FH<5H5@H<CI;<H<9HMD=75@<M9HC;F5D<GC:HKC;FCIDG5F9G=A=@5F 
H<9 D95?F5=B:5@@ C:HMD9 ## =G AI7< <=;<9FH<5B H<5H C:HMD9 # (9LH H<989G=;B 
<
F5=B:5@@J5@I9GK=H<F9HIFB:F9EI9B7=9GC:      5B8 M95FGK9F975@7I

@5H98 65G98 CB 

< F5=B:5@@ 85H5G9HG 7C@@97H98 =B H<9 
  D9F=C8 =B5@@M  H<9
89G=;B<M9HC;F5D<G7CFF9GDCB8=B;K=H<H<989G=;B
<F5=B:5@@J5@I9GK9F9C6H5=B98
6M A5DD=B; :FCA HKC HMD=75@ <M9HC;F5D<G HMD9 # 5B8 ## .<9G9 89G=;B <M9HC;F5D<G
K9F9IG985GH<9=BDIHG:CF.,HC7CBGHFI7HH<989G=;B<M8FC;F5D<G


 
567

Fig. 2.IAI@5H=J98=A9BG=CB@9GG<M9HC;F5D<5B89G=;B<M9HC;F5D<=B&5B;-CB##
Application of Win-TR55 for calculating designed flood … 1097

3.3 Land use map and Curve number map

#BCF89FHC9GH=A5H9FIBC:::FCAGHCFAF5=B:5@@ H<9.,AC89@IG9GH<97IFJ9BIA

69F(A9H<C85GDF9G9BH98=BG97H=CB K<=7<F9EI=F9GD5F5A9H9F(.<989H9F

A=B5H=CBC:(89D9B8GCBH<9K5H9FG<987<5F57H9F=GH=7G=B7@I8=B;GC=@5B8@5B87CJ

9F7CB8=H=CBG =;5 K<=7<H<9AC89@F9DF9G9BHG5G<M8FC@C;=7GC=@;FCID @5B8IG9 

HF95HA9BH 5B8<M8FC@C;=77CB8=H=CB!=J9BH<CG9K5H9FG<9885H5 H<9(A5D75B69
7CBGHFI7H98 IG=B; !#- HCC@G =; 6 G<CKGH<9 @5B8 7CJ9F5B8 ( A5DG C:H<9 5B
<5H K5H9FG<98 .<9 :=;IF9 =B8=75H9G H<5H H<9F9 5F9 H<F99 F9;=CBG K=H< ( J5@I9G C:
 5B8 =BK<=7<H<9F9;=CBGK=H<(J5@I9GC:5B8 8CA=B5H9ACGH@MH<9
K<C@9K5H9FG<98-=A=@5FHCH<9(9GH=A5H=CB H<9H=A9C:7CB79BHF5H=CB.75B8H=A9
C:HF5J9@.HK9F95@GC75@7I@5H98:FCAH<9HCDC;F5D<MA5D6MIG=B;!#-HCC@


56

Fig. 3. &5B8IG95B8IFJ9BIA69FA5DC:H<95B<5HK5H9FG<98

3.4 Determination of design hydrograph

.<9 89G=;B <M8FC;F5D< :CF H<9 5B <5H K5H9FG<98 K5G 7CBGHFI7H98 :FCA H<9 85=@M
F5=B:5@@K=H<F9HIFBD9F=C8GC:      5B8 M95FGIG=B;H<9.,A9H<

C8I9HC=HGGA5@@5F95 H<95B<5HK5H9FG<98CB@M=B7@I89GCB9GI6
65G=B5B8CB9
@5F;9F957<.56@9GIAA5F=9GH<9D95?8=G7<5F;95HGI665G=B5B8CIH@9HC:H<9K5H9F

G<98 =;6G<CKG5B9L5AD@9C:H<989G=;B<M8FC;F5D<5HH<9CIH@9HC:H<9K5H9F

G<987CFF9GDCB8=B;K=H<5F9HIFBD9F=C8C: M95FG

Table 2"M8FC;F5D<D95?H56@9C:5B7<5H6F=8;9K5H9FG<98

*95?:@CK6M,5=B:5@@,9HIFB*9F=C8
-I6
+ + + + + + +
F95 7AG 7AG 7AG 7AG 7AG 7AG 7AG

MF 
MF 
MF 
MF 
MF 
MF 
MF
-I6          
)IH9H         
 
1098 N. T. Doan


56

Fig. 4.15H9FG<98C:5B<5H5B89G=;B<M8FC;F5D<

4 Conclusion

#BH<=G GHI8M  H<9 ., A9H<C8 K5G9LD@CF98 5B8 5DD@=98HC9GH=A5H9H<9 89G=;B <M

8FC;F5D<G K=H<F9HIFB D9F=C8G :FCA  HC   M95FG.<9=B=H=5@ F9GI@HG G<CKH<5HH<=G

A9H<C8=GGI=H56@9:CFH<975@7I@5H=CBC:H<989G=;B<M8FC;F5D<:CFGA5@@5B8A98=IA
K5H9F
G<98 *5FH=7I@5F@M  H<975@7I@5H=CB C: H<9H=A9 C:7CB79BHF5H=CB=BH<=G A9H<C8 
K<=7< 577CIBHG :CF H<9 H=A9 C: HF5J9@=B; 6CH< CB 5 <=@@ G@CD9 5B8 H<9 F=J9F  =G ACF9
577IF5H9H<5B H<9 HF58=H=CB5@ A9H<C8G.<9577IF57MC: FIBC:: G=AI@5H=CBIG=B;7CA

6=B98!#-5B8.,K5G5@GC7CB:=FA98=BH<9GHI8MC:,5A5B534.<9A9H<C8C:
., G<CKG 5 ;F95H DCH9BH=5@ HC 69 5DD@=98 HC ;9B9F5H9 <M8FC;F5D<G 5B8 89J9@CD 5
85H565G9C:<M9HC;F5D<G5B8<M8FC;F5D<G:CFCH<9FACIBH5=BCIGF9;=CBG<5J=B;G=A=

@5F7CB8=H=CBG=B(CFH<9FB0=9H(5A"CK9J9F =HG<CI@869BCH98H<5HH<9G=AD@=:=75

H=CBGC:., CBH<9CB9<5B8 G<CK=HG95G9C:5DD@=75H=CBG CBH<9CH<9F<5B8 H<9G9

G=AD@=:=75H=CBG @=A=H H<9 IG9 C: H<9 DFC798IF9G 5B8 75B DFCJ=89 F9GI@HG H<5H 5F9 @9GG
577IF5H9H<5BACF989H5=@98A9H<C8G

References
 <CK0.DD@=98"M8FC@C;M#B"M8FC@C;M
D5;9G
 <CK0."5B86CC?C:DD@=98"M8FC@C;M
 &905B(;<=B<"M8FC@C;=75@DF=B7=D@9G;F=7I@HIF5@*I6@=G<=B;#B7 
 &9=B<.<5B<GH=A5H=CBC:*'*5B8*' :CF0=9HB5A*<=GG9FH5H=CB"5BC=15H9F
/B=J9FG=HM
 (;C &9 &CB; -HI8M CB G7=9BH=:=7 65G=7G :CF DFCDCG=B; GH5B85F8G C: 89G=;B98 :@CC8  G95
8M?9G=BH<97CBH9LHC:7@=A5H97<5B;9 G95@9J9@F=G9=B0=9H(5A5B8A=H=;5H=CBA95GIF9G

 ,5A5B5 !09B?5H5,5=B:5@@FIBC::AC89@=B;69HK99B.,
<M8FC@C;=7K5H9FG<98AC8

9@ 5B8CJ9F@5B8H=A9C: 7CB79BHF5H=CB AC89@  ,'(  #BH9FB5H=CB5@ $CIFB5@ C: 8

J5B79G=BCADIH9F-7=9B795B8.97<BC@C;M  '5F7< 

 /B=H98-H5H9G9D5FHA9BHC:;F=7I@HIF9/F65B"M8FC@C;M:CF-A5@@15H9FG<98G.97<

B=75@F9@95G9.,
 Using HCM Method in estimation the highway capacity
and recommendation traffic management for a section of
national highway #5 - VietNam

Dr. Tran Trung Hieu1, M.Sc. Trinh Xuan Lam1

1
University of Transport Technology, 54 Trieu Khuc, Thanh Xuan, Hanoi, Vietnam
trantrunghieu@utt.edu.vn

Abstract: The HCM method estimated the highway capacity base on the speed-
flow relationship in the ideal conditions and other data collected in many state
highways in the US, which support to define the capacity in conjunction with
the level of services that a road or a highway facilities will adapt or could serve
in the specific conditions of traffic, terrain, and geometry. In order to define the
level of services and estimate the capacity of a specific roadway, the traffic and
infrastructure data collection will be implemented firstly, then the highway will
be divided by the sub-sections which have typical characteristics to estimate the
capacity. This paper describes the data collection in the section of KM 55+582
to KM77+830 of National Highway 5 and briefs the analysis and estimation the
capacity and level of services of study section base on the Highway Capacity
Manual approach and recommends the improvement solutions.

Keywords: HCM, highway capacity, management of transportation, FFS, free

flow speed.

1 Introduction

There are several estimation methods of United States ( USA), Europe and Russia
were being applied in the word wide, in which the Highway Capacity Manual-HCM
[1,2,3,4], a method comes from the United States is the most widely used. The HCM
method considers the estimation of capacity and the speed-flow relationship in the
ideal conditions and other data collected in many state highways in the North Ameri-
ca, In order to define the level of services and estimate the capacity of a specific
roadway, the traffic and infrastructure feature data collection and evaluation of study
section will be implemented firstly, then the highway will be divided by the sections
which have same natures to estimate the capacity by the sections.
The article describe the evaluation the existing conditions of National Highway 5-
NH5 section KM 55+582 to KM77+830 and the analysis and estimation the capacity
and level of services of study section base on the Highway Capacity Manual ap-
proach.
The HCM model introduces the speed-flow curve (shown in Fig. 4) is a series of
free flow speed, which were developed base on the statistical data collecting in many
state highways in United States.

© Springer Nature Singapore Pte Ltd. 2020 1099

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_176
1100 T. T. Hieu and T. X. Lam

HCM 2010 defines capacity of a system element is the maximum sustainable hour-
ly flow rate at which persons or vehicles reasonably can be expected to traverse a
point or uniform section of a lane or roadway during a given time period under pre-
vailing roadway and traffic conditions [3,5]. It is interesting that HCM complements
to capacity a concept of Level of Services (LOS). LOS is a quantitative index to
measure the quality of traffic flow. The HCM defines six levels of service ranging
from A to F, for each service measure, or for the output of a mathematical model
based on multiple performance measures. The capacity will be the maximum of flow,
in which the level of services are accepted (normally level D or E).

2 Capacity analysis in the study section

2.1 Brief of National Highway #5

National Highway #5- NH5 is trunk road connecting the two biggest cities in North-
ern Vietnam, Hanoi City and Haiphong City. Despite the new parallel expressway
became in operation since 2016 the traffic volume in the NH5 is quite large due to the
low operation cost and easy in accessibility than the new expressway. The study sec-
tion (KM 61 to KM 67+240) is a dual lane highway and has 2 lanes for 4 wheel vehi-
cles and 1 lane for 2 wheel vehicles in each direction.

Fig 1. The study section map.

The traffic volume were collected form the traffic counting done by the road mainte-
nance company twice per years since 2014 , in which each counting set is carried out
in 3days with two days counting in 16 hours and 1 day in 24 hour. The traffic count-
ing are classified by the vehicles type such as passenger car, light truck, heavy truck,
mini-bus, conventional bus, motorcycle, bicycle and so on.
Using HCM Method in estimation the highway capacity … 1101

2.2 Free flow speed measurement

As mentioned above, free flow speed (FFS) is basic factor in capacity analysis by the
HCM method. FFS can be determined directly from field measurements or can be
estimated by the formulas [6]. FFS is the mean speed of passenger cars measured
during periods of low to moderate flow (up to 1,400 pc/h/ln - passenger cars per hour
per lane). If the FFS can be field measured, that determination is preferable.
In the case of the study, the field measurement of FFS was implemented twice in
March and June 2018. The test car is passenger car (saloon type) and speeds was de-
termined by the both the car speed meter and the application software Speedometer in
smartphone. The speeds were measured by the chainage with the space of each meas-
urement about 300-500 m. There are about 30 samples was taken by the each direc-
tion. The measurements were combined with the observation the drive to define the
impact of the other car to their performance.
Main characteristic of FFS field measurement as follow: FFS was impacted by the
local driver behavior (passing cars, two wheel vehicles crossing), the impact of high
proportion of large vehicles; Safety space with other cars is about 70m (base on the
drive experience). It is equivalent to the speed of more than 60 km/h; FFS was im-
pacted by the limit speed in NH5 (80 km/h), in addition the road conditions (the
roughness and friction) did not allow the speed higher than 80 km/h.

2.3 Data refinery

Elimination the over margin data: The data will be refined to exclude those did not
reflect the nature of free flow speed. Hence the speeds that are smaller than 60 km/h
will be excluded.
Data analysis: The collected data will be analyzed by the statistical method to pro-
vide the distribution and frequency for comparing with the average speed (as HCM
recommended approach) by the chainage. The free flow speed by the chainage in both
direction of study section had been described in the below charts (Fig. 2).

Fig. 2. Free flow speed by the chainage in Hanoi-Haiphong outbound: Saverage= 66.9 km/h (left
figure); and in Haiphong- Hanoi outbound: Saverage= 67.3 km/h (right figure)
1102 T. T. Hieu and T. X. Lam

Table 1. Speed parameter on the study section.

3 Results and conclusion

3.1 Comparison FFS between the survey and the estimation

The value of measured free flow speed (in yellow lines) were compared with the val-
ue of free flow speed that estimated by the formula in the Chapter 14 of HCM 2010
(in blue lines) are shown in Fig. 3.

Fig. 3. Hanoi- Haiphong outbound (left figure) and Haiphong-Hanoi outbound (right figure).

3.2 Summary the capacity analysis of level of services by the chainage of study
section

Base on the adopted parameters of flow and free flow speed, applying to HCM's mod-
el in the Fig. 4, the level of service (LOS) and capacity of the study section is defined
and shown in the Table 2.

Table 2. Summary the capacity analysis of level of services by the chainage of study section.
Using HCM Method in estimation the highway capacity … 1103

Fig. 4a. Capacity Estimation by directions Hanoi-Haiphong of study section.

Fig. 4b. Capacity Estimation by directions Haiphong- Hanoi of study section.

Please note that the capacity and LOS was estimated by lanes of 4 wheel vehicle and
2 wheel vehicle in considering the practice of limited mixing these vehicles during the
field study implementation. The LOS of two wheel vehicle applies the value collected
in [7,8].
1104 T. T. Hieu and T. X. Lam

3.3 The recommended of motorcycle lane

The motorcycle path is introduced in the existing 2- wheel vehicle lanes to separate
the 4- wheel vehicle and 2- wheel vehicles that support the safety and smoothness of
traffic flow as well as the capacity of the major traffic flow. The separated barrier type
of plastic, i.e., the tuff curb is recommended.
In conclusion, the HCM methodology can be applied in conditions of mix traffic in
Vietnam but need to consider some disadvantages of the free flow speed- FFS may
not reach the value that is instructed by HCM and the lane discipline. The motorcycle
lane is introduced as a solution to improve the road safety and lane discipline.

Reference
1. Highway Capacity Manual (HCM) edition 2010, Transportation Research Board 2010,
Washington DC.
2. Lighthill, M.J., Whitham, On kinematic waves. Part II: A theory of traffic flow on long
crowded roads, G.B. (1955), London.
3. Roger P. Roess and Iena S. Prassas, The Highway Capacity Manual: A Conceptualand Re-
search History, Volume 1: Uninterrupted Flow, Springer International Publishing Switzer-
land 2014.
4. Roger P. Roess and Iena S. Prassas and William Mc Shane, Traffic Engineering, Pearson
Prentice Hall 2004, New Jersey.
5. Assessment of Highway Capacity Estimation Methods, Transport Research Record 1572,
Transportation Research Board, Washington DC.
6. Volume VI: Definition, Interpretation, And Calculation Of Traffic Analysis Tools
Measures of Effectiveness, US Department of Transport, Federal Highway Administration
2007.
7. Sulkiflee Mamaa, and Pichai Taneerananon, Effective Motorcycle Lane Configuration
Thailand: A Case Study of Southern Thailand, Engineering Journal Vol 20.
8. Hussain H, Radin Umar R.S, Ahmad Farhan M.S, Estimating Capacity for Unin-
terrupted Motorcycle path in Malaysia, Dadang M.M, 2004.
 Green infrastructure modelling for assessment of urban
flood reduction in Ho Chi Minh city

Nguyen Hong Quan1, 2, Nguyen Duy Hieu1, Tran Thi Van Thu1, Matthew Buchanan3,
Nguyen Duc Canh 4, Mariana da Cunha Oliveira Santos 5, Pham Dang Manh Hong
Luan1, Tran Tuan Hoang6, Huynh Luu Trung Phung7, Khuu Minh Canh8, Mitchel
Smith3
1
Center of Water Management and Climate Change (WACC), Vietnam National University ±
Ho Chi Minh City (VNU-HCM),Vietnam
2
Institute for Computational Science and Technology, Ho Chi Minh City, Vietnam
3
TUFLOW FV, BMT group, Brisbane, Australia
4
Faculty of Environment and Labour Safety, Ton Duc Thang University, Vietnam
5
Institute of City Planning and Urban Design, University of Duisburg-Essen, Essen, Germany
6
Sub-Institute of Hydrometeorology and Climate Change, Ho Chi Minh city, Vietnam
7
Department of Science and Technology Ho Chi Minh city, Vietnam
8
Center of applied GIS, Department of Science and Technology Ho Chi Minh city, Vietnam
nhquan@wacc.edu.vn

Abstract. Urban flooding has become a severe problem in Ho Chi Minh City.
Green infrastructure (GI) have been considered as a promising solution. GI is
not only beneficial for urban flood reduction but also improved microclimatic
conditions, landscape values. Recently, with the advancement of science and
technology, the mathematical models have been developed and applied in vari-
ous fields including urban flood. This paper presents the initial results of apply-
ing 2 urban water models, namely Tuflow and SWMM (Stormwater manage-
ment model). The Tuflow mathematical model was used to simulate 2 dimen-
sional (2D) surface model while the SWMM was developed for 1 dimensional
(1D) surface and drainage water system. These models are later served in as-
sessing the effectiveness of green infrastructure development in urban flood re-
duction. The models were developed at a specific study area (Tham Luong -
Ben Cat Catchment) in Ho Chi Minh city. Results can be used for urban plan-
ning especially for urban flood management in Ho Chi Minh city.

Keywords: 1, 2 D urban flood model, green infrastructure, Ho Chi Minh city,

Tuflow, SWMM.

1. Introduction

Urban flooding is becoming an increasingly important issue, which has caused devas-
tation and economic losses. The 21st century has already seen many large-scale food-
related disasters in Hanoi, Ho Chi Minh city, Vietnam (2008); Seoul, Korea (2011);
Guangdong, Beijing, China (2007); Bangkok, Thailand (2011); Taipei, Taiwan

© Springer Nature Singapore Pte Ltd. 2020 1105

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_177
1106 N. H. Quan et al.

(2001); and New Orleans, USA (2005). Disaster statistics appear to indicate that flood
events are becoming increasingly frequent. The numbers of people and financial,
economic and insured damages affected by floods have also increased. 178 million
people and total losses of exceeded $40 billion were reported to be affected by floods
in 2010 alone [1].
The main cause of flooding in the city can be explained by increased impervious
surfaces as a result of urbanization creating higher volume of surface runoff. Impervi-
ous land surface blocks rainfall infiltration flow. In addition, human activities such as
reducing trees, and constructing buildings, roads and drain systems change urban
hydrological conditions.
Efforts to prevent flooding have been made using different approaches. In the past,
cities typically focused on extensive flood-control infrastructure, such as channeliza-
tion, pump station, dams, levees, or large-scale retention systems. However, these
practices can be enormously expensive and maybe not able to cope with combined
pluvial ± fluvial flood. Recently Green infrastructure (GI), refers to the use of the
natural landscape, instead of engineered structures, to capture and treat rainwater
where it falls, is gaining interest as a new approach to stormwater management. The
approach is not only sustainable and environmentally friendly, but cost-effective as
well. GI includes, but is not limited to, green roofs, downspout disconnection, trees
and tree boxes, rain gardens, vegetated swales, pocket wetlands, infiltration planters,
vegetated median strips, curb extensions, permeable pavements, reforestation, and
protection and enhancement of riparian buffers and floodplains. GI can alleviate local
urban flooding by minimizing runoff volume and peak discharges. In addition, en-
hancing water quality, water saving, groundwater recharge, energy saving and im-
proving urban aesthetics are other benefits of GI [2].
Developing efficient GI could overcome the problems in Ho Chi Minh City. De-
spite growing enthusiasm for these practices, a number of obstacles and challenges to
integrating GI into flood reductions have been identified. Many observers believe that
the biggest barriers are lack of information on performance and cost-effectiveness,
and uncertainty whether green GI will contribute to achieving reduction of flooding
and water quality improvements [3, 4]. Therefore, examining the effectiveness of GI
for a local area is necessary.
Software tools have been used for water resource management since the mid-1960s
and the modeling tools that have the ability of simulating the stormwater runoff quali-
ty and quantity started emerging from 1970s [5]. After the GI controls were identified
as a suitable solution of managing urban stormwater, these tools were updated with
the components that can evaluate the effectiveness of GI practices. Outputs of these
modeling tools include runoff volume reduction, reduction of runoff rate, and reduc-
tion of pollutant loading, due to the implementation of different GI practices [6, 7].
In this study, we apply several mathematical models i.e. SWMM (1 dimension ± 1 D)
and Tuflow (2 dimension ± 2 D) to examine the effectiveness of GI for case study of
Tham Luong ± Ben Cat catchment, Ho Chi Minh city. The catchment is is facing
serious flooding problems.
Green infrastructure modelling for assessment … 1107

2. Methodology

2.1. Study areas

Figure 1 depicts study area and its location. The study area of approximately 1,220 ha
locates in the Northern part of the city and entirely within Tham Luong ± Ben Cat
catchment in Ho Chi Minh city. Cho Cau is the main channel which divides the area
into half approximately7KHXSSHUKDOIEHORQJVWRGLVWULFW¶VWHUULWRU\DQGWKHORZHU
half covers part of Go Vap and Tan Binh districts. Land-use in the study area is fea-
tured predominantly by residential one, while other include lands for public uses,
open waters, trees and roads (Figure 1).

Fig. 1. Map of study area

2.2. Defining Green Infrastructure areas
In the context of urban complexity, a holistic approach over the urban settlement is
made in order to identify opportunities for defining GI employment. Land use and
built morphology data is compared with spatial modeling of flood scenarios, precipi-
tation data and topography in order to select GI solutions suitable for this specific
context [8].
2.3. Tuflow Model
Tuflow Model is used to simulate 2D flow caused by tide and rainfall. It is a nu-
merical engine for simulating free-surface water flow for urban waterways, rivers and
other types of surface water. Third party Graphical User Interfaces (GUI) and Geo-
graphic Information Systems (GIS) are used to provide the modeling environments. In
this study, QGIS is used as a support tool to import and retrieve results.
Spatial data for model development includes Digital Elevation Model(DEM) and
land use. The DEM data is extracted from Lidar image with the resolution is 1m.
Temporal hydro-meteorological data are collected local agency.
1108 N. H. Quan et al.

2.4. EPA-SWMM
EPA-SWMM Ver 5. is a well-known package in urban storm simulation. The model
was developed by Huber in 1971 [9] and revised by the US Environmental Protection
Agency to upgrade into Ver 5.1 by 2015. EPA-SWMM is a computer program to
simulate hydrological, hydraulic and water quality of closed/open drainage systems in
urban and rural areas. The model can simulate both an individual rainfall event and
continuous events. We collected drainage data from local agency to setup the model.

3. Results

3.1. Tuflow Model

Hydraulic models typically have water level boundaries at the downstream and
flow boundaries at the upstream. Water level boundaries representing the ocean or a
lake are specified using 1D or 2D HT (water level-time) or HS (sinusoidal water lev-
el) boundaries. HS for a sinusoidal tide and a HT for the storm surge component, the
combination giving a storm tide). For flood models, occasionally, an upstream water
level boundary is used in the absence of reliable river flow estimates. In this study,
HT boundaries are assigned for both upstream and downstream. Boundary data is
extracted from Mike 11[10].
The model is calibrated reasonably with observed data (less than 10% differences).
For scenarios development, we applied the 3-hour designed rainfall event and several
GI measures in the areas as scenarios (mostly based on utilizing opened space). The
results are showed in figure 2.

Fig. 2. Results of Tuflow model simulation with (right) and without GI (left)
3.2. EPA-SWMM
The EPA ± SWMM model was setup for the area with detail drainage system. The
model is calibrated with observed data. After that, we implement GI solutions includ-
ing rain barrels, permeable pavement, bio-retention cell or rain garden, which are
implemented at each subcatchment according to classified structure types. The results
of flood depths at serve flooded locations are showed in table 1.
Green infrastructure modelling for assessment … 1109

Fig. 3. SWMM model settings and flooded level with (right) and without GI (left)
Table 1. Flood reduction with and without GI

Model Scenario Node Node Node Node Node

68 66 52 7 6
No GI (Calibration) 0.55 0.54 0.42 0.40 0.35
Rain Barrel (RB) 0.54 0.53 0.41 0.38 0.33
RB + Permeable Pavement (PP) 0.39 0.39 0.31 0.21 0.18
RB + PP + Bio-Retetion Cell/Rain Garden 0.29 0.29 0.24 0.21 0.15
% Reducing (with GI/without GI) 47 46 41 47 56

The model results presented in the table indicate the expectation of the flooding
reduction. For example, the flood depth of Node 68 is reduced 47% by 03 GI controls
(Rain Barrel + Permeable Pavement + Bio-Retetion Cell/Rain Garden ) in the Subcatchment
68.

4. Conclusion

This paper presents the results of applying the SWMM (1D) and Tuflow (2D)
models of GI simulation for urban flood reduction in Tham Luong-Ben Cat catch-
ment, Ho Chi Minh city. Both models have certain advantages in coupling GI solu-
tions. While Tuflow model can provide visually demonstration of effects of GI (by
incorporating opened space), the SWMM offered various GI solutions including
measures at household scales. The models were calibrated with observed data with
reasonable confidence. Given the identified possible GI solution to be implemented in
the areas, the models demonstrate how floods can be reduced. Results can be used for
GI planning to reduce flooding as well as other functions such as improved microcli-
matic conditions, landscape values in Ho Chi Minh City.
1110 N. H. Quan et al.

Acknowledgments

The authors acknowledge and appreciate the inputs provided through the recent
iniWLDWLYH RQ ³*UHHQ ,QIUDVWUXFWXUH´ VXSSRUWHG E\ WKH +R &KL 0LQK 'HSDUWPHQW RI
Science and Technology through the Institute for Computational Science and Tech-
nology (ICST) (contract number: 23/2017-+Ĉ-KHCNTT). We thank TUFLOW FV,
BMT group for giving the software available.

References
1. Jha, A., et al., Five Feet High and Rising: Cities and Flooding in the 21st Century. 2011,
Policy Research Working Paper 5648, World Bank: Washington DC.
2. Taylor, W.J., Stormwater Management Program Effectiveness Literature Review: Low
Impact Development Techniques. 2013, Prepared for Association of Washington Cities
and Washington State Department of Ecology.
3. Pauleit, S., R. Ennos, and Y. Golding, Modeling the Environmental Impacts of Urban
Land Use and Land Cover Change²A Study in Merseyside, UK. Landscape and Urban
Planning 2005. 71(295±310).
4. James, P., et al., Towards an Integrated Understanding of Green Space in the European
Built Environment. Urban Forestry & Urban Greening 2009. 8: p. 65±75. 5
5. Zoppou, C., Review of urban storm water models. Environmental Modelling & Software,
2001. 16(195±231).
6. Jayasooriya, V.M. and A.W.M. Ng, Tools for Modeling of Stormwater Management and
Economics of Green Infrastructure Practices: a Review Water Air Soil Pollut, 2014. 225:
p. 2055.
7. Haris, H., et al., Urban Stormwater Management Model and Tools for Designing Storm-
water Management of Green Infrastructure Practices. IOP Conference Series: Earth and
Environmental Science, 2016. 32(1): p. 012022.
8. Mariana da Cunha Oliveira Santos, Water sensitive urban design: Addressing flooding re-
silience in Ho Chi Minh City in IFoU 2018: Reframing Urban Resilience Implementation:
Aligning Sustainability and Resilience session Urban Design and Management: Infrastruc-
tures and Services. 2018.
9. +XEHU : ³6WRUP :DWHU 0DQDJHPHQW 0Rdel, Volume I - )LQDO 5HSRUW´ (3$ 5HSRUW
11024 DOC 07/71 (NTIS PB203289). 1971, Metcalf and Eddy, Inc., University of Florida,
and Water Resources Engineers, Inc., U.S. Environmental Protection Agency: Washing-
ton, DC.
10. Nguyen Hong Quan, et al., Water quality modelling for impact assessment of industrial-
ized and urbanized activites in downstream of Dong Nai river, in VNU - HCM Interna-
WLRQDO &RQIHUHQFH IRU (QYLURQPHQWDQG 1DWXUDO 5HVRXUFHV ,&(15  ³*UHHQ JURZWK
climate change and protection of endangered FRDVWDO UHJLRQV´   -19/6/2014, Ho
Chi Minh city, Vietnam.
 Review of Unmanned Aerial Vehicles (UAVs) Operation
and Data Collection for Driving Behavior Analysis

H.Q. Pham1, M. Camey2, K.D. Pham3, K.V. Pham4, L.R. Rilett1

1
University of Nebraska - Lincoln, 330G Prem S.Paul Research Center, NE 68588 USA
2
University of Nebraska ± Lincoln, Bureau of Business Research, NE 68588 USA
3 University of North Texas, 1155 Union Circle, TX 76203 USA
4
University of Transport and Communication, 3 Lang Thuong, HN, Vietnam
huong.pham@huskers.unl.edu

Abstract. In recent years, several innovative transportation data collection tech-

nologies have developed, including global positioning satellites, Bluetooth de-
tectors, video detectors, lidar detectors, and radar detectors. While these technol-
ogies have been useful for data acquisition of both points (e.g., volume counts at
a given location) and space data (e.g., travel time between two locations), they
are still limited to gathering data on multiple vehicles over both time and space.
This paper will examine the use of unmanned aerial vehicles (UAV) for trans-
portation data collection. First, a literature review will be provided over the
framework of UAV operations and current popular platforms with a focus on
their capabilities, costs, and limitations. The review emphasis on the use of UAVs
for speed behavior analysis, gap acceptance and merging behavior. Lastly, the
challenges and consideration of UAV technologies for collecting traffic data in
developing countries will be discussed with applications on Vietnam situations.

Keywords: UAVs, Drones, Driving Behavior.

Introduction
Unmanned aerial vehicles (UAVs), also known as drones, are among the most advanced
robotics technology for disaster remediation. They are often employed with other nav-
igation assistant robots including unmanned ground vehicles (UGVs), unmanned ma-
rine vehicles (UMVs), and unmanned surface vehicles (USVs) [1]. UAVs are defined
as semi-autonomous or fully autonomous aircraft that can support and transport cam-
eras, sensors, communication equipment or other payloads. Cost-effective UAVs with
advanced sensing technologies are quickly being adopted by diverse industries, includ-
ing intelligent transport system (ITS) [2].
UAVs advantages have been recognized in multiple applications in transportation
engineering. One of the most popular applications of UAVs among US DOTs projects
is bridge inspection [3]. UAVs play an important role on traffic surveillance and mon-
itoring [4], traffic congestion and safety [5]. UAV Data acquisition can be real-time or
offline based on project requirements. The majority of studies mentioned in this paper
including Gu et al. (2019) [5], Zhou et al. (2019) [6], Salvo et al. (2015) [13] used

© Springer Nature Singapore Pte Ltd. 2020 1111

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_178
1112 H. Q. Pham et al.

offline approaches in which the video data is acquired and refined after the UAV flight
completion. Sharma et al. (2017) [7] employed real-time data acquisition and pro-
cessing techniques to develop a new approach for driver reputation calculation.
This paper provides an overview of common operational framework and platforms
for UAV operations in section 1 and 2. Section 3 shows different UAV data collection
approaches for driving behavior analysis. Sections 4 will discuss current challenges in
UAVs operation and consideration for UAVs application in Vietnam (VN).

1 Frameworks for UAV operations

A literature review found several sources related to UAV operations frameworks. In

accordance with previous studies, Khan et al. (2017) [8] proposed a complete frame-
work with seven stages for UAV operations: (i) scope definition, (ii) flight planning,
(iii) flight implementation, (iv) data acquisition, (v) data processing and analysis, (vi)
data interpretation and (vii) optimized traffic application. The first module involves the
definition of the scope of the study, including objective, network elements and perfor-
mance measures. It could be an intersection, a roadway segment, a ramp or a combina-
tion of them. The type of traffic parameters to be derived from the UAV videos also
define the type of UAV flight. Flight planning stage can be classified into three main
categories: safety, environment, and path planning aspects. This stage outlines the
equipment to be used in the operation, flight path. In flight implementation stage, the
UAV flight over the study area is conducted. Based on the characteristics of the opera-
tion, flights can be conducted manually or autonomously. The data acquisition can be
performed online, in real-time, or offline. The data processing and analysis can be car-
ried out using semi-automated video analysis and fully automated video analysis. In the
data interpretation stage, the data visualization techniques are applied. The optimization
stage helps to improve the existing traffic models with the study-specific parameters.

2 UAVs Platforms

Much work has recently been conducted to categorize UAVs platforms. Generally,
UAVs can be divided into three types: fixed-wing, rotorcraft, and vertical takeoff and
landing (VTOL) vehicles [2]. Fixed-wing UAVs fly similarly to a traditional aircraft,
and airframes can vary just as much as traditional aircraft. Multirotor UAVs are de-
signed similar to helicopters but propelled by multiple rotating propellers attached to
arms extending from the UAV body. VTOL UAVs can be considered as a combination
of multirotor and fixed-wing designs and its platform type is still under development.
In general, the fixed-wing flight is much more effective in covering large areas, ide-
ally for mapping applications or kilometer-scale measurements. Multirotor UAVs have
flexible mobility and the ability to hold the position and rotate, so they are perfect for
applications requiring high accuracy vehicle placement (see details in table 2). The US
FAA defines small UAVs as those weighing less than 25 kg (55 lbs)[10]. Table 1 lists
several UAV platforms, the vehicles weight, altitude, and payload can vary widely [11].
Review of Unmanned Aerial Vehicles (UAVs) … 1113

Table 1. Most Common UAV Platforms ± adapted from Jalayer et al. 2019 [11]

Cate- Power Weight Endur- Altitude capacity

UAVs
gory source (lb) ance (ft)

DJI Mavic Rotary Battery 1.6 27 m 16,404

DJI Phantom 4 Rotary Battery 3.0 28 m 19,685

Trimble UX5 Fixed Battery 2.5 50 m 16,404

Table 2 shows cost, advantages and limitations of four main UAV platforms that were
tested. Those included Bergen hexacopter ± a six bladed multi-rotor UAV, a small DJI
Phantom Vision 2 quadcopter, a smaller and more agile Blackout mini H quad UAV
for confined space inspection and micro-UAV platforms [12].

Table 2. Applications, Cost, Advantages and Limitations of 4 UAV platforms

Bergen Hexa- DJI Phantom

Mini H quad Micro-UAVs
copter quadcopter

Images

traffic monitor- Flying through Send live video of

High resolu-
Applications ing, construc- and providing the insides of
tion image
tion sites real-time video pump stations

$799, real-time $382 almost

$5400, set of video via Wi-Fi ready to fly
Cost (2014) < $200
spare batteries extender and $430 with a sin-
smartphone app gle camera

Walkera micro
flies back safety ability to highly agile and
camera transmit
Advantages "home" in lost enable "fly to customizable in
the video live to a
connection KRPH´PRGH confined spaces
smart devices

(FAA) sug- Shorter flight

Without sen- Shorter flight time
Limitations gested heights WLPH DW ¶
sors at 10 minutes.
of up to 250 ft more difficult
1114 H. Q. Pham et al.

3 UAVs Data Collections for Driving Behavior Research

3.1 Speed behavior on freeway loops

Zhou et, al. (2019) [6] incorporated civil UAVs to implement an effective method to
extract speed data and lateral offsets. Vehicular movements and truck types on three
freeway loops (radius 60m) were recorded by drone at a height of 140 to 180 meters.
The vehicle location in the predecessor frame were firstly predicted by Kalman filter
and Hungarian algorithm. Then the longitudinal speed data were extracted automati-
cally by the improved vehicle detection algorithm in Piotr, a MatLab computer-vision
toolbox. The steps for extracting vehicle trajectory as follows: (i) extract each frame
from the video, (ii) detect the moving objects using pixel subtraction, (iii) assign detec-
tions with a specific ID, (iv) determine similarity between consecutive frame and track
vehicles, and (v) record and export trajectory data. The lateral offsets were derived by
a semi-automatic distance-measure program which requires manual clicking on the
front right tire and the nearest inner side road pile. This helps preventing the noises by
drone vibration. Mentioned by the author, the result of the study including speed-radius
distribution and the centrifugal force coefficient CFC values could serve for further
studies of autonomous vehicles (AVs) and non-AV vehicles in mixed traffic flow.

3.2 Gap acceptance, driver aggressiveness and merging behavior

Salvo et al. (2015) [13] implemented UAV flights at an altitude of 60 meters above an
urban intersection near the University of Palermo, Italy to evaluate the gap acceptance
and driver aggressiveness. The study used a probe vehicle to compare the velocity pro-
files from GPS (real data) and micro UAVs. The results from v-t trend analysis showed
a very similar pattern between two acquired speed profiles with just small deviations.
It is verified the reliability of UAVs approach. However, the authors mentioned the
UAVs approach can be used as an additional method for traffic data acquisition but can
not completely replace other traditional methods of traffic detectors. Gu et al. (2019)
[5] employed only one UAV (DJI Inspire 1) for trajectory data acquisition to analyze
merging behavior at a 220m length acceleration lane. The UAV video recorded under
some conditions: the wind force below grade 4, sunny weather, no electromagnetic in-
terference with a total length of 410m to catch the vehicles movement during the peak
and non-peak periods. In data processing and analysis stage, the vehicle trajectory data
were extracted from the videos using the Tracker, an open source video analysis and
modeling tool. The error caused by the projection of the videos could be eliminated
through the calibration process using the length of lane marking.

3.3. Multi UAVs applications

Sharma et al. (2017) [7] proposed a driver behavior detection system using multi-UAVs
in an urban environment. The goal was to develop a new approach for driver reputation
Review of Unmanned Aerial Vehicles (UAVs) … 1115

calculation. A total of five UAVs (rotor wing) were operated over 2.5 GHz radio fre-
quency and 4.5 channels to achieve the accurate judgement of driver activity. The sys-
tem performance analysis was conducted by following four steps: i) network formation
analysis of the vehicles and the UAVs, ii) Simulation-based analysis, iii) Behavior de-
tection, and iv) real-time vehicle tracking and reputation marking of vehicles. For data
acquisition, the authors suggested the increase in the number of UAVs implemented,
the research approach provides better reliability and prolonged connectivity.

4 UAVs operational regulations and considerations for Vietnam

The US FAA released the first Section 333 Exemption [14] in September 2014. The
regulation then reviewed by the 2018 FAA Reauthorization Act for non-hobby UAV
users and scientific researchers. The special Authority for Unmanned System is more
accurately referred to the 49 U.S.C. 44807 [15@ WR GHWHUPLQH ZKHWKHU ³FHUWDLQ XQ
PDQQHGDLUFUDIWV\VWHPVPD\RSHUDWHVDIHO\LQWKHQDWLRQDODLUVSDFHV\VWHP1$6´
The US FAA required the pilot in command (PIC) to pass the Airman Knowledge Exam
and receive a valid FAA Remote Pilot license [9]. Indian Director General of Civil
Aviation (DGCA) has released guidelines on UAV policies under the title of ³5HTXLUH
PHQWVIRU2SHUDWLRQRI&LYLO5HPRWHO\3LORWHG$LUFUDIW6\VWHP53$6´>16]. It was
newly approved December 1, 2018. The policy proposes 200ft Above Ground Level
(AGL) which is half of AGL in the US FAA Section 333 for UAVs operations.
In China, the Civil Aviation Administration of China (CAAC) released an Interim
Provisions on Light and Small Unmanned Aircraft Operations (UAS Operation Provi-
sions) in December 2015. The maximum empty weight of deployment is 116 kilograms
and a calibrated air speed of no greater than 100 kilometers per hour [17].
In Vietnam, UAVs, also known as fly-cams, are governed in the Decree 36/2008
ND-CP dated 28/3/2008 for hobbyist users only. However, our regulations did not in-
cOXGH WKH GHWDLOHG UXOHV IRU 8$9V¶ VSHHG OLPLW ZHLJKW DQG PD[LPXP DOWLWXGHV RI
8$9V¶IOLJKWRSHUDWLRQV$IWHUWKHHYHQWVRIWZR8$9VIOLJKWFUDVKHVLQ0D\DQG
June 21 2015, Official letter 6321/BQP-TM was released by the Ministry of Defense
dated 21/7/2015 to enhance the restriction of spontaneous flights. PIC permit for com-
mercial fly-cams are required to ensure social security and safety.
Data Privacy of UAV operation is the key concern in the cybersecurity era. (Finn
and Wright 2012) [18] indicated UAVs data privacy have not effectively addressed in
the US, UK, and EU legislation. The US Constitution proposed the Fourth Amendment
to protect citizens from unreasonable searches, especially in privacy expectation loca-
tions. In Europe, the usage of aerial surveillance technologies is covered by the Charter
of Fundamental Rights of the European Union 2000. For the UK, video monitoring by
UAVs devices has been regulated by the Data Protection Act 1998 and the Regulation
of Investigatory Powers Act (RIPA) 2000. The UK and EU Data Protection Acts both
stipulate the rights of citizens to be informed and request of its copies while a surveil-
lance system is in operation. However, the policies are just applied for CCTV database.
The privacy norms are regulated clearly in India DGCA that UAV operator/ remote
pilot shall be liable to ensure not compromised citizen privacy in any manner [16].
1116 H. Q. Pham et al.

In conclusion, the UAVs offer many technical and logistic advantages in trajectory
data collection, but their use is also affected by limiting factors including instrumental
factors itself (low payload, EDWWHU\UDQJH«DQGSULYDF\SURWHFWLRQFRQFHUQV,QFRQ
sideration of Vietnam conditions, a multi-layered regulation with the collaboration be-
tween related legislators would be crucial to utilize the UAVs technology responsibly.

References
1. R. R. Murphy, S. Tadokoro, and A. Kleiner, Disaster Robotics, Springer Handbook of Ro-
botics, pp. 1577±1604, 2016.
2. W.W. Greenwood, J.P. Lynch, D. Zekkos, Applications of UAVs in Civil Infrastructure,
Journal of Infrastructure Systems, Vol 25, No 2, 01-Jan-2019.
3. J. Seo, L. Duque, and J. P. Wacker, Field Application of UAS-Based Bridge Inspection,
Transportation Research Record, vol. 2672, no. 12, pp. 72±81, Jun. 2018.
4. X. Liu, L. Gao, Z. Guang, and Y. Song, A UAV Allocation Method for Traffic Surveillance
in Sparse Road Network, Journal of Highway and Transportation Research and Develop-
ment (English Edition), vol. 7, no. 2, pp. 81±87, Jun. 2013.
5. X. Gu, M. Abdel-$W\4;LDQJ4&DLDQG-<XDQ³8WLOL]ing UAV video data for in-
GHSWKDQDO\VLVRIGULYHUV¶FUDVKULVNDWLQWHUFKDQJH PHUJLQJ DUHDV´ $FFLGHQW $QDO\VLV 
Prevention, vol. 123, pp. 159±169, Feb. 2019.
6. H. Zhou, J. Wang, H. Huang, M. Hou, J. Cheng, Comparative Study of the Operating Char-
acteristics between Cars and Trucks on Freeway Loops: An Empirical Study Based on Drone
Videos | TRB Annual Meeting, Trb.org, 2019
7. V. Sharma, H.-C. Chen, and R. Kumar, "Driver behavior detection and vehicle rating using
multi-UAV coordinated vehicular networks," Journal of Computer and System Sciences,
vol. 86, pp. 3±32, Jun. 2017.
8. M. A. Khan, W. Ectors7%HOOHPDQV'-DQVVHQVDQG*:HWV³8$9-Based Traffic Anal-
\VLV $ 8QLYHUVDO *XLGLQJ )UDPHZRUN %DVHG RQ /LWHUDWXUH 6XUYH\´ 7UDQVSRUWDWLRQ 5H
search Procedia, vol. 22, pp. 541±550, 2017.
9. (1%DUPSRXQDNLV(,9ODKRJLDQQLDQG-&*ROLDV³8QPDQQHG$HULDO$LUFUDIW6\V
tems for transportation engineering: Current practice and future FKDOOHQJHV´,QWHUQDWLRQDO
Journal of Transportation Science and Technology, vol. 5, no. 3, pp. 111±122, Oct. 2016.
10. E. Dourado, R. Hagemann, and A. Thierer, Operation And Certification Of Small Unmanned
Aircraft Systems. No. 15-01. FAA-2015-0150 Notice, 2015.
11. 0-DOD\HU02¶&RQQHOO+=KRX36]DU\6'DV$SSOLFDWLRQ2I8$97R,QVSHFW$QG
Inventory Interchange Assets To Mitigate Wrong-Way Entries, Trb.org, 01-Jan-2019.
12. C%URRNVHWDO³(YDOXDWLQJWKH8VHRI8QPDQQHG$HULDO9HKLFOHVIRU7UDQVSRUWDWLRQ3XU
SRVHV´7UERUJ-Jan-2015.
13. G. Salvo, L. Caruso, and A. Scordo, "Gap acceptance analysis in an urban intersection
through a video acquired by a UAV," Unipa.it, Jan. 2015.
14. 6SHFLDO$XWKRULW\IRU&HUWDLQ8QPDQQHG$LUFUDIW6\VWHPV6HFWLRQ´ Faa.gov, 2012.
15. ³3HWLWLRQLQJIRU([HPSWLRQXQGHU6HFWLRQ´ Faa.gov, 2018.
16. *2,7(&+1,&$/&(175(³5HTXLUHPHQWVIRU2SHUDWLRQRI&LYLO5HPRWHO\3LORWHG
$LUFUDIW6\VWHP53$6´-Jan-2018.
17. ³5HJXODWLRQRI'URQHV3HRSOH¶V5HSXEOLFRI&KLQD´ Loc.gov, 2015.
18. R. L. Finn and D. Wright, "UAS: Surveillance, ethics, and privacy in civil applications,"
Computer Law & Security Review, vol. 28, no. 2, pp. 184±194, Apr. 2012.
 New cooperative approach between universities and
enterprises, towards a reality urban planning. Lesson
learnt from Nhon Ly, Quy Nhon city

Nguyen, Thai Huyen1 and Dang, Thuc Trang2

1
Institute of International Training and Cooperation, Hanoi Architectural University, Vietnam
2
Company Scene Arch Plus, Vietnam
thaihuyennguyen@gmail.com
trang.dangthuc@sceneplus.vn

Abstract. In the process of development and international integration, the ques-

tion of training associated with the practice has always been a challenge for all
universities. Among the training, which requires a practical environment, the
architectural formation in Vietnam is also in the process of transforming itself
to adapt to the reality needs and international integration. From 2017, Hanoi
Architectural University has been a pioneer in developing new international
training model, in co-operation with schools of architecture in France and en-
terprise, which calls work-shop. The organization of the workshop is based on
the cooperation between Hanoi Architectural University, the city authorities
where the workshop takes place and an architectural cabinet. The workshop
proposes to understand the complex relationship between space and time, the
process and mutations of space at different scales and human influence on the
transformation of the community. In this paper, by experiences from our work-
VKRSDW1KѫQ/ê4X\1KѫQSURYLQFHZHZLVKWRLQWURGXFHWKHDGYDQWDJHVRI
the method in this new specialized training as well as pointing out the difficul-
ties in the training process towards a reality urban planning.

Keywords: Landscape architect training, Vietnam, urban planning, coopera-

tion, workshop.

1 Context of the birth of Landscape Architecture and its

training in Vietnam

Landscape Architecture is a major with young developing history comparing with

other tradition design majors such as architecture or planning. Before 1800, Land-
scape Architecture had not been mentioned as a separate major but combined with
planning. The terminology « Landscape Architect » was first used as a professional
career title by Frederick Law Olmsted in 1863. The development history of Land-
scape Architecture was later affirmed with the birth of American Society of Land-
scape Architects (ASLA) in 1899 and International Federation of Landscape Archi-
tects (IFLA) in 1948.

© Springer Nature Singapore Pte Ltd. 2020 1117

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_179
1118 T. H. Nguyen and T. T. Dang

In Vietnam, Landscape Architecture has been a new major so far. However, not ex-
cluded from the common law of the world, the strong urban development process
during the past years has led to the demand of manpower in Landscape design. From
that reality, many domestic training organizations started to open training programme
of Landscape. Generally, these programmes can be divided into two groups according
to training goals. The first is the training programme of Landscape Architecture Engi-
neer at Universities of Agriculture ± Forestry. The second is the training programme
of Landscape Architect at Universities of Architecture. At Hanoi Architectural Uni-
versity, the programme of Landscape Architect in French in association with three
schools from France (Normandie National School of Architecture, Bordeaux National
School of Architecture and Landscape, Toulouse National School of Architecture) has
been proceeded since 2010

2 Training on Landscape Architect is closed to reality through

field design WORKSHOP

The model of training Architect in workshops is a familiar model in the environments

of Architect training in Vietnam. In front of the demand that training to be closed to
reality, after a lot of changes, the model of Architect training in workshops at Hanoi
Architectural University has been more and more completed with a lot of new types
of exercises and working method and closer to reality. Besides, the training coopera-
tion programmes of Architect training between Hanoi Architectural University and
Schools of Architecture and Landscape in France have also brought a lot of contribu-
tion during the training innovation of the university.
Field design workshop exercise, which is also called Workshop, is an exercise of
project, which is executed in 1 week at a locality outside Hanoi. The participants are
4th or 5th year students of Hanoi Architectural University and students from partner
universities from France, under the instruction of Vietnamese and French lecturers.
The exercises are organized at the beginning of the 1st term of their 4th and 5th year
in the training programme.
Since 2017, Hanoi Architectural University has developed a cooperation method
between enterprises and localities in their training to combine with reality. Workshop
exercise are organized under the sponsorship of localities and enterprises. The work-
VKRS¶V UHVHDUFK UHVXOW DQG UHFRPPHQGDWLRQV DUH XVHG LQ UHDOLVWLF H[HFXWLRQ RI SODn-
ning, architecture and landscape design of enterprises and localities.

3 Workshop in Nhon Ly commune, Quy Nhon city ± Model of

Universities ± Localities ± Enterprises cooperation

Nhon Ly is a coastal commune of Quy Nhon city, about 22 km away from Quy Nhon
FHQWHU:RUNVKRSZLWKWKHWRSLF³5HVHDUFKRQKRXVHVDQGXUEDQIRUPLQJODndscape,
public spaces of Nhon Ly fishing village ± 4X\1KRQFLW\´ZDVFRQGXFWHGIURPQG
to 28th August, 2018 in Nhon Ly commune, Quy Nhon city. The lead organization
New cooperative approach between universities … 1119

was Institute of International Training and Cooperation, Hanoi Architectural Univer-

sity, with cooperation and supports from Scene Plus Architects Company and Quy
1KRQ3HRSOH¶V&RPPLWWHH
Workshop contributes a part to the sustainable community based tourism (CBT),
exploiting the strength and outstanding features of Nhon Ly commune with the fa-
mous place Eo Gio. This workshop has set out the research directions to exploit the
strengths and characteristics of the location, community, terrain, climate ... The ap-
proaches and problem solving at the workshop with the participation of schools -
businesses - society have been new methods to ensure the implementation of flexible
and sustainable solutions with long-term goals.
Workshop is organized into two stages.
The first stage is the "Survey and Concept" step based on a week of site surveys.
Collecting data and interpreting existing maps / research records for Nhon Ly, meet-
ing with local authorities and people to grasp the guidelines, planning orientations and
challenges posed to localities directions in the development process.At the end of the
survey period, the groups had a preliminary report to the representatives of local au-
thorities, Nhon Ly people, enterprise and leaders of the Institute of International Co-
operation and Training, experts and lecturers of the Hanoi Architectural University, to
acknowledge comments.
The second phase was continued by students during the next week at Hanoi Archi-
tectural University. Based on the comments, the groups edit, consolidate and finalize
the report and propose preliminary ideas. At the end of this second week, the groups
made official reports to the council consisting of French, Vietnamese experts, profes-
sors, local representatives and enterprise. The report and proposal of preliminary ideas
are then finalized and sent to enterprise and representatives of local authorities as
research materials and considered for application.

4 The result and scientific applications of the workshop:

identifying the urban form, landscape and public space of
fishing village in Nhon Ly commune, proposing planning
concept 1/2000

As PRVWRIWKHILVKHUPHQ¶VYLOODJHLQ9LHWQDP1KRQ/\LVORFDWHGQHDUE\DSHQLQVXOD
offering two moorages, one on the north side protected from the south wind and one
on the south side protected from the north wind. Each one stands along a beach,
which consWLWXWHVD UHDO KDUERU,W¶VWKH SODFHIURP ZKLFK WKH ILVKHUPHQ OHDYHWR JR
fishing and the place where they come back with their catch. It is the center of the life
RIVXFKILVKHUPHQ¶VYLOODJH7KURXJKWKHFHQWXULHVDQGPDLQO\LQEHWZHHQVDQG
2005 as far as architecture is concerned, the people of Nhon Ly have elaborated a
functional, ecological, beautiful and symbolic village of fishermen.
The urban form of Nhon Ly fishing village is shaped based on the habits and pro-
fessional characteristics of fishermen so that they can live conveniently and safely in
ecological and very humane environments. The main streets lead to the beach (where
WKHUHLVDILVKLQJSRUWLQWKHFHQWHURIWKHILVKLQJYLOODJH7KH\DUHWKH³UDGLDO³VWUHHWV
1120 T. H. Nguyen and T. T. Dang

they follow the slope and lead naturally all the rain water to the sea. They are no
straight in order to break the wind coming from the sea; they are narrow to make
VKDGRZV 7KH VWUHHWV FRQQHFWLQJ WKH UDGLDO RQHV DUH WKH ³WUDQVYHUVDO ³VWUHHWV WKH\
IROORZWKHOHYHOFXUYHV,W¶VRQ which that are located the shops, they lead to the plac-
es of community life such pagodas and temples.
The houses in Nhon Ly were built almost in the 1960s and 1970s. Through the ini-
tial survey, many of small houses were beautiful, creative and harmonious. The hous-
es have one or two levels, in order not to break the view from the other houses on the
VHD)URPPRVWRIWKHKRXVHVLW¶VSRVVLEOHWRVHHWKHVHDDQGZKDWLVJRLQJRQWKHUH
DQGWKDW¶VLPSRUWDQWIRUILVKHUPHQ
In particular, the architectural form of many houses in Nhon Ly fishing village
contains heritage values. The surveyed houses in Nhon Ly were built during the peri-
od of 1960 - 2005. Through the survey, the research group pointed out the architec-
tural value of the very beautiful little houses, very well designed, very well colored,
creative, harmonious proportions, with elegant wooden doors and windows. At that
time, the fishermen invest not in the size, not in the luxurious but in the simple beau-
ty.
In recent years, the Community based tourism model has tended to form in some
coastal areas of Quy Nhon city. However, compared to the potential of nature and
culture, this development is not commensurate, it is also spontaneous and unprofes-
sional. Local communities involved in the tourism organization process are passive.
There are no policies for people to benefit from their community tourism activities.
Nhon Ly is a typical fishing village that is facing rapid urban development. The ap-
pearance of tourist elements is an inevitable consequence when the Resort is built
here. That brings strong impacts on Nhon Ly's transformation. In fact, the locality
receives nearly 1000 visitors a day. This is a great pressure for a locality on the rise of
tourism.
The question is how to balance economic development and tourism with preserv-
ing the local architectural form, and accepting urban development naturally, but both
elements must be satisfied harmony, sustainability.
In the site survey, 5 groups of students have identified the characteristics of Nhon
Ly's identity as the proposed foundation, that is:

x Nhon Ly fishing village cemetery is located on the sand hill, has a favorable
geo-graphical position, located in the center, is the first access point of the area,
with high altitude, panoramic views of the area.
x The architectural space of the houses making fish sauce
x Homestay area
x Spontaneous public space in the village includes vacant land with views of the
sea, steep seaward paths.
x The seafront facade of Nhon Ly fishing village
New cooperative approach between universities … 1121

Fig. 1. &RQFHSW³6WDUWLQJIURPWKHHQG´ Fig. 2. &RQFHSW³)ROORZLQJWKHIODYRURIWKH

VHD´

Fig. 3. &RQFHSW³2QHGD\WREH1KRQ/\ Fig. 4. &RQFHSW³-RXUQH\WRFRQQHFWPHPo-

SHRSOH´ ULHV´

Fig. 5. &RQFHSW³$SSHDUDQFHRI1KRQ/\ILVKLQJYLOODJH´

Since then, preliminary proposal ideas have been established (see Fig.1 to Fig.5 -
Source: Institute of International Training and Cooperation, Hanoi Architectural Uni-
versity.
1122 T. H. Nguyen and T. T. Dang

x Starting from the end: the idea of reconstructing Nhon Ly's cemetery space in-
to a cemetery park, a destination for tourists to capture local history and culture.
x Following the flavor of the sea: the idea of improving the Landscape Architec-
ture space, forming tourist routes to discover, experience, visit traditional fish
sauce making establishments here.
x One day to be Nhon ly people: proposed concept of homestay in the fishing
village of Nhon Ly towards the experience of life in fishing villages for tourists.
x Journey to connect memories: the idea of re-creating public spaces that con-
nect the memories of Nhon Ly, which are public spaces with regional souls.
Appearance of Nhon Ly fishing village: The research team proposed to re-
build Nhon Ly fishing village based on the available structures of fishing villag-
es (embankments, slopes, sandbanks...), identifying architectural elements that
need to be preserved or need development control to ensure the typical land-
scape of fishing villages, creating beauty, bringing a distinct identity to Nhon
Ly.

5 Conclusion

With the interest, participation, support and coordination of the locality, enterprise
DQG SHRSOH ZRUNVKRS ³5HVHDUFK RQ KRXVHV DQG XUEDQ IRUPLQJ ODQGVFDSH SXEOLF
spaces of Nhon Ly fishing village ± 4X\1KRQFLW\´RUJDQL]HGE\+DQRL$UFKLWHFWXr-
al University and Scene Plus Architects has brought about high practicality. The ap-
proach and problem solving in the implementation process is new to ensure the im-
plementation of flexible and sustainable solutions with long-term goals by the partici-
pation of the community from the beginning of the project.
The goals and requirements set for training have been ensured: practicality, feasi-
bility, teamwork skills as well as the ability to work independently, the application of
skills learned in the exercise in an efficient and fluent way, practical interoperability
with problems and solutions offered by students.
Finally, the implementation of design workshop on site, in conjunction with local
and enterprise, is capable of creating employment opportunities for students after
graduation, engaging training with practice and connecting enterprises.

References
1. Landscape Architecture French Program (2009), HAU, Hanoi, Vietnam.
2. Nguyen TH.: Teaching and learning of landscape architecture project. Scientific Confer-
ence of 45th years Hanoi Architectural University, HAU, Hanoi, Vietnam (2014).
3. Nguyen TH.: Application of French experiences in teaching landscape architecture.
Scientific Research of Hanoi Architectural University, Hanoi, Vietnam (2016).
4. 'DQJ771KѫQ/ê)LVKLQJ9LOODJHDQHZ,GHDVIRUDWRXULVWVHWWOHPHQWDW the fishing vil-
ODJH:RUNVKRS³5HVHDUFKRQKRXVHVDQGXUEDQIRUPLQJODQGVFDSHSXEOLFVSDFHVRI1KRQ
Ly fishing village ± 4X\1KRQ&LW\´4X\1KRQ9LHWQDP
 Smart Monitoring For Urban Biodiversity Preservation

Thi Hai Yen Pham1, Isam Shahrour1,*, Ammar Aljer1,*, Alain Lepretre1,*, Celine
Pernin1,*, Sana Ounaies1,*,
1
Laboratoire de Génie Civil et géo-Environnement, Université de Lille, 5900 Lille,
France

Abstract. The rapid urbanization causes increased pressure on biodiversity,

with dramatic consequences such as reduction of green areas, space fragmenta-
tion and species loss or transformation. Considering the importance of biodiver-
sity for the quality of life in cities as well as for the biological and environmen-
tal equilibrium, cities should take great care for the protection and enhancement
of the biodiversity. All the stakeholders of the city should be involved in the re-
alization of this objective.
The biodiversity conservation requires first of all monitoring in order to scan
the current situation, to track the biodiversity evolution and to take the right
measurements to stop the biodiversity degradation and even more to ensure its
development. However, the biodiversity monitoring is very complex, because
of the huge variety of parameters of the biodiversity as well as the long-term
timescale. The recent developments of Smart Technology could be used for bi-
odiversity monitoring in cities.

Keywords: Smart monitoring, urban biodiversity, sensors networking, smart

city.

1 Introduction

Biodiversity designates the variability of ecological organisms and their environment.

It is the foundation of ecosystem services as material basis for humans. Urban biodi-
versity provides a series of benefits to cities ranging from the more directly perceived,
such as water supplies and recreation facilities (parks) to less tangible effects of large
biodiversity areas, such as hosting species, which help cure diseases or contribute to
long-term climate stability [10]. Intensive urbanization leads to biodiversity loss with
negative impact on human well-being. Consequently, the conservation of urban bio-
diversity becomes a major issue for cities. This conservation requires adequate moni-
toring to understand the current situation, to track biodiversity evolution and to take
the right measurements to stop biodiversity degradation and even more to ensure its
development. However, biodiversity monitoring is very complex; because of the large
variety of biodiversity parameters as well as the long-term time scale. This paper
presents a methodology for this monitoring using smart technology.

© Springer Nature Singapore Pte Ltd. 2020 1123

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_180
1124 T. H. Y. Pham et al.

2 Monitoring Indicators

Urban biodiversity concerns living organisms, including genetic variation, and

habitats. Species range from rural fringe to urban core [2]. Since, it is not possible to
monitor all spices everywhere, it is very important to set up the monitoring targets,
which are necessary for decision-making [7]. Table 1 summarizes the indictors, which
should be followed for urban biodiversity preservation. They include 3 categories:
biodiversity, causal factors and management system.

Table 1. Urban Biodiversity indicators

Biodiversity Coverage Proportion of green areas

Connectivity/Fragmentation of green areas
Biology Abundance and distribution of selected species
Change in number of native species
Proportion of invasive alien species
Causal factors Physical factors Temperature, Humidity of air
Temperature, Humidity of soil
Human activities Changing in land used
Soil pollution
Water pollution
Air pollution
Noise pollution
Light pollution
Urban hazards Flood
Fires
Industrial
Management Participation of stakeholder Involvement of citizens in protecting urban
system biodiversity
Finance Budget for urban biodiversity projects

Biodiversity sectors.
Biodiversity indicators include 2 sub-groups:
Coverage indicators: they concern the proportion of green areas as well as connec-
tivity and fragmentation of green areas.
Biology indicators: they concern abundance and distribution of selected species,
change in number of native species and the proportion of invasive alien species.

Causal factors.
Causal factors indicators include 3 sub-groups:
Physical factors, which concern temperature and humidity of the soil and air.
Human activities, which concern pollution of soil, water and air as well as noise
and light pollutions. Urban hazards, which concern flood, fire and industrial activity.
Smart Monitoring For Urban Biodiversity Preservation 1125

Management factors.
Management factors indicators are organized in 2 sub-groups:
Stakeholders participation, which include involvement of citizens in protecting ur-
ban biodiversity Finance, which concerns the budget dedicated for urban biodiversity.

3 Urban biodiversity smart monitoring.

A major objective of biodiversity measurement is to feed policies with pertinent in-

formation to reduce the biodiversity loss as a component of sustainable economic
development strategy [4]. Today, biodiversity monitoring does not only rely on tradi-
tional manual surveys but uses new technologies, such as genetic bar coding, camera
traps and drones.
According to Talari et al. (2017), the smart monitoring consists of four layers: per-
ception layer, network layer, database layer and application layer (Figure 1).
The perception layer includes a group of Internet-enabled devices that are able to
perceive, detect objects, gather information, and exchange information with other
devices through t communication networks. Cameras, sensors, Global Positioning
Systems (GPS) are the devices of perception layer. Forwarding data from the percep-
WLRQ OD\HU WR WKH DSSOLFDWLRQ OD\HU XQGHU WKH FRQVWUDLQWV RI GHYLFHV¶ FDSDELOLWLHV QHW
ZRUNOLPLWDWLRQDQGWKHDSSOLFDWLRQV¶FRQVWUDLQWs is the task of the network layer. In-
ternet technologies such as Wi-Fi, 2G, 3G, 4G, 5G carry the information over long
distances based on the application. With the database layer, database management
system (DBMS), structured query language (SQL) are used to store and analysis.
Since applications aim to create smart monitoring, the last layer which is the applica-
tion layer is where the information is received and processed.

Perception layer Gateway, Wifi, 2G, 3G,4G, 5G

Network layer Gateway, Wifi, 2G, 3G,4G, 5G

Database layer Database management system (DBMS), Structured query language

(SQL)
Application layer Smart monitoring

Fig. 1. Smart monitoring layer

Figure 2 shows the monitoring system used to follow the biodiversity indicators pre-
sented in table 1. The following sections describe this system.
1126 T. H. Y. Pham et al.

Application in smart urban biodiversity monitoring

Biodiversity Caused factors Management

Cameras traps, Cameras monitoring

Wireless sensors net-
remote sensing work
Tracking Public place monitoring
Physical factors Meas-
urements
Identification
Pollution monitoring (Air, People and objects
water, noise, light) tracking
Data gathering Security systems (fire, earth- Social network sup-
quake,...) porting

Fig. 2. Application in smart urban biodiversity monitoring

Camera traps.
The biology indictors particularly focus on fauna change. It is monitored by cam-
era traps technology which allows observation of species that are not easily seen.
Camera traps are also useful for detecting events, behaviours or associations not easi-
ly observed during sightings. Digital camera traps have replaced film-loaded ones,
greatly expanding our abilities to detect rare or secretive species. Satellite-borne cam-
eras can detect and monitor some animals in open habitats [3]. In some cases, the use
of camera traps has improved monitoring results dramatically, with a 60%
known/identifiable population becoming a 95% known population in a matter of
months1.
Remote sensing

Remote sensing is used to gather information from distant objects by measuring the
radiation that they reflect or emit. They use 2 types of sensors: active and passive [5].
Remote sensing is used to evaluate the coverage indictors. It helps monitoring in large
scale. Since, there are various components of biodiversity; the remote sensing varies
according to components. Suitability of this approach depends upon the kind of envi-
ronment in which biodiversity monitored [5].
Conservation requires methods to identify species and location as well as to track
individual plants and animals. These methods have become better, faster, less intru-
sive and cheaper. Remote sensing technology allows detailed and frequent assess-
ments of VSHFLHV¶Kabitats as well as the impact of human actions on species.
Smart Monitoring For Urban Biodiversity Preservation 1127

Wireless sensor networks (WSNs).

Wireless sensors network and IoT are integrated to measure the impact of causal
factors on urban biodiversity. A WSN can be defined as a network of tiny devices,
called sensor nodes, which are spatially distributed and work cooperatively to com-
municate information gathered from the monitored field through wireless links. Data
gathered by the different nodes are sent to a sink which either uses the data locally or
is connected to other networks, for example, the Internet (through a gateway) [8].
WSNs make diverse proper data available and might be applied in lots of uses like
healthcare as well as government and environmental services [1]

Fig. 3. Wireless sensor network (WSN) [8]

This technology is used to measure physical environment elements such as temper-

ature and humidity of air and soil, monitoring the pollution of air, water, noise and
light as well help predict the risk of hazards from nature and human.
WSN technology functions best when integrating camouflage, low power-
consuming devices, sophisticated signal processing software and hardware, and suita-
ble packaging that can withstand hostile environmental conditions. WSN is a fast
emerging field and ecologists and conservation practitioners alike can benefit signifi-
cantly from new understanding of their target species or environments [6]

Data from citizens

Citizens are affected by the biodiversity. They could be composed of different

groups with health, welfare, intellectual, recreational, spiritual, and financial interests
that will be affected by changes in biodiversity. We define interested parties as a sub-
set of stakeholders who specifically support the scientific measurement of biodiversi-
ty. They may be individuals but generally, they are organized in associations or insti-
tutions [4]
A lot of data can be collected and transmitted by citizens. Smartphone apps could
be used in biodiversity monitoring. Residents can send sightings of invasive non-
1128 T. H. Y. Pham et al.

native species and tree disease sightings. This contribution has some benefits such as
real-time data, large cover and data geo-localization [9]

4 Conclusion

Preservation of urban biodiversity constitutes a key issue for urban sustainable de-
velopment. Intensive urbanization causes biodiversity deterioration with negative
impact on both ecosystem equilibrium and human well-being. This preservation re-
quires adequate monitoring to understand biodiversity evolution as well as the causal
factors and their impact on biodiversity. Smart monitoring well help to take the right
measurements to stop biodiversity degradation and even more to ensure its develop-
ment. The paper presented the monitoring indicators related to biodiversity, causal
factors and management system as well as an advanced monitoring system based on
the use of camera traps, remote sensing, sensors networks and citizens observations.
References
1. $ODPUL $WLI HW DO  ³$ 6XUYH\ RQ 6HQVRU-Cloud: Architecture , Applications , and
$SSURDFKHV´ 2013.
2. &%'³&LWLHVDQG%LRGLYHUVLW\2XWORRN$*OREDO$VVHVVPHQWRIWKH/LQNVEHWZHHQ
8UEDQL]DWLRQ %LRGLYHUVLW\ DQG (FRV\VWHP 6HUYLFHV´ Executive Summary 64. Retrieved
(http://www.cbd.int/authorities/doc/cbo-1/cbd-cbo1- summary-en-f-web.pdf).
3. 'HKJDQ$OH[HWDO³(PHUJLQJ7HFKQRORJLHVWR&RQVHUYH%LRGLYHUVLW\´ Trends in
Ecology & Evolution (October). Retrieved (http://dx.doi.org/10.1016/j.tree.2015.08.008).
4. Green, Rhys E. et al. 2005. ³A Framework for Improved Monitoring of Biodiversity: Re-
sponses to the World SummLWRQ6XVWDLQDEOH'HYHORSPHQW´ 19(1):56±65.
5. .KDUH6LGGKDUWKDDQG6.*KRVK³6PDUW6DWHOOLWH5HPRWH6HQVLQJ7HFKQRORJLHV
)RU%LRGLYHUVLW\0RQLWRULQJ$QG,WV&RQVHUYDWLRQ´ 431±40.
6. Marvin, David C. et al. 2016. ³,QWHJUDWLQJ7HFKQRORJLHVIRU6FDODEOH(FRORJ\DQG&RQVHU
YDWLRQ´ Global Ecology and Conservation 7:262±75. Retrieved
(http://dx.doi.org/10.1016/j.gecco.2016.07.002).
7. McDonald-0DGGHQ (YH HW DO  ³0RQLWRULQJ 'RHV 1RW $OZD\V &RXQW´ Trends in
Ecology and Evolution 25(10):547± 50.
8. Network, Wireless Sensor. 2014. ³Wireless Sensor Network: Recent Developments and Po-
tential Synergies´
9. 3ODQ %LRVHFXULW\ $FWLRQ DQG 1DWLRQDO3ROOLQDWRU6WUDWHJ\ ³(QYLURQPHQWDO &LWL]HQ
6FLHQFH´ (476):1±5.
10. Puppim de Oliveira, J. A. et al. 2011. ³&LWLHVDQG%LRGLYHUVLW\3HUVSHFWLYHVDQG*RYHUQ
ance Challenges for Implementing the Convention on Biological Diversity (CBD) at the
&LW\/HYHO´Biological Conservation 144(5):1302± 13.
11. 7DODUL6DEHUHWDO³$5HYLHZRI6PDUW&LWLHV%DVHGRQWKH,QWHUQHWRI7KLQJV&RQ
FHSW´Energies 10(4):1±23.
 Part VI
Data mining and Machine Learning
 A Daily Work Report Based Approach for Schedule
Risk Analysis

Chau Le1[0000-0002-2582-2671] and H. David Jeong2[0000-0003-4074-1869]

1
Ph.D. Student, Texas A&M University, College Station TX 77843, USA
2
Professor, Texas A&M University, College Station TX 77843, USA

Abstract. The Critical Path Method (CPM) has been the dominant scheduling
method for the past several decades. Most State Departments of Transportation
(DOTs) in the U.S. currently apply CPM to estimate project duration for major
and complex projects. The method, however, has limitations, and one of them is
its poor ability to analyze schedule risks. While previous studies have identified
various factors that cause uncertainties in schedules, the estimated duration from
CPM is deterministic. To quantify schedule risks, the Program Evaluation Re-
view Technique which uses three-point estimates for activity duration and Monte
Carlo Simulation (MCS) based methods that assign uncertainties to work activi-
ties through the probability distributions of activity durations have been limitedly
used in the industry. After running the simulation thousands of times, the proba-
bility distribution of the total project duration can be developed to assist risk anal-
ysis, such as determining a schedule contingency for the project. In current prac-
tices, the use of the MCS method is still limited due to various reasons. One of
the most challenging issues is the difficulties in estimating the probability distri-
butions of activity durations objectively. This study aims to leverage historical
digital daily work report (DWR) data available in the DOTs’ databases to deter-
mine the probability distributions of the production rates of work activities then
estimate the distributions of activity durations when quantities of the activities
are provided. Since the DWRs record the daily accomplished quantity of each
work item in the construction phase, the actual production rates of the work items
can be calculated to obtain a more accurate and realistic duration estimate for a
future project. DWR data collected from a DOT were used to conduct a case
study that demonstrates the value of this new approach for schedule risk analysis.

Keywords: Critical Path Method, Monte Carlo Simulation, Daily Work Report
Data, Project Scheduling, and Schedule Risk Analysis.

1 Introduction

The consistent change in the duration and sequence of work activities has a significant
impact on the accuracy of the planned schedule of a construction project, hence the
need for the determination of the uncertainty associated with the project schedule (Os-
trowski 2006). As a dominant scheduling method for more than half a century, the Crit-
ical Path Method (CPM) has been used widely in the construction industry (Kelly et al.

© Springer Nature Singapore Pte Ltd. 2020 1131

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_181
1132 C. Le and H. D. Jeong

2014). However, CPM, a deterministic scheduling method, often fails to deal with
schedule risks (Ramdass and Lim 2011). As a complement to CPM, the Program Eval-
uation and Review Technique (PERT) and Monte Carlo Simulation (MCS) allow for
the uncertainty in the duration of work activities. The two methods, however, have lim-
itations. One major drawback is the subjectivity in the three-point estimation of activity
duration for PERT and the determination of the probability distributions of work items
for MCS.
This study investigated the use of digital daily work report (DWR) data for historical
data driven schedule risk analysis. Since the DWRs record daily accomplished quantity
of each work item in the construction phase, the data provide a reliable and objective
source for the analysis. The combined usage of regression analysis and MCS allows for
the development of the probability distributions of the work activities in estimating the
project duration with the consideration of the effects of influential factors on production
rates. A case study was conducted to illustrate the proposed approach with historical
DWR data obtained from a DOT for the case study.

2 Background

2.1 Deterministic scheduling methods

The top two scheduling methods that are mostly used by Departments of Transportation
(DOTs) are the bar chart method (for small projects) and CPM (for primary and com-
plex projects). A bar chart is used to represent a project schedule. The chart lists the
activities that need to be performed. Each activity is illustrated by a bar, and the length
of the bar is proportional to the planned duration of the activity (FHWA 2002). The
advantages of bar charts are simple to develop and easy to understand. However, bar
charts do not show the relationships between the activities, which impose difficulties
when construction changes occur (FHWA 2002). Also, using bar charts does not help
identify the activities that control the overall project schedule, hence the limited use of
bar charts for large and complex construction projects (FHWA 2002).
CPM also allows for a graphical representation of a project schedule like bar charts.
However, CPM overcomes the disadvantage of the bar chart method by taking account
of the relationships between the activities. CPM not only can provide a good estimate
of project duration but also is a useful tool to verify and monitor the project schedule
(FHWA 2002). The disadvantages of the method are the requirement of experience and
knowledge from the scheduler to use effectively and the need for regular updates to
reflect the actual progress of the project (FHWA 2002). However, with the development
of a large number of sophisticated CPM computer applications, CPM has been used
extensively for various purposes such as resource and cost allocation, progress report-
ing, and schedule updating (Harmelink and Rowings 1998; Mattila and Park 2003).

2.2 Probabilistic scheduling methods

In terms of risk analysis, the disadvantage of CPM is its poor ability to analyze schedule
risks (Ostrowski 2006). While previous studies have identified various factors that
A Daily Work Report Based Approach... 1133

cause uncertainties in schedules, the estimated duration from CPM is deterministic.

PERT acknowledges that the CPM schedule is just approximations and is associated
with some level of uncertainty. PERT allows for three estimates of duration for each
activity: optimistic (a), pessimistic (b), and most probable (m). With the assumption
that each activity has a beta distribution, PERT provides the formulas to calculate the
expected duration of an activity and the variance of the activity duration. Based on the
central limit theorem, the total duration of the project is assumed to follow a normal
distribution when the number of independent activities gets larger than 30. The ex-
pected project duration and its variance are subsequently formulated by summing the
corresponding estimates of the activities on the critical path of the project. However,
the use of PERT has been criticized by numerous studies. Some of the problems are the
subjective estimates of a, b, and m by estimators, the assumption of the beta distribution
of all activities, the ignorance of near-critical paths in calculating the project duration,
and the assumption of a normal distribution for the project duration (Cottrell 1999;
Salas-Morera et al. 2018). From a practical perspective, Hajdu and Bokor (2016) sug-
gest that schedulers should focus more on determining the accurate values of a, b, and
m than on exploring different distribution models of activity duration due to the insig-
nificant impact of distribution types on the project duration.
MCS overcomes the disadvantages of PERT by allowing various distribution models
of activity duration, considering all possible paths of the schedule, and not making the
normality assumption of the project duration. In current practices, the use of the MCS
method is still limited due to various reasons. One of the most challenging issues is the
difficulties in estimating probability distributions of activity durations objectively.

3 A daily work report based approach for schedule risk

analysis

DWRs contain digital records of daily accomplished quantity of each work item during
the construction phase. The actual production rate of a work item can be calculated
using the historical DWR data to obtain a more accurate and realistic duration estimate
for a future project. This study aims to leverage historical DWR data to determine the
probability distributions of the production rates of work activities objectively. Based
on these distributions, the quantities of the work items, and the logic among the activi-
ties, MCS can be applied to identify the probability distribution of the project duration
for schedule risk analysis.
The production rate of a work item in a past project is the total installed quantity of
the work item divided by the total number of working days on which the work was
performed. These production rates are a reliable source to identify the actual production
rate distribution of a work item. The most appropriate distribution model can be iden-
tified by fitting various distribution models to the past production rates and comparing
the fitness among the models. However, this approach encounters some problems.
Firstly, the variance of the chosen distribution can be very high due to the consideration
of all possible values of past production rates in DWR data. The wide range of produc-
1134 C. Le and H. D. Jeong

tion rate distributions can result in a large variability of the project duration, hence dif-
ficulties in making scheduling decisions. Secondly, this approach ignores the impact of
possible influential factors (e.g., quantities, location, and project type) on production
rates. For a given project condition, the realistic range of production rates can be esti-
mated in a much narrower range.
To overcome the drawbacks above, regression models can firstly be developed from
the DWR data to predict production rates using a set of independent variables repre-
senting major influential factors. The regression model of an activity provides not only
the expected production rate of the activity but also the probability distribution of the
production rate, which is typically used to determine the confidence interval of the out-
put. This feature, however, has rarely been leveraged for schedule risk analysis in the
literature. The distribution and its standard deviation are not fixed but varied depending
on the input values of the regression model. The achieved distributions of production
rates are then used as input in the MCS model to estimate the project duration.

4 Case study

The section presents a case study that applies the proposed approach for schedule risk
analysis. Table 1 shows the list of work items and schedule-related information of a
hypothetical example of a rural reconstruction project. In this case study, the project
duration was calculated in working days. To obtain the project duration in calendar
days, additional information regarding non-working days (e.g., weekend, holidays, and
adverse weather days) are needed. The historical DWR data of a DOT were used to
develop regression models and production rate distributions for the case study.

Table 1. Project schedule information

No Work Item Unit Quantity Logic Duration
1 Mobilization ls 1 5 days
2 Topsoil-Salvaging and Placing m3 53,642 1 LR+MCS
3 Drainage Pipe m 1,464 2SS+3 LR+MCS
4 Excavation-Unclassified m3 425,673 2,3SS+10 LR+MCS
5 Special borrow m3 142,120 4SS+4,3FF +3 LR+MCS
6 Crushed Aggregate Course m3 59,691 5FF + 3 LR+MCS
7 Plant Mix Surfacing t 25,353 6 LR+MCS
8 Miscellaneous works ls 1 4,7 15 days
Note: SS–Start to Start, FF–Finish to Finish

From the historical DWR data, linear regression (LR) models were developed to predict
production rates of work items based on the quantities of the items. Quantities were
chosen as a predictor since they seem correlated with production rates. Fig. 1 shows the
scatter plots and the correlation coefficients between quantities and production rates of
two activities. The output distribution of production rates had a Student’s t-distribution
with the degree of freedom of (n - 2), where n was the number of the projects used to
develop the regression model. The mean of the output distribution was the expected
production rate, and the standard deviation varied depending on the input quantity.
A Daily Work Report Based Approach... 1135

Fig. 1. Correlation between quantities and production rates

Given the quantities and the logic in Table 1, the project duration was a function of
production rates of the work items. The shape, mean, and standard deviation of each
production rate distribution were obtained from the corresponding regression model.
By running a 10,000-iteration simulation, the production rate distributions of the work
items and the probability distribution of the project duration were developed (see Fig.
2 and Fig. 3). According to Fig. 3, while the expected project duration is 75 working
days, the 99% confidence interval of the expected duration is from 70 to 81 days.

Fig. 2. Production rate distribution of Topsoil–Salvaging and Placing (m3/day)

Fig. 3. Distribution of the project duration (day)

These historical data driven activity duration distributions can empower schedulers and
build confidence in determining the total required time for a project and help them de-
fend their schedule with historical evidence.
1136 C. Le and H. D. Jeong

5 Conclusion

The primary contribution to the body of knowledge of this study is a DWR-based ap-
proach for schedule risk analysis. By applying regression analysis to the DWR data, the
probability distributions of the production rates of work activities can be developed
while considering the effects of project characteristics or influential factors on produc-
tion rates. These output distributions from the regression models are in turn used as
input for the MCS model. The project duration determination is a function of the pro-
duction rates of the work items, the quantities of work items, and the established logic
among work items. By applying MCS, the distribution of the project duration is
formulated. The project duration distribution is then used for schedule risk analysis,
such as identifying the level of certainty associated with a project duration.
Further research might explore the effect of multiple influential factors on produc-
tion rates with the use of and comparison among different regression models besides
simple linear regression like the case study. Additional data of the actual projects in
which schedulers use their judgment and experience to establish activity duration dis-
tributions for schedule risk analysis using PERT or MCS will be needed for a compar-
ison between the proposed approach and the existing methods. Sensitivity analysis can
also be applied to compare the various effects of the production rates of different work
activities on the project duration.

References
Cottrell, W. D. (1999). "Simplified Program Evaluation and Review Technique (PERT)."
Journal of Construction Engineering and Management, 125(1), 16-22.
FHWA (2002). "FHWA Guide for Construction Contract Time Determination Procedures."
Federal Highway Administration, Washington, D.C.
Hajdu, M., and Bokor, O. (2016). "Sensitivity analysis in PERT networks: Does activity duration
distribution matter?" Automation in Construction, 65, 1-8.
Harmelink, D. J., and Rowings, J. E. (1998). "Linear Scheduling Model: Development of
Controlling Activity Path." Journal of Construction Engineering and Management,
124(4), 263-268.
Kelly, P. M., Nagata, M. F., Sanders, M. C., and Thorndike, C. L. (2014). "The Great Debate: Is
CPM Past Its Prime?" AACE International: Cost Engineering.
Mattila, K. G., and Park, A. (2003). "Comparison of Linear Scheduling Model and Repetitive
Scheduling Method." Journal of Construction Engineering and Management, 129(1),
56-64.
Ostrowski, V. M. (2006). "Construction CPM Scheduling — Precision Without Accuracy."
AACE International Transactions.
Ramdass, S., and Lim, J. (2011). "A Simple Qualitative Approach for Schedule Risk Analysis."
AACE International Transactions.
Salas-Morera, L., Arauzo-Azofra, A., García-Hernández, L., Palomo-Romero, J. M., and Ayuso-
Muñoz, J. L. (2018). "New Approach to the Distribution of Project Completion Time
in PERT Networks." Journal of Construction Engineering and Management, 144(10).
$'HHS/HDUQLQJ%DVHG3URFHGXUHIRU6DIHW\(YDOXDWLRQ
RI6WHHO)UDPHV8VLQJ$GYDQFHG$QDO\VLV

0DQK+XQJ+D4XDQJ9LHW9XDQG9LHW+XQJ7UXRQJ
1DWLRQDO8QLYHUVLW\RI&LYLO(QJLQHHULQJ*LDL3KRQJ5RDG+DQRL9LHWQDP

KXQJKP#QXFHHGXYQ
6HMRQJ8QLYHUVLW\*XQMD'RQJ*ZDQJMLQ*X6HRXO6RXWK.RUHD

YLHWYT#YLPDUXHGXYQ
7KX\ORL8QLYHUVLW\7D\6RQ'RQJGD+DQRL9LHWQDP

WUXRQJYLHWKXQJ#WOXHGXYQ

$EVWUDFW7KHELJJHVWFKDOOHQJHRIDQDGYDQFHGDQDO\VLVLVWKDWLWLVWLPHFRQ
VXPLQJFRPSDUHGWRDQHODVWLFDQDO\VLV7KLVOHDGVWRWKHOLPLWDWLRQRIDSSOLFDWLRQ
RIDGYDQFHGDQDO\VLVLQFRPSOLFDWHGGHVLJQSUREOHPVVXFKDVRSWLPL]DWLRQGHVLJQ
DQGUHOLDELOLW\DQDO\VLVRIVWUXFWXUHVZKHUHDYHU\ODUJHQXPEHURIVWUXFWXUDODQDO
\VHVDUHUHTXLUHG7RVROYHWKLVSUREOHPDGHHSOHDUQLQJ'/EDVHGSURFHGXUH
IRU VDIHW\ HYDOXDWLRQ RI VWHHO IUDPHV LV LQWURGXFHG LQ WKLV VWXG\ $Q DGYDQFHG
DQDO\VLVLVXVHGWRFDSWXUHWKHQRQOLQHDULQHODVWLFEHKDYLRUVRIVWUXFWXUHV7ZR
'/PRGHOVDUHSURSRVHGIRUHVWLPDWLRQRIWKHVWUXFWXUDOORDGFDUU\LQJFDSDFLW\
DQGHYDOXDWLRQRIWKHVWUHQJWKVDIHW\RIWKHVWUXFWXUHE\SUHGLFWLQJWKHVWUXFWXUDO
XOWLPDWHORDGIDFWRUJUHDWHURUVPDOOHUWKDQ$WZRVWRU\VSDFHVWHHOIUDPHLV
VWXGLHG

.H\ZRUGV$GYDQFHGDQDO\VLV6WHHOIUDPH'HHSOHDUQLQJ0DFKLQHOHDUQLQJ

 ,QWURGXFWLRQ

$GYDQFHGDQDO\VLVPHWKRGVKDYHEHHQSURPRWHGDVHIIHFWLYHDOWHUQDWLYHVWRPHPEHU
EDVHGGHVLJQPHWKRGVLQGHVLJQRIVWHHOIUDPHVVLQFHWKHVHDSSURDFKHVFDQGLUHFWO\FDS
WXUHVWUXFWXUDOQRQOLQHDULQHODVWLFEHKDYLRUVDQGHOLPLQDWHWKHVHSDUDWHPHPEHUFDSDFLW\
FKHFNVHQFRPSDVVHGE\WKHVSHFLILFDWLRQHTXDWLRQV>@+RZHYHUWKHDSSOLFDWLRQRI
DGYDQFHGDQDO\VLVPHWKRGVLQSUDFWLFDOGHVLJQKDVEHHQOLPLWHGGXHWRWKHLUWLPHFRQ
VXPLQJFRVWHVSHFLDOO\IRUFRPSOLFDWHGSUREOHPVVXFKDVRSWLPL]DWLRQGHVLJQDQGUHOL
DELOLW\DQDO\VLVRIVWUXFWXUHVZKHUHDJUHDWQXPEHURIVWUXFWXUDODQDO\VHVDUHUHTXLUHG
>@$VROXWLRQWRWKLVSUREOHPLVWRXVHPHWDPRGHOVE\XVLQJPDFKLQHOHDUQLQJPHWK
RGV VXFK DV DUWLILFLDO QHXUDO QHWZRUN $11 .ULJLQJ SRO\QRPLDO UHVSRQVH VXUIDFH
PHWKRG560DQGVXSSRUWYHFWRUPDFKLQHV690$PHWDPRGHOLVDQDSSUR[LPDWH
PDWKHPDWLFDOUHSUHVHQWDWLRQXVHGWRGHVFULEHKLJKOHYHODEVWUDFWLRQVLQGDWD7KDQNVWR
PHWDPRGHOVWKHFRPSOH[QRQOLQHDUUHVSRQVHVRIVWUXFWXUHVFDQEHDSSUR[LPDWHGZLWK
RXWSHUIRUPLQJWLPHFRQVXPLQJDGYDQFHGDQDO\VHV
5HFHQWO\GHHSOHDUQLQJ'/DQHZHVWEUDQFKRIWKHUHSUHVHQWDWLRQOHDUQLQJPHWKRGV
KDVDGYDQFHGLQDGGUHVVLQJYDULRXVGHVLJQDQGDQDO\VLVSUREOHPVRIVWUXFWXUHVVXFKDV

© Springer Nature Singapore Pte Ltd. 2020 1137

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_182
1138 M.-H. Ha et al.

GDPDJHGHWHFWLRQKHDOWKPRQLWRULQJHWF>@0RVW'/PRGHOVDUHEDVHGRQ$11V
WKDWLQFOXGHPXOWLSOHOHYHOVRIUHSUHVHQWDWLRQE\XVLQJVLPSOHEXWQRQOLQHDULQWHUFRQ
QHFWHGOD\HUVZLWKPDQ\QHXURQVSHUOD\HUZKHUHWKHSUHVHQWDWLRQDWWKHIROORZLQJOD\
HUVKDVKLJKHUOHYHOVRIDEVWUDFWLRQWKDQWKHSUHYLRXVRQH7KURXJKWKHOD\HUVLPSRUWDQW
LQIRUPDWLRQLVDPSOLILHGZKLOHQRQFULWLFDOLQIRUPDWLRQLVJUDGXDOO\UHGXFHGDQGHOLP
LQDWHG:LWKVXFKDFRPSOLFDWHGDQGIOH[LEOHRUJDQL]DWLRQV\VWHP'/FDQKDQGOHFRP
SOH[DQGKLJKGLPHQVLRQDOGDWD)XUWKHUPRUHGHYHORSPHQWDQGXVHRID'/PRGHOGR
QRWUHTXLUHDKLJKH[SHUWLVHRIWKHXVHUVWKHUHIRUHWKHVHPHWKRGVDUHDSSOLFDEOHLQPDQ\
ILHOGVRIWHFKQRORJ\PHGLFDOEXVLQHVVDQGVFLHQFH
,QWKLVVWXG\ZHGHYHORSD'/EDVHGSURFHGXUHIRUVDIHW\HYDOXDWLRQRIVWHHOIUDPHV
$GYDQFHGDQDO\VLVLVXVHGWRFRQVLGHUQRQOLQHDULQHODVWLFEHKDYLRUVRIVWUXFWXUHV7ZR
'/PRGHOVDUHGHYHORSHGZKHUHWKHILUVWPRGHOLVWRHVWLPDWHWKHVWUXFWXUDOORDGFDU
U\LQJFDSDFLW\DQGWKHVHFRQGRQHLVWRHYDOXDWHWKHVWUXFWXUDOVDIHW\E\SUHGLFWLQJWKH
XOWLPDWHORDGIDFWRUJUHDWHURUVPDOOHUWKDQ7KHXOWLPDWHORDGIDFWRULVWKHUDWLRRI
VWUXFWXUDOORDGFDUU\LQJFDSDFLW\DQGDSSOLHGORDGLQJ$WZRVWRU\VSDFHVWHHOIUDPHLV
VWXGLHGWRGHPRQVWUDWHWKHHIILFLHQF\RIWKHGHYHORSHGPRGHOV7RWKHEHVWRIDXWKRUV¶
NQRZOHGJHQRVWXGLHVXVLQJ'/WRHYDOXDWHWKHXOWLPDWHORDGIDFWRURIQRQOLQHDULQH
ODVWLFVWHHOIUDPHVKDVEHHQSXEOLVKHGLQWKHOLWHUDWXUH

 'HHS/HDUQLQJLQ$UWLILFLDO1HXUDO1HWZRUN

/HDUQLQJIRUPRID'11PRGHOFDQEHFODVVLILHGDVVXSHUYLVHGIRUPIRUODEHOHG
GDWDXQVXSHUYLVHGIRUPIRUXQODEHOHGGDWDDQGVHPLVXSHUYLVHGIRUPIRUWUDLQ
LQJGDWDWKDWLQFOXGHVERWKODEHOHGDQGXQODEHOHGGDWD)RUWKHVDIHW\HYDOXDWLRQSURE
OHPVRIVWHHOIUDPHVWKHVXSHUYLVHGOHDUQLQJLVXVHGEHFDXVHWKHLQSXWDQGRXWSXWGDWD
DUHVSHFLILHG7KHEDVLFLVVXHVRIDVXSHUYLVHGOHDUQLQJIRUPDUHVXSHUYLVHGOHDUQLQJ
IHHGIRUZDUG QHXUDOQHWZRUNDQGEDFNSURSDJDWLRQDOJRULWKPWKDWDUH SUHVHQWHGLQWKH
IROORZLQJVHFWLRQV
 ,QDVXSHUYLVHGOHDUQLQJIRUPWKHWUDLQLQJGDWDLVH[SUHVVHGDV
1
 7 ^ ;  < `
L L L 
 

WK
ZKHUH ; L LVWKHLQSXWYHFWRUIHDWXUHYHFWRURIWKHGDWD L  <L LVWKHRXWSXWYHFWRURU
WK
WKHODEHORIWKHGDWD L  1 LVWKHQXPEHURIGDWDVDPSOHV6LQFHWKHH[DFWDQVZHUVDUH
JLYHQWKHUHVXOWVRIWKHWUDLQLQJSURFHVVFDQEHFRQWUROOHGE\PLQLPL]LQJWKHHUURUEH
WZHHQWKHSUHGLFWHGDQGH[DFWYDOXHVDVIROORZV)RUH[DPSOHZLWKWKHLQSXW ; L WKH
WUDLQLQJPRGHOJLYHVWKHSUHGLFWHGUHVXOW <L
7RLPSURYHWKHWUDLQLQJPRGHOZHFDQ
XVHWKHRSWLPL]DWLRQSURFHVVWRPLQLPL]HWKHPHDQVTXDUHHUURUIXQFWLRQJLYHQDV

 1 
 (06( ¦
1 L
 <L  <L
  

 $ IHHGIRUZDUG 11 LQFOXGHV PDQ\ OD\HUV DQG RQH OD\HU KDV PDQ\ XQLWV 7KH ILUVW
OD\HULVWKHLQSXWOD\HUZKHUHHDFKXQLWLQWKLVOD\HULVDIHDWXUHRIWKHGDWD7KHQXPEHU
A Deep Learning-Based Procedure for Safety Evaluation... 1139

RIXQLWVLQWKHLQSXWOD\HUFDQEHHTXDORUGLIIHUHQWWRWKHQXPEHURIIHDWXUHVRIWKHGDWD
7KHODVWOD\HULVWKHRXWSXWOD\HULQZKLFKHDFKXQLWLVDODEHORIWKHGDWD7KHQXPEHU
RIXQLWVLQWKLVOD\HUPXVWEHHTXDOWRWKHQXPEHURIODEHOVRIWKHGDWD5HPDLQLQJOD\HUV
DUHFDOOHGDVWKHKLGGHQOD\HUVFRQWDLQLQJPDQ\XQLWVLQHDFKOD\HU7KHKLGGHQOD\HUV
SOD\ WKHUROHRIWUDQVIHUULQJWKHLPSRUWDQWIHDWXUHVIURPWKHLQSXWOD\HUWRWKH RXWSXW
OD\HU,QDIHHGIRUZDUG11WKHLQIRUPDWLRQPRYHVLQRQO\RQHGLUHFWLRQIURPWKHLQSXW
OD\HUWKURXJKWKHKLGGHQOD\HUVWRWKHRXWSXWOD\HU)LJ SUHVHQWVDQH[DPSOHRID
IXOO\ FRQQHFWHG IHHGIRUZDUG 11ZLWKKLGGHQOD\HUV$VFDQEHVHHQLQ WKLVILJXUH
RQO\WKHRXWSXWVRIWKHXQLWVLQDOD\HUDUHFRQQHFWHGWRWKHLQSXWVRIWKHQH[WOD\HUXQLWV
7KLVPHDQVWKDWWKHUHLVQRIHHGEDFNLQWKHV\VWHP

 I  
Z  I  
 X Y X Y 
Z
Z
[ \
\







[
I  I 


   
XL YL XL YL \L
\L
[L








[, \+
\+
 I  I 
Z ,- X-
 
Y- 

X.

Y.

Z .+
Z -.

,QSXWOD\HU +LGGHQOD\HU +LGGHQOD\HU 2XWSXWOD\HU ([DFWRXWSXW


)LJ)HHGIRUZDUGQHXUDOQHWZRUNZLWKWZRKLGGHQOD\HUV

WK
7KHLQSXW XL M  RIWKHXQLW L RIWKHOD\HU M WK LVFDOFXODWHGDVIROORZV
 M 
1XQLW
 XL M  ¦Z
O 
 M
OL  

WK
ZKHUH ZOL M  LVWKHZHLJKWDWWKHFRQQHFWLRQEHWZHHQWKHXQLW O WK RIWKHOD\HU  M   
WK
LVWKHQXPEHURIXQLWRIWKHOD\HU  M   
WK
DQGWKHXQLW L RIWKHOD\HU M WK  1 XQLW
 M 

 ,QHDFKOD\HUDQDFWLYDWLRQIXQFWLRQ I   LVXVHGWRWUDQVIRUPWKHLQSXWVLJQDO XL M  
WK
WRWKHRXWSXWVLJQDO YL M  LQWKHXQLW L LQWKHOD\HU M WK DVIROORZV

 YL M  I  XL M    

1RWHVWKDWZHFDQFKRRVHGLIIHUHQWDFWLYDWLRQIXQFWLRQVIRUGLIIHUHQWOD\HUV6RPHSRS
XODU DFWLYDWLRQ IXQFWLRQV DUH VLJPRLG OLQHDU VRIWSOXV VRIWPD[ UHFWLILHG OLQHDU XQLW
5H/8HWF8VHUVKDYHWRFKRRVHDSSURSULDWHDFWLYDWLRQIXQFWLRQVIRUWKHLUSUREOHPV
 *HQHUDOO\WKHIRUZDUGSDVVVWDUWVE\UDQGRPO\LQLWLDOL]LQJWKHZHLJKWV7KHLQIRU
PDWLRQXVLQJWKRVHZHLJKWVPRYHVLQRQO\RQHGLUHFWLRQIURPWKHLQSXWOD\HUWKURXJK
WKHKLGGHQOD\HUVWRWKHRXWSXWOD\HU7KHZHLJKWYDOXHVDUHWKHQUHYLVHGFRQWLQXRXVO\
XQWLOWKHHUURUEHWZHHQWKHSUHGLFWHGRXWSXWDQGH[DFWRXWSXWRUORVVIXQFWLRQLVVPDOO
1140 M.-H. Ha et al.

HQRXJK 7KLV PHDQVWKDW WKHWDVN RIWUDLQLQJSURFHVV RID '11 PRGHO LV WR ILQG WKH
RSWLPDOZHLJKWSDUDPHWHUVWRPLQLPL]HWKHORVVIXQFWLRQ7RRSWLPL]HWKHZHLJKWSD
UDPHWHUVWKHEDFNSURSDJDWLRQDOJRULWKPLVFRPPRQO\XVHGEDVHGRQWKHJUDGLHQWGH
VFHQW PHWKRG ,Q WKH EDFNSURSDJDWLRQ DOJRULWKP WKH EDFNZDUG SDVV LV SHUIRUPHG WR
UHYLVHWKHZHLJKWSDUDPHWHUVIURPWKHRXWSXWOD\HUWKURXJKWKHKLGGHQOD\HUVWRWKH
LQSXWOD\HULQRUGHUWRILQGWKHQHZZHLJKWVZKLFKOHDGWRVPDOOHUORVVIXQFWLRQ/RVV
IXQFWLRQVDUHDOVRGLYLGHGLQWRW\SHVUHJUHVVLRQDQGFODVVLILFDWLRQORVVIXQFWLRQVFRU
UHVSRQGLQJ WR UHJUHVVLRQ DQG FODVVLILFDWLRQ SUREOHPV 6RPH SRSXODU UHJUHVVLRQ ORVV
IXQFWLRQVDUHPHDQBVTXDUHGBHUURUPHDQBDEVROXWHBHUURUHWF6RPHH[DPSOHVRIFODV
VLILFDWLRQORVVIXQFWLRQVDUHORJBORVVIRFDOBORVVH[SRQHQWLDOBORVVHWF

 'DWD*HQHUDWLRQ

*HQHUDWLQJGDWDLVDFUXFLDOVWHSIRUWUDLQLQJPRGHOV'DWDPXVWEHHQVXUHGLQWHUPVRI
VDPSOHTXDQWLW\LQSXWVDQGRXWSXWV,QWKLVVWXG\WKHLQSXWVDUHOLPLWHGWRWKHSURSHU
WLHVRIFURVVVHFWLRQVRIEHDPVDQGFROXPQV7KHUHDUHIHDWXUHVIRUHDFK:VKDSHG
VHFWLRQFRQVLGHUHG DV OLVWHG LQ 7DEOH 7KH RXWSXW LV WKH XOWLPDWHORDGIDFWRU RIWKH
VWUXFWXUHWKHUDWLRRIWKHVWUXFWXUDOORDGFDUU\LQJFDSDFLW\WRWKHLQLWLDODSSOLHGORDG
ZLWKWKHUHJUHVVLRQSUREOHPV,QWKHFODVVLILFDWLRQSUREOHPVLIWKHXOWLPDWHORDGIDFWRU
RIWKHVWUXFWXUHLVVPDOOHUWKDQRXWSXWHTXDOV]HURRWKHUZLVHLWHTXDOV

7DEOH3URSHUWLHVRI:VHFWLRQFRQVLGHUHGDVLQSXWV

 &URVVVHFWLRQSURSHUWLHV  &URVVVHFWLRQSURSHUWLHV
 :DUSLQJFRHIILFLHQW  6KHDUDUHDIRUVKHDUDORQJORFDO\D[LV
 *\UDWLRQUDGLXVRIDERXWZHDNEHQGLQJD[LV  6KHDUDUHDIRUVKHDUDORQJORFDO]D[LV
 6HFWLRQPRGXOXVDERXW\D[LV  3ODVWLFVHFWLRQPRGXOXVDERXWORFDO\D[LV
 6HFWLRQPRGXOXVDERXW]D[LV  3ODVWLFVHFWLRQPRGXOXVDERXWORFDO]D[LV
 7RUVLRQDOFRQVWDQW  6HFWLRQGHSWK
 %HQGLQJLQHUWLDDERXWORFDO\D[LV  6HFWLRQIODQJHZLGWK
 %HQGLQJLQHUWLDDERXWORFDO]D[LV  6HFWLRQIODQJHWKLFNQHVV
 &URVVVHFWLRQDUHD  6HFWLRQZHEWKLFNQHVV
 7RGHWHUPLQHWKHXOWLPDWHORDGIDFWRURIDVWHHOIUDPHDQDGYDQFHGDQDO\VLVXVLQJ
SODVWLFKLQJHEHDPFROXPQHOHPHQWLVXVHGWKURXJKWKH3UDFWLFDO$GYDQFHG$QDO\VLV
3URJUDP3$$3>@&RPSDUHGWRSODVWLF]RQHPHWKRGVWKLVDSSURDFKFRQVXPHVPXFK
VKRUWHUFRPSXWDWLRQDOHIIRUW7KLVLVYHU\XVHIXOVLQFHWKHDPRXQWRIGDWDWREHFUHDWHG
LQWKLVVWXG\LVTXLWHODUJH

 &DVH6WXG\

$WZRVWRU\VSDFHVWHHOIUDPHSUHVHQWHGLQ)LJLVVWXGLHG&URVVVHFWLRQVRIEHDPV
DQGFROXPQVDUHGLYLGHGLQWRDQGJURXSVUHVSHFWLYHO\7KLVPHDQVWKDWWKHUHDUH
LQSXWYDULDEOHVIRUWUDLQLQJPRGHOV&ROXPQPHPEHUVDUHGHVLJQHGXVLQJ:DQG
:ZKHUHDV:::DQG:DUHXVHGIRUEHDPPHPEHUV7KHLQLWLDOVWRU\
RXWRISOXPEQHVVDFFRUGLQJWRWKH;D[LVLV   7KHPDWHULDOLVWKH$VWHHOZLWK
A Deep Learning-Based Procedure for Safety Evaluation... 1141

WKH\LHOGVWUHQJWKRI03DDQGWKHHODVWLFPRGXOXVRI03D7KHZLQGORDGV
DUHN1LQWKH;GLUHFWLRQDWHYHU\EHDPFROXPQMRLQWVDQGWKHYHUWLFDOGLVWULEXWHG
ORDGVRQWKHEHDPVDUHN1P
7KHUHJUHVVLRQPRGHOIRUHVWLPDWLQJWKHXOWLPDWHORDGIDFWRURIWKHIUDPHLVGHYHO
RSHGILUVW7KHQHWZRUNRIWKHWUDLQLQJPRGHOLV7KHQXPEHURIHSRFKV
LV  7KH REMHFWLYH IXQFWLRQ RI WKH WUDLQLQJ PRGHO LV WKH ³PHDQVTXDUHGHUURU´
06( 7KH DFWLYDWLRQ IXQFWLRQ LV ³/HDN\5H/8¶ WKH RSWLPL]HU LV ³$GDP´ DQG WKH
QXPEHURIEDWFKVL]HVLVHTXDOWR³VDPSOHV´7KHQXPEHURIWKHWHVWGDWDLVHTXDO
WR%\FKDQJLQJWKHQXPEHURIWKHWUDLQLQJGDWDZHREWDLQHGWKHFXUYHRIWKH
REMHFWLYHIXQFWLRQDFFRUGLQJWRQXPEHURIWKHWUDLQLQJGDWDDVSUHVHQWHGLQ)LJ)URP
WKLVILJXUHZKHQWKHQXPEHURIWUDLQLQJVDPSOHVLVJUHDWHUWKDQWKH06(RIWKH
WUDLQLQJ PRGHOLVRQO\7KHUHIRUHZHFDQFRQFOXGHWKDWWKH'/PRGHOFDQ
HVWLPDWHWKHXOWLPDWHORDGIDFWRURIWKHIUDPHZLWKKLJKDFFXUDF\ZKHQWKHQXPEHURI
WUDLQLQJVDPSOHVLVJUHDWHQRXJK
S * UR X S 
S
RX RX
*U *U

*URXS
* UR X S 

P
*URXS
*URXS

*URXS

S   * UR X S  S 
RX RX
*U *U
*URXS

* UR X S 
P
*URXS
*URXS

*URXS

       

=
   
<   P
;
  P

)LJ7ZRVWRU\VSDFHIUDPH
ϯ͘ϬͲϬϯ
ǀĞƌĂŐĞ06( ŽĨWKHŵŽĚĞů

Ϯ͘ϱͲϬϯ >
Ϯ͘ϬͲϬϯ
ϭ͘ϱͲϬϯ
ϭ͘ϬͲϬϯ
ϱ͘ϬͲϬϰ
Ϭ͘ϬнϬϬ
Ϭ ϱ͕ϬϬϬ ϭϬ͕ϬϬϬ ϭϱ͕ϬϬϬ ϮϬ͕ϬϬϬ
EƵŵďĞƌŽĨƚƌĂŝŶŝŶŐƐĂŵƉůĞƐ 
)LJ$YHUDJH06(RIXOWLPDWHORDGIDFWRUUHJUHVVLRQPRGHORIWKHIUDPH
7KHFODVVLILFDWLRQPRGHOIRUHYDOXDWLQJWKHVDIHW\RIWKHIUDPHLVQRZFRQVLGHUHG7KH
SDUDPHWHUVRIWKHWUDLQLQJPRGHODUHWKHVDPHZLWKWKHFDVHRIUHJUHVVLRQPRGHOH[FHSW
IRUWKHREMHFWLYHIXQFWLRQZKLFKLV³DFFXUDF\´7KHFXUYHRIWKHDFFXUDF\RIWKHWUDLQLQJ
PRGHOLVSUHVHQWHGLQ)LJ)URPWKLVILJXUHZKHQWKHQXPEHURIWUDLQLQJVDPSOHVLV
JUHDWHUWKDQWKHDFFXUDF\RIWKHWUDLQLQJPRGHOLVJUHDWHUWKDQ7KLVSURYHV
WKDWWKHFODVVLILFDWLRQPRGHOXVLQJ'/LVKLJKO\DFFXUDWHZKHQWKHQXPEHURIWUDLQLQJ
VDPSOHVLVJUHDWHQRXJK
1142 M.-H. Ha et al.

ϭ͘ϬϬ
Ϭ͘ϵϱ >

ǀĞƌĂŐĞĐůĂƐƐŝĨŝĐĂƚŝŽŶ
ĂĐĐƵƌĂĐLJŽĨŵŽĚĞů
Ϭ͘ϵϬ
Ϭ͘ϴϱ
Ϭ͘ϴϬ
Ϭ͘ϳϱ
Ϭ͘ϳϬ
Ϭ͘ϲϱ
Ϭ͘ϲϬ
ϭ͕ϬϬϬ Ϯ͕ϬϬϬ ϱ͕ϬϬϬ ϭϬ͕ϬϬϬ ϮϬ͕ϬϬϬ
EƵŵďĞƌŽĨƚƌĂŝŶŝŶŐƐĂŵƉůĞƐ 
)LJ$YHUDJHDFFXUDF\RIXOWLPDWHORDGIDFWRUFODVVLILFDWLRQPRGHORIWKHIUDPH

 &RQFOXVLRQ

7KHIROORZLQJFRQFOXVLRQVFDQEHGUDZQIURPWKLVZRUN
¾ $'/EDVHGPRGHOIRUHVWLPDWLQJWKHXOWLPDWHORDGIDFWRURIQRQOLQHDULQHODVWLF
VWHHOIUDPHVLVSURSRVHG
¾ $QRWKHU '/EDVHG PRGHO IRU HYDOXDWLQJ WKHVDIHW\ RI QRQOLQHDU LQHODVWLF VWHHO
IUDPHVLVGHYHORSHG
¾ 5HVXOWV RI WKH VWXG\ RI D WZRVWRU\ VSDFH IUDPHSURYHWKDWWKHGHYHORSHG '/
EDVHGPRGHOVDUHKLJKO\DFFXUDWH

$FNQRZOHGJHPHQW

7KLVUHVHDUFKLVIXQGHGE\9LHWQDP1DWLRQDO)RXQGDWLRQIRU6FLHQFHDQG7HFKQRORJ\
'HYHORSPHQW1$)267('XQGHUJUDQWQXPEHU

5HIHUHQFHV
&KLRUHDQ*&6HFRQGRUGHUIOH[LELOLW\EDVHGPRGHOIRUQRQOLQHDULQHODVWLFDQDO\VLVRI'
VHPLULJLGVWHHOIUDPHZRUNV(QJLQHHULQJ6WUXFWXUHV
7KDL+7.LP6(1RQOLQHDULQHODVWLFWLPHKLVWRU\DQDO\VLVRIWUXVVVWUXFWXUHV-RXUQDORI
&RQVWUXFWLRQDO6WHHO5HVHDUFK
7UXRQJ 9+ .LP 6( $Q HIILFLHQW PHWKRG RI V\VWHP UHOLDELOLW\ DQDO\VLV RI VWHHO FDEOH
VWD\HGEULGJHV$GYDQFHVLQ(QJLQHHULQJ6RIWZDUH
7UXRQJ9+1JX\HQ3&.LP6($QHIILFLHQWPHWKRGIRURSWLPL]LQJVSDFHVWHHOIUDPHV
ZLWK VHPLULJLG MRLQWV XVLQJ SUDFWLFDO DGYDQFHGDQDO\VLVDQGWKH PLFURJHQHWLFDOJRULWKP
-RXUQDORI&RQVWUXFWLRQDO6WHHO5HVHDUFK
7UXRQJ9+.LP6($UREXVWPHWKRGIRURSWLPL]DWLRQRIVHPLULJLGVWHHOIUDPHVVXEMHFW
WRVHLVPLFORDGLQJ-RXUQDORI&RQVWUXFWLRQDO6WHHO5HVHDUFK
&KD<-&KRL:%\N|]WUN2'HHSOHDUQLQJEDVHGFUDFNGDPDJHGHWHFWLRQXVLQJ
/H&XQ<%HQJLR<+LQWRQ*'HHSOHDUQLQJ1DWXUH
7KDL+7.LP6(3UDFWLFDODGYDQFHGDQDO\VLVVRIWZDUHIRUQRQOLQHDULQHODVWLFDQDO\VLVRI
VSDFHVWHHOVWUXFWXUHV$GYDQFHVLQ(QJLQHHULQJ6RIWZDUH
 Prediction of buckling coefficient of stiffened plate
girders using deep learning algorithm

George Papazafeiropoulos1, Quang-Viet Vu2*, Viet-Hung Truong3*, Minh-Chinh

Luong3, Van-Trung Pham2
1
National Technical University of Athens, Athens, Greece
gpapazafeiropoulos@yahoo.gr
2
Faculty of Civil Engineering, Vietnam Maritime University, 484 Lach Tray Street, Le Chan
District, Hai Phong City, Vietnam.
vietvq@vimaru.edu.vn
pvtrung@vimaru.edu.vn
3
Faculty of Civil Engineering, Thuyloi University, 175 TaySon, Dongda, Hanoi, Vietnam
truongviethung@tlu.edu.vn
chinhlm@tlu.edu.vn
* Corresponding author

Abstract. This paper aims at introducing a new method to determine the buck-
ling coefficient kb of the stiffened plate girders under pure bending using deep
learning, one of the most powerful algorithms in machine learning. Firstly, out-
put data kb is generated from eigenvalue buckling analyses based on input data
(various geometric dimensions of the girder). This procedure is implemented by
using the Abaqus2Matlab toolbox, which allows the transfer of data between
Matlab and Abaqus and vice versa. After that, 2,200 training data are used to
build the model for predicting kb using deep learning. Finally, 200 test data are
used to evaluate the accuracy of the model. The results obtained from this mod-
el are also compared with analogous results of previous works with a good
agreement.

Keywords: Stiffened plate girders; stiffeners; Abaqus2Matlab; deep learning.

1 Introduction
In deep plate girders having slender webs, the use of longitudinal stiffeners is very
important since they can improve the web strength against bend-buckling. In design
of stiffened plate girders, the buckling coefficient kb plays the most important role
because it is employed to identify the girder buckling strength. Therefore, the accurate
computation of kb is necessary. The kb of longitudinally stiffened web plates in the
AASHTO LRFD specifications [1] was firstly suggested by Frank and Helwig [2].
The kb was separately proposed for cases when ds/Dc t 0.4 and ds/Dc  0.4, where
ds/Dc = 0.4 is recommended to be the optimum location of a single stiffener. Howev-
er, in these equations the effect of the presence of flanges, web thickness, stiffener
dimensions, aspect ratio, etc., on kb is not considered. Recently, Cho and Shin [3], and

© Springer Nature Singapore Pte Ltd. 2020 1143

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_183
1144 G. Papazafeiropoulos et al.

Kim et al. [4] suggested new equations to compute k b, in which they proposed the
optimum stiffener location to be ds/Dc = 0.425 and ds/Dc = 0.42, respectively. Never-
theless, these equations have not yet considered some factors mentioned above.
In recent years, deep learning (DL), a new category of machine learning methods,
has been used successfully in solving numerous structural analysis problems e.g.,
damage detection [5], health monitoring [6], prediction [7], etc. Most DL models are
based on an ANN which includes representation multiple levels by using simple but
nonlinear interconnected layers with a large amount of neurons per layer, in which the
following layers present higher abstraction levels than the previous ones. Through
these layers, important information is intensified, while non-critical information is
gradually decreased and excluded. For a complex organization system, DL can solve
for complicated and high-dimensional data. In addition, the use of the DL model does
not require a high expertise of the users, therefore this method is applied in many
fields of technology, medical, business, and science.
In this study, we develop a DL-based procedure for prediction of kb of longitudi-
nally stiffened web plate girders subjected to pure bending. Datasets, consisting of
input data (various geometric dimensions of the girder) and output data kb, are gener-
ated from eigenvalue buckling analyses. This procedure is performed by using a
toolbox, Abaqus2Matlab [8], which allows the data transfer between Matlab and
Abaqus and vice versa. After that, 2,200 training data are employed to establish the
model to predict the value of kb using deep learning. The number of hidden layers and
the number of neurons in each layer, optimizer and activation function are chosen to
get the best result in training data. Finally, 200 test data are utilized to estimate the
model accuracy.

2 Deep Learning in Artificial Neural Network

In the current study, only the supervised learning is taken into consideration
since the input and output data are specified.
In supervised learning form, the training data is expressed as
N
T ^ Xi , Yi ` , i 1
(1)
where Xi is the input vector of the data i th or the feature vector of the data i th ; Yi is
th
the output vector or the label of the data i ; N is the number of data samples. Since
the exact values are given, the training process results can be controlled by minimiz-
ing the error between the exact and predicted values. For instance, the training model
gives the predicted result Yi' corresponding to the input Xi . To enhance the training
model, the optimization process is employed to minimize the mean-square error
(MSE) function given as follows:
1 N
¦ Y  Y '  . (2)
2
E MSE i i
N i 1

A feedforward neural network (FNN) includes many layers and one layer has
many units. The first layer plays the role of the input layer in which each unit is a data
feature. In the input layer, the number of units and the number of the data feature can
be equal or different from each other. The last layer is the output layer where each
Prediction of buckling coefficient of stiffened plate... 1145

unit is a data label. The number of units and the number of the data labels must be
identical. The remaining layers are the hidden layers which contain many units in
each layer. Fig. 3, presents an example of a fully connected FNN with 2 hidden lay-
ers.
(2)
(1) f () (1)
w11 (2) f () (2)
(1) u1 v1 u1 v1 (3)
w11
w11
x1 y1' y1

......

......

......

......
x2
f () f ()

......
(1) (1) (2) (2)
ui vi ui vi yi' yi
xi

......

......
......

......
...

xI yH ' yH
(1) f () f ()
wIJ uJ
(1) (1)
vJ (2) uK
(2) (2)
vK
(3)
wKH
wJK

Input layer Hidden layer 1 Hidden layer 2 Output layer Exact output

Fig. 1. FNN with two hidden layers

The input of a unit is computed based on the weight function given as follows:
( j 1)
N unit
ui( j ) ¦w ,
l 1
( j)
li
(3)

where ui( j ) is the input of the i th unit of the j th layer; wli( j ) is the weight at the connec-
tion between the unit l th of the  j 1th layer and the i th unit of the j th layer; Nunit ( j 1)
is
the number of units of the  j 1th layer. In each layer, to transform the input signal to
the output signal in the unit, an activation function f   is employed as follows:
vi( j ) f ui( j )  (4)
th th
where v is the output of the i unit in the j layer. It is noted that different activa-
( j)
i

tion functions can be chosen for different layers. The users should select the suitable
activation functions for their problems.
In multilayer FNN, a BP algorithm is commonly used to compute the influence
of each weight corresponding to a loss function based on the gradient descent method.
In the BP algorithm, the backward pass is implemented to update the weight functions
from the output layer through the hidden layers to the input layer in order to search
the new weights so that they minimize the loss function. In the BP algorithm, the
weight functions can be updated as follows:
W( j 1) W( j )  HwW( j )  P'W( j 1) (5)
where W is the weight functions matrix of the j epoch; H is a learning rate used
( j ) th

to control the ratio of the weights adjusted; P is a momentum parameter used to

maintain the influence of the previous changes of the weights on the current move-
ment direction in the weight space.

3 Data Generation

In this study, a steel plate girder is considered with the height and the thickness of the
web chosen to be D = 3.0 m and t = 9.0 mm respectively, which are identical with the
model used in the paper of Viet et al. [9]. The flange width and thickness of the girder
1146 G. Papazafeiropoulos et al.

are set equal to 600 mm and 54 mm, respectively for verification purposes with the
results given in the paper of Elbanna et al. [10]. Datasets include input data consisting
of nine parameters of the girders and output data consisting of kb. These input pa-
rameters and their ranges which are chosen to satisfy the AASHTO LRFD require-
ments and cover all dimensions in practical designs are shown in Table 1.
Table 1. Input parameters of steel plate girders
No Girder geometric dimension Constraint (upper and lower bounds)
1 Top flange width (bfc) D/6 (500 mm) d bft/tft d 24
2 Top flange thickness (tfc) 8 ± 50 mm
3 Bottom flange width (bft) D/6 (500 mm) d bft/tft d 24
4 Bottom flange thickness (t ft) 8 ± 50 mm
5 Stiffener width (bs) 6 d bs/ts d 12
6 Stiffener thickness (ts) 8 ± 30 mm
7 Web thickness (tw) D/300 (10 mm) to D/150 (20 mm)
8 Stiffener location (ds) 0 to 3,000 mm
9 Aspect ratio (a) 0.5 to 3.0
The output data kb is generated from eigenvalue buckling analyses based on input
data through the Abaqus2Matlab toolbox [8], which integrates Matlab and Abaqus.
This procedure has been introduced by Viet et al. [11] for steel plates. For steel plate
girders, firstly a verification is made to demonstrate the accuracy of the finite element
(FE) model as shown in Fig. 2. The results obtained from these models are compared
with those given by Elbanna et al. [10] with a good agreement as shown in Table 2.
Secondly, Abaqus2Matlab is used to generate 2,200 output data for kb based on ran-
domly choosing 2,200 combinations of input parameters in their range.
ञढट

रढट रढट
ञय ठय
रय रय

ऀ ऀ औ

रढर रढर
झ ञढर

a) Geometry of steel plate girder b) Applied load in the model

Fig. 2. Steel plate girder model
Table 2. Comparison of kb obtained from present study and Elbanna et al. [10]
Aspect ratio Stiffener Buckling coefficient, kb Difference
a location, ds (%)
FEA Elbanna et al. [10]
1 0.21D 176.5 177.25 0.42
2 0.21D 176.94 180.07 1.74
Prediction of buckling coefficient of stiffened plate... 1147

4 Results and discussion

In order to build a suitable deep learning model for prediction of k b, several fea-
tures of this model should be adjusted to get the best results. In this study, ReLU acti-
vation function is employed and the Adam optimizer is selected for all training model.
The number of hidden layers is chosen to be 2, while the number of neurons in each
layer is selected as 12 which is higher than the number of input data. It is noted that
2,200 datasets generated from Abaqus2Matlab toolbox are randomly divided by 2,000
training data and 200 test data. 7KHREMHFWLYHIXQFWLRQRIWKHWUDLQLQJPRGHOLV³PHDQ-
squared-errRU´06(.
Fig. 3 shows the comparison of average MSE between training data and test data of
the model starting from a high value and decreasing to smaller values. In this case, the
accuracy of training and test model is high since the average MSEs obtained from
these models are very small. The minimum values of average MSE of training and
test data are 10.93 and 37.5, respectively at the 10,000 epochs. Therefore, the relative
errors of training and test model are 4.9% and 6.46%. This means that the deep learn-
ing model developed herein is reliable for prediction of k b of longitudinally stiffened
web plate girders subject to pure bending.

Fig. 3. Comparison of average MSE between training data and test data

In order to demonstrate the efficiency of this model, two cases of kb obtained from
this model are used to validate with the results given in [10] as presented in Table 3.

Table 3. Comparison of kb obtained from the DL model and Elbanna et al. [10]
Aspect ratio Stiffener Buckling coefficient, kb Difference
a location, ds (%)
DL model Elbanna et al. [10]
1 0.2D 171.7 164.2 4.4
2 0.2D 176.1 167.4 4.9

As seen in this table, the results obtained from deep learning model are close to
those given in [10]. Therefore, it is concluded that the developed DL model can be
used to accurately predict the value of kb of longitudinally stiffened web plate girders
subject to pure bending.
1148 G. Papazafeiropoulos et al.

5 Conclusion

This paper presents an efficient method for predicting kb of longitudinally stiffened

web plate girders subjected to pure bending using DL. Datasets consist of 2,200 train-
ing data and 200 test data. Nine input parameters consisting of (ds, bs, ts, bfc, tfc, bft, tft,
tw, and a) are taken into consideration to cover the effect of all geometrical features of
plate girders. Output result is kb which can be generated based on eigenvalue buckling
analyses through Abaqus2Matlab toolbox, which allows the transfer of data between
Matlab and Abaqus and vice versa. The efficiency of the DL model is verified by
comparison of its results with those obtained from the literature which showed a good
agreement. It is demonstrated that the DL model is reliable and can accurately predict
kb of longitudinally stiffened web plate girders under pure bending.

Acknowledgments

This research is funded by Vietnam National Foundation for Science and Technology
Development (NAFOSTED) under grant number 107.01-2018.327.

References
1. AASHTO, AASHTO LRFD Bridge design specifications, Sixth, 2012.
2. K.H.Frank, T.A.Helwig, Buckling of webs in unsymmetric plate girders, Eng. Journal,
Second Quart. (1995) 43±53.
3. E.Y. Cho, D.K. Shin, Elastic web bend-buckling analysis of longitudinally stiffened I-
section girders, Int. J. Steel Struct. 11 (2011) 297±313.
4. H.S. Kim, Y.M. Park, B.J. Kim, K. Kim, Numerical investigation of buckling strength of
longitudinally stiffened web of plate girders subjected to bending, Struct. Eng. Mech. 65
(2018) 141±154.
5. Pathirage, CSN, Li, J, Li, L, Hao, H, Liu, W, Ni, P: Structural damage identification based
on autoencoder neural networks and deep learning. Engineering Structures 172: 13-28
(2018).
6. Rafiei, MH, Adeli, H: A novel unsupervised deep learning model for global and local
health condition assessment of structrures. Engineering Structures 156: 598-607 (2018).
7. M. Ahmadi, H. Naderpour, A. Kheyroddin: Utilization of artificial neural networks to pre-
diction of the capacity of CCFT short columns subject to short term axial load. Archives of
civil and mechanical engineering 14: 510-517(2014).
8. G. Papazafeiropoulos, M. Muñiz-Calvente, E. Martínez-Pañeda, Abaqus2Matlab: A
suitable tool for finite element post-processing, Adv. Eng. Softw. 105 (2017) 9±16.
9. Q. Vu, G. Papazafeiropoulos, C. Graciano, S. Kim, Optimum linear buckling analysis of
longitudinally multi-stiffened steel plates subjected to combined bending and shear, Thin
Walled Struct. 136 (2019) 235±245.
10. A.A. Elbanna, H.M. Ramadan, S.A. Mourad, Buckling enhancement of longitudinally and
vertically stiffened plate girders, J. Engineeing Appl. Sci. 61 (2014) 351±370.
11. Q. Vu, V. Truong, G. Papazafeiropoulos, C. Graciano, S. Kim, Bend-buckling strength of
steel plates with multiple longitudinal stiffeners, J. Constr. Steel Res. 158 (2019) 41-52.
 Monitoring Pedestrian Flow on Campus with Multiple
Cameras using Computer Vision and Deep Learning
Techniques

Peter Kok-Yiu Wong1[0000-0003-1758-675X] and Jack C.P. Cheng2[0000-0002-1722-2617]

1 The Hong Kong University of Science and Technology, Hong Kong
kywongaz@connect.ust.hk
2
The Hong Kong University of Science and Technology, Hong Kong
cejcheng@ust.hk

Abstract. This paper proposes a robust method for multi-camera person re-iden-
tification (ReID), which can accurately track pedestrians across non-overlapping
cameras. Closed-circuit television (CCTV) are widely used to capture pedestrian
movement in different places. By integrating CCTV with computer vision and
deep learning techniques, trajectory of individual pedestrian can be efficiently
acquired for analyzing pedestrian walking behaviors. Many existing ReID meth-
ods aim to extract discriminative human features to distinguish a person from
others. Recent state-of-the-art performance is achieved mostly by obtaining fine
features from each body part. However, these part-based feature extraction meth-
ods did not consider which parts are more useful for person ReID. Therefore, this
paper proposes a weighted-parts feature extraction method, such that features of
specific body parts are more influential to identity prediction. After comparing
the performances of utilizing each part alone, several parts are considered more
view-invariant and discriminative. Higher weights are then imposed on these spe-
cific parts to extract more useful human features for person ReID. Experimental
results with videos on a college campus show that the ReID accuracy of our pro-
posed method notably outperforms many existing ones.

Keywords: Computer vision, Deep learning, Human feature extraction, Multi-

camera re-identification, Pedestrian flow analytics.

1 Introduction

Wide networks of closed-circuit television (CCTV) have been commonly deployed in

different places, such as college campuses, public estates, mass transit stations. Under-
standing pedestrian movement enables urban designers to analyze pedestrian behavior
in a neighborhood, such that pedestrian-friendly infrastructural design can be studied
[1]. Traditionally, video analytics require intensive manual effort to monitor multiple
cameras, which are usually scattered among a building or district with non-overlapping
field-of-view. Long-term movement of individual person needs to be tracked across
multiple cameras, which requires an accurate identity mapping when given two or more

© Springer Nature Singapore Pte Ltd. 2020 1149

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_184
1150 P. K.-Y. Wong and J. C. P. Cheng

persons captured by different cameras. This problem can be termed as person re-iden-
tification (ReID). Nevertheless, in a congested district, manually observing the long-
term and widespread movement of each person can be erroneous and time-consuming.
An image-based ReID problem is that given a set of images taken from different
cameras, the identity of a person on each image is established such that the same person
from different images is correctly matched. This task can be automated by incorporat-
ing computer vision techniques into cameras. Many algorithms have been proposed
which can extract from an image certain visual features that distinguish human from
background, or identities among different persons. Some typical features include color,
texture, edges etc. [1]. Zheng et al. [2] suggested that a robust ReID method should
extract discriminative features among many persons. Recently, convolution neural net-
work (CNN) with deep learning are commonly used techniques for visual feature ex-
traction, which also benefits person ReID. Some CNNs like ResNet have been widely
used to extract features of pedestrians, which are further processed for similarity com-
parison in ReID. Among open datasets for ReID researches, CNN-based methods with
deep learning outperformed traditional methods [2].
However, as more challenging datasets were published for ReID researches, ordi-
nary CNN experienced a bottleneck in performance enhancement. Yao et al. [3] ex-
plained that ordinary CNN in ReID tends to ignore some very fine body parts which
are particularly discriminative. This claim has encouraged studies of part-based feature
extraction. Yao et al. [3] and Zhao et al. [4] clustered a person into a few sub-regions
based on approximated body posture. Features in each of these sub-regions are pro-
cessed by an independent identity classifier and the outputs are fused eventually. Ex-
periment demonstrated a notably higher ReID accuracy of part-based CNNs than ordi-
nary CNNs. Later, several variants of this concept were proposed, which mainly differ
in how a person is partitioned into sub-regions. Liu et al. [5] sampled features in differ-
ent scales which contain fine features around different body parts. Sun et al. [6] uni-
formly divided an image into fix-size parts and proposed a multi-stage training strategy
to refine the approximated parts. These part-based mechanisms require no extra label-
ing of body parts, and can be incorporated into most ordinary CNNs, with significantly
improved ReID accuracy.
Among these part-based and ordinary CNN-based methods, the uniform partition
method by Sun et al. [6] achieved the highest accuracy in most public datasets. Never-
theless, it still showed some misclassified samples. One possible limitation is that they
treated each body part equally important upon final classification stage. Liu et al. [7]
suggested that the importance of different features to distinguishing persons may vary
among different circumstances. Variable illumination, human pose and similar clothing
typically compromise the reliability of some features. Instead, some body parts are pos-
sibly more invariant and discriminative, such that higher reliance can be placed on them
for ReID. Therefore, this paper proposes a method to extract more discriminative hu-
man features. The part-based method of Sun et al. [6] is chosen as our baseline, from
which the importance of each body part is investigated. After this analysis, our pro-
posed ReID method imposes higher weights to the more important body parts, which
shows even higher accuracy than the baseline and other methods.
Monitoring Pedestrian Flow on Campus with Multiple... 1151

2 Proposed Methodology

Our baseline method is Part-based Convolutional Baseline (PCB) [6]. Fig. 1 shows the
proposed PCB-weighted with individual weight imposed on each part loss. Except the
weighting scheme bounded in red, the preceding operation follows the original PCB.
The ordinary ResNet is the backbone network which produces a global feature tensor
ܶ. It is uniformly divided into several sub-regions which roughly represent each body
part of a person. The subsequent average pooling layer extracts features from each part.
The resulting feature map ݃ is squeezed by the 1x1 conv layer to form a reduced feature
map ݄, whose features are more specific to each body part. In training stage, each piece
of feature map goes into a separate fully connected layer for identity prediction. Cross-
entropy loss values are produced by each branch, where the model tries to minimize the
average loss. That is, each body part is equally important in the optimization. In testing
stage, either feature map ݃ or ݄ of each person is exported for similarity comparison
among others. If model training is successful, ݃ and ݄ should contain discriminative
features of a person. Our study does not consider the refined part pooling (RPP) strat-
egy. Readers are referred to the original paper [6] for more detailed explanation.

Fig. 1. Framework of the Proposed PCB-weighted with Weighted Part Loss (revised from [6])

Based on our literature review, equal weighting among each body part is sub-optimal.
Therefore, this paper proposes a strategy of using weighted part losses in model train-
ing. In training stage, the model is optimized against the weighted average loss as
shown in (1), where ‫ݓ‬௜ and ݈௜ are respectively the weight and loss of the ݅ ௧௛ part, ‫ ܮ‬is
the weighted average loss among all ‫ ݌‬parts divided in one image. This paper adopts 6-
parts division (‫=݌‬6) which was found optimal in [6].
೛
σ೔సభሺ௪೔ ‫כ‬௟೔ ሻ
‫ ܮ‬ൌ ೛
σ೔సభ ௪೔
(1)

The contribution of features in each part to the final ReID accuracy is investigated.
Firstly, a PCB model was trained with the configuration shown in Table 1. Only 1-stage
training was involved, without the RPP strategy. Horizontal flip and normalization of
images were replaced by random erasing [8]. The trained model learned to produce 6-
parts feature maps ݃ and ݄, as in Fig. 1. Subsequently, each piece of ݄ was isolated as
final feature representation of each person in testing stage. The ReID accuracy with
each piece was then analyzed one by one to compare the usefulness of each individual
body part. Features of ݄ was analyzed because they are more part-specific, such that
the effect of isolating each part is more obvious. DukeMTMC-reID dataset [9] obtained
1152 P. K.-Y. Wong and J. C. P. Cheng

on a college campus was used for training and testing. Rank-1 accuracy and mean av-
erage precision (mAP) are the evaluation metrics.

Table 1. Configuration of training a PCB model in our study

Item Setting
Backbone feature extractor ResNet50 pre-trained from ImageNet
Training dataset DukeMTMC-reID [9]
Training epochs 60
Batch size 32
Initial learning rate 0.0003
Learning rate decay 0.1 (at 40th epoch)
Loss function Cross-entropy loss
Data augmentation Random erasing [8]
With RPP strategy [6] No

Table 2 summarizes the performance in the testing dataset by utilizing each body part
alone. Parts 1-6 of an image are uniformly divided from top to down. It is shown that
features of parts 2, 3, 5 resulted in the highest Rank-1 accuracy or mAP, while parts 1
and 6 alone led to worst performance. It is possibly because the RPP strategy was not
utilized, images may not be precisely aligned or cropped along the boundary of human
body. These may make parts 1 and 6 mostly cover noisy information like background
rather than the actual human head or shoes. Furthermore, the resolution of images was
not very high, such that the small faces and shoes are blurry which provided limited
cues for ReID. Instead, parts 2, 3, 5 capture richer features of a human body. Therefore,
with the aforementioned conditions, losses of parts 2, 3, 5 should have larger weights
than other parts in model training.

Table 2. Testing performances in DukeMTMC-reID by selecting each part from feature map ݄

Part(s) Rank-1 accuracy (%) mAP (%)

1 52.7 35.8
2 62.9 44.1
3 63.6 43.6
4 62.1 42.8
5 62.9 44.1
6 49.8 32.7

3 Result and Discussion

Among the 6 part losses (݈௜ , ݅=1, 2, 3, 4, 5, 6), higher weights were assigned to parts 2,
3, 5, where ‫ݓ‬ଶ ൌ ‫ݓ‬ଷ ൌ ‫ݓ‬ହ ൌ ʹǤͲ and ‫ݓ‬ଵ ൌ ‫ݓ‬ସ ൌ ‫ ଺ݓ‬ൌ ͳǤͲ. This paper adopts two
Monitoring Pedestrian Flow on Campus with Multiple... 1153

discrete values to easily visualize the advantage of weighting some parts differently.
Optimizing the weights of each part would be a potential future work. With the weights
imposed on the part losses, a PCB-weighted model was re-trained, according to the
configuration in Table 1. The RPP strategy was not adopted. In testing stage, the feature
map ݃ was exported to represent each person.
Table 3 compares the performance in DukeMTMC-reID testing dataset of our pro-
posed PCB-weighted method, against the baseline PCB and other recently published
methods. It is shown that our imposing higher weights on particular body parts results
in considerable improvement in ReID accuracy. Our PCB-weighted method notably
outperforms some recently published methods, including the baseline PCB by 2.3% in
Rank-1 accuracy and 5.5% in mAP. This supports our claim that some body parts pro-
vide more discriminative features. Placing higher reliance on these specific parts would
benefit person ReID. Fig. 2 demonstrates the improved performance of our PCB-
weighted against the original PCB. Despite tKHZRPDQµ¶ORRNHGTXLWHGLIIHUHQWin
another camera due to the red scarf cap, our PCB-weighted relied more on other body
parts like the black cotton socks around her knees (part 5), which were more view-
invariant. Instead, the original PCB IRXQGDQRWKHUPDQµ¶ possibly due to the sim-
ilarity of overall clothing color, rather than focusing on specific parts that were more
discriminative.
On the contrary, Fig. 3 shows some failure results of our PCB-weighted. It mis-
PDWFKHGµ¶ZLWKµ¶SRVVLEO\EHFDXVHRIWKHLGHQWLFDOZKLWHFDUDWSDUWZKLFK
occluded the person. Relying too much on this part may then mislead our model. In-
stead, the original PCB which considers each part equally important may better handle
such condition with incomplete features of a person. In other words, our PCB-weighted
method does not filter out the background clutters before feature extraction. Prepro-
cessing the images to isolate the targeted person would be a potential future work for
additional robustness in person ReID.

Table 3. Testing performances in DukeMTMC-reID of our proposed and other methods

Method Rank-1 accuracy (%) mAP (%)

MLFN [10] 81.2 62.8
DuATM [11] 81.8 64.6
PCB (no RPP) [6] 81.7 66.1
PCB-weighted (ours) 84.0 71.6

Fig. 2. Improved results of our PCB-weighted (left) against original PCB (right). Q: query iden-
tity, G1: the most similar identity in the ranked gallery.
1154 P. K.-Y. Wong and J. C. P. Cheng

Fig. 3. Degraded results of our PCB-weighted (left) against original PCB (right). Q: query iden-
tity, G1: the most similar identity in the ranked gallery.

4 Conclusion and Future Work

Our proposed method, PCB-weighted, can extract more discriminative human features
for person ReID across multiple cameras. By selectively focusing more on specific
body parts that are more discriminative, our PCB-weighted notably outperforms some
state-of-the-art ReID methods, including the baseline PCB which considers the body
parts equally important. For further improvement, a more optimized weighting among
body parts could be studied. Image preprocessing for filtering out background clutter
could also enhance the robustness of ReID upon occluded human features.

References
1. A. Bedagkar-Gala, S.K. Shah, A survey of approaches and trends in person re-identification,
Image Vision Comput. 32 (2014) 270-286.
2. Liang Zheng, Yi Yang, Alexander G. Hauptmann, Person Re-identification: Past, Present
and Future, (2016).
3. H. Yao, S. Zhang, Y. Zhang, J. Li, Q. Tian, Deep Representation Learning with Part Loss
for Person Re-Identification, arXiv preprint arXiv:1707.00798. (2017).
4. L. Zhao, X. Li, Y. Zhuang, J. Wang, Deeply-learned part-aligned representations for person
re-identification, (2017) 3219-3228.
5. X. Liu, H. Zhao, M. Tian, L. Sheng, J. Shao, S. Yi, J. Yan, X. Wang, Hydraplus-net: Atten-
tive deep features for pedestrian analysis, (2017) 350-359.
6. Y. Sun, L. Zheng, Y. Yang, Q. Tian, S. Wang, Beyond part models: Person retrieval with
refined part pooling (and a strong convolutional baseline), (2018) 480-496.
7. C. Liu, S. Gong, C.C. Loy, X. Lin, Person re-identification: What features are important?
(2012) 391-401.
8. Z. Zhong, L. Zheng, G. Kang, S. Li, Y. Yang, Random erasing data augmentation, arXiv
preprint arXiv:1708.04896. (2017).
9. E. Ristani, F. Solera, R. Zou, R. Cucchiara, C. Tomasi, Performance measures and a data set
for multi-target, multi-camera tracking, (2016) 17-35.
10. X. Chang, T.M. Hospedales, T. Xiang, Multi-level factorisation net for person re-identifica-
tion, (2018) 2109-2118.
11. J. Si, H. Zhang, C. Li, J. Kuen, X. Kong, A.C. Kot, G. Wang, Dual attention matching net-
work for context-aware feature sequence based person re-identification, (2018) 5363-5372.
 An Image Mosaicking Method Base on The Curvature of
Cost Curve for Tunnel Lining Inspection

Cuong Nguyen Kim1, Kei Kawamura2, Masando Shiozaki3, and Amir Tarighat4
1
Mientrung University of Civil Engineering, Tuyhoa, Phuyen, Vietnam
nguyenkimcuong@muce.edu.vn
2
Graduate School of Sciences and Technology for Innovation of Yamaguchi University, Ube,
Yamaguchi, Japan
Email: kay@yamaguchi-u.ac.
3
Sumitomo Mitsui Construction Co., Ltd., Tokyo, Japan
MasandoShiozaki@smcon.co.jp
4
Dept.of Civil Engineering, Shahid Rajaee Teacher Training University, Iran
tarighat@srttu.edu

Abstract. The paper proposes an image mosaicking method for tunnel lining
inspection. The conventional methods only used the cost value of the pixel
being processed based on similarity metric to estimate an image-matching
location. To improve the image-mosaicking efficiency, the curvature metric has
adapted to improve matching point accuracy. Experimental results for an actual
tunnel demonstrate that the curvature measurement can select the precise
matching points accurately for assisting defect inspection without depending on
the speed of inspection vehicle.

Keywords: Image mosaicking, Tunnel lining inspection, Curvature metric.

1 Introduction

The conventional visual inspection usually is affected by human error, subjectively.

Therefore, automatic crack detection for concrete tunnel lining also tempts many
researchers. For illustration, Zhang et al. 2014 [1] proposed the crack detection
system. The original tunnel images are collected by CMOS line scan cameras. These
color images are transferred into gray-scale images for further image processing.
Subsequently machine learning techniques are applied to classify and compare the
accuracy of different classifiers. Shen et al., 2015[2] proposed a subway tunnel crack
detection method based on wireless multimedia sensor network. This system includes
four components: a vehicular wireless multimedia sensor node, a station sink node, a
cable transmission access module, and a central server. This system obtains
significantly meaningful results with two noise-removal steps for crack detection, but
it needs a very complex equipment and even more expensive.
Ukai and Nagamine [3] used a tunnel scanner with line sensor cameras, which does
not require a special vehicle, to continuously scan images for detecting tunnel wall

© Springer Nature Singapore Pte Ltd. 2020 1155

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_185
1156 C. N. Kim et al.

deformations. Besides, image-collection speed is about 20 Km/h relatively slow to be

able to affect conventional traffic flow. This image scanning system is used on
railways, but it has not been applied to road tunnels. Besides, the image matching
algorithm is not presented in detail.
This study utilizes the tunnel scanning approach to provide the layout panorama for
defect inspection on the tunnel lining surface. The goals are to achieve reasonable
accuracy, active image-acquisition time, and manageable inspection cost. For image
matching, the authors use color pixel-intensity-based feature and the curvature
measure to find the appropriate matching points. Moreover, in the curvature metric,
median filter is used for refinement to smooth image motion quantity (IMQ) in the
consecutive image stitching process. This is different from standard image matching
techniques that make use of key-point (feature)-based image matching methods such
as SIFT, SURF [4] ), which may fail when the tunnel lining images contain noises or
do not have many discriminated features.
The contribution of the inspection system in this research is to create the extended
panoramic images in both longitudinal and circumferential directions to assist
inspectors who have a wider field of view or an enhanced visualization of large area
the tunnel lining.

2 Proposed method

2.1 Image acquisition system

Fig.1 shows the image acquisition system consisting of six digital video cameras
(from V1 to V6), three led illuminators attached to a steel framework shaped
compatible within a half of the cross-section of the tunnel, and a car. This device is
capable of sliding from the side to the top of the car so that the full tunnel lining
surface image can be captured by several passes through the tunnel [5].
Quality of the collected data depends on many factors, such as distance between the
tunnel wall and the inspection car, the illumination, and the specification of the video
camera. This system is simple and does not require an exclusive vehicle. The video
data are converted into the consecutive images with the specific overlapped region
depending on the scanning speed of the car. By design, the resulting image motion is
only the translation, thus the rotation is not a concern. Therefore, the direct method,
which makes use of color-pixel intensity in the image-matching location, is applied to
stitch images.

2.2 Image-stitching procedure

The image-stitching procedure for the entire tunnel lining involves three steps shown
in Fig.2. Captured image data are first retrieved from the six video cameras, each of
which contains a number of images depending on the length of the tunnel.
An Image Mosaicking Method Base on The Curvature... 1157

Video camera Steel frame

Illuminator LED light
SONY HDR-CX630V Sliding from side to top

Image acquisition system Inspection Car

camera angle of view between 5o- 6o

Fig. 1. Image acquisition system

In Step 1, the consecutive images for each camera are stitched into a panorama in the
longitudinal direction of tunnel for each camera. In Step 2, these panoramic images
obtained in Step 1 are stitched in the circumferential direction of tunnel for each
region corresponding to each pass of the inspection car. Subsequently, in Step 3, the
connected images of the six regions (R1-R6) in step 2 are stitched together to make a
layout panorama, which consists of a full view of the entire tunnel lining surface.
The key to image stitching is to find the correct matching point for a pair of
consecutive images. To this end, the authors propose a new image-matching
algorithm, as follows: First, an initial image-matching location (IML) is estimated
using Sum Absolute Difference (SAD). Moreover, in this paper, a novel matching
method using Curvature of the cost curve at the Nearest-neighbor Pixels (CNP) is
used to match a consecutive image pair.

2.3 Image-matching procedure

Figure 3 indicates the automatic matching process of two consecutive images in the
tunnel longitudinal direction (X-axis). Here, image 1 is the referenced image, and
image 2 is the registered image of the image-matching process. A search area is set in
advance. Each movement step of the search point is with respect to the location,
which enables image 2 to be shifted onto image 1 to find the appropriate image-
matching location by measuring the similarity and sharpness in the brightness of all
pixel pairs in the overlapped region of the two images.
Furthermore, to accelerate the search and measurement process, the search point
and similarity measurement are skipped in the search area and overlapped region with
predefined values, respectively.
1158 C. N. Kim et al.

a consecutive frame pair stitching

Camera

v06 䞉䞉 䞉䞉䞉 䞉䞉
䞉䞉䞉䞉

䞉䞉䞉䞉
䡚

v01 䞉䞉 䞉䞉䞉 䞉䞉

Camera panoramic stitching for each region

v06
䡚

v01

Fig. 2. The image stitching procedure for the entire tunnel Fig. 3. Searching process of
Image matching location

M N

¦¦¦
i m j n
{ R ,G , B }
| ( I 1  I 2 ) |( i , j )
SAD
( N  n  1) u ( M  m  1) (1)

In these equations, I1 and I2 are the intensity of pixels at coordinates (i, j) of images
1 and 2 in the overlapped region, respectively; SAD is the sum of difference between
the pixel values of images 1 and 2 for each color channel (R, G, and B). (m, n) and
(M, N) are as the lower left and upper right coordinates (pixel) of the overlapped
region of the image pairs, respectively. In this function, the higher similarity of the
overlapped region of two images yields the smaller score of SAD.
The traditional similarity metrics are affected by artifacts such as non-uniformed
brightness, periodic structures, featureless, and noises. These factors result in the
image-matching error. Therefore, in some cases, the global minimum cost value of the
VLPLODULW\ PHWULF GRHVQ¶W \LHOG ZHOO FRUUHVSRQGing to the ground truth (G-T image
mosaicking). Otherwise, the local minimum cost is coincident with the G-T image
mosaicking. The authors consider to the local minimum cost.

2.4 Curvature metric with full search method

The conventional similarity metrics only exploited the information of the matching
point of the pixel being processed without considering the information of neighbor
pixels. Therefore, the results of similarity metrics frequently cause to many errors.
The author proposes a direct curvature metric method based on density gradient of the
neighbor pixels on full search area as the following equations:

§0 1 0· § C(k  E ,l) ·
m ¨ ¸ ¨ ¸ (2)
CUR(k,l) ¨ 1 m 1 ¸ u ¨ C(k,l D ) C(k,l) C(k,l D ) ¸
¨ 0 1 0 ¸ ¨¨ ¸¸
© ¹ © C(k  E ,l) ¹
An Image Mosaicking Method Base on The Curvature... 1159

C(k,l): Cost value of the centre pixel using cost functions ; (DE) is the skipped pixel
number or a size of search window. CURm(k,l), is the curvature metric value ; m refers
to the 4-connected and 8-connected neighborhood pixels at the (k,l) coordinates,
respectively. The larger CUR value means higher matching possibility.

2.5 Case study

The tunnel inspected in this experiment is an actual single-core circular tube with the
length of 230 m, the width of 10.25m, and the height of 4.7m in Yamaguchi
Prefecture, Japan. The video digital cameras are SONY HDR-CX630V, with LED
light 600lx and camera angle of view between 5 0- 60. The distance between the
cameras and the tunnel wall is 3m. The dataset acquired by each camera consists of
1,558 images with a resolution of 1920 × 1080 pixels for each picture. It is useful to
use a photographic laser distance meter to maintain the constant distance between the
tunnel lining and the inspection car to ensure the constant resolution. The speed of the
inspection car is maintained at 30± 5 km/h corresponding to the desired image motion
quantity from 500 to 700 pixels.
Further, a prototype software written in C++ was developed to implement the full
image stitching algorithm.

3 Performance evaluation

3.1 Evaluation of image matching accuracy based on the parameters of the

video acquisition system
The speed of the inspection car is set at about 30 km/h, with possible variation in the
range 25 to 35 km/h with respect to the reliable range of image motion quantity of
[500 700] pixel. The error rate is computed from the following equation (3):
N
1
AME
N
¦ IME u100%
i 1
i

­0 if IMQi  [500 700]

IMEi ®
¯1 Otherwise (3)
where N is a number of images in each camera, IMEi is ith image matching error.
IMQi is image motion quantity at ith measurement; and AME is the accumulative error
rate for each camera. This equation indicates that an IMQ value is out of range [500
700] pixels considered as an IME.
The Table 1 shows comparison results of the proposed curvature metrics for six
cameras. As can be seen, all of CUR and CNP denoted the average curvature of the
cost curve have error ratio very small. Comparing among the results of CUR metrics,
CUR4 is the highest accuracy with the average accuracy of six cameras of 1.95%.
Otherwise, CNP8 defined the average curvature metric of the 8-connected nearest
neighbor curvature has the best results of the error ratio of 1.35%. It is demonstrated
1160 C. N. Kim et al.

that the higher connected nearest neighbor has the more confident matching point
accuracy obtains.
Table 1. Error rate of curvature metric

Camera CUR4 CNP4 CUR8 CNP8

V1 1.09 0.90 0.96 0.83

V2 0.45 0.39 0.83 0.39

V3 0.71 0.64 2.12 0.90

V4 0.90 0.90 1.16 0.96

V5 0.45 0.45 0.77 0.51

V6 8.10 7.52 19.78 4.50

Average 1.95 1.80 4.27 1.35

4 Conclusions

The SAD metric considered the global minimum value of the similarity metric so that
the results of the matching points had many errors due to the influence of the noises.
The CUR metric focused on the candidates which had the local maximum values of
the cost curve. So far, the CNP metric tested the stability of the candidates based on
the mean curvature value.
The main advantage of the average curvature metric was to eliminate the artifacts
of the neighbor pixels surrounding the candidate matching point before measuring the
sharpness at the peaks of the cost curve. Therefore it improved the computational time
and the matching accuracy compared to our previous studies.

References

1. Zhang,W., Zhang, Z., Qi, D., Liu,Y.: Automatic crack detection and classification
method for subway tunnel safety monitoring, Sensors, Vol.14, pp.19307-19328, 2014.
2. Shen, B., Zhang, W., Y., Qi, D., P., and Wu, X., Y: Wireless Multimedia Sensor
Network Based Subway Tunnel Crack Detection Method, Volume 2015, pp.184639-
184649, 2015.
3. Ukai, M. and Nagamine, N. : A high-performance inspection system of tunnel wall
deformation using continuous scan image, 9th World Congress Railway Research,
2011.
4. Byrne, J., Laefer ' ) DQG 2¶.HHIIH (: Maximum feature detection in aerial
unmanned aerial vehicle datasets, Journal of Applied Remote Sensing, Vol. 11, 2017.
5. Kawamura, K., Koga, M., Matsumoto, J., Shiozaki, M., and Sawamura, S. : A Study
On Panoramic Images Generation Of Tunnel Wall, Japan Society of Civil Engineers,
Vol(71), No.2, pp.142-151, 2015.
Application of Neural Network to predict the workability
Parameters of Self-Compacting Concrete

Cuong H. NGUYEN1, Linh H. TRAN2

and Khoa N. HO3

1
National University of Civil Engineering, Hanoi, Vietnam
2
Hanoi University of Science and Technology, Hanoi, Vietnam
3
National University of Civil Engineering, Hanoi, Vietnam

cuongnguyen.dhxdhn@gmail.com,linh.tranhoai@hust.edu.vn,
khoahn@nuce.edu.vn

Abstract. The paper presents a new application of classical multi-layer percep-

tron neural network to approximate the parameters of fresh self-compacting
concrete. The approximation is needed to determine the working parameters at
construction site and can be used to estimate the components used at mixing
station to achieve the desired concrete quality. A number of real field tests were
conducted and six basic parameters were measured for each test. The numerical
results showed the high accuracy of proposed solution.

Keywords: Self-compacting Concrete; Nonlinear Approximation; Feed for-

ward Neural Network; Multi-layer Perceptron.

1 Introduction

Self-compacting concrete (SCC) is a fresh concrete mixture with high flexibility, self-
flowing under the weight to completely fill all the edges of the form without vibra-
tion. 6&&¶VSHUIRUPDQFH is influenced by climatic conditions. Workability decreases
with time, especially in high temperature and low humidity conditions, reducing SCC
advantages. Decrease in workability not only affects the ease of construction but also
affects the quality and strength of SCC [1].
Vietnam has a tropical monsoon climate, with adverse impacts on the workability
of SCC. Therefore, the research and development of a model predicting the variation
of the parameters of SCC for Vietnam is urgent. For estimating model, the setting at
the mixing station will be proposed on site at the time of transport, temperature and
humidity. At the same time, through this model, it is proposed that with this workload
requirement and the initial intensity, the material components must be mixed with the
proportions. In this paper, we present a method using artificial neural networks
(ANN) to predict the parameters of the SCC depending on the environment tempera-
ture, the concrete temperature and the retention time.

© Springer Nature Singapore Pte Ltd. 2020 1161

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_186
1162 C. H. Nguyen et al.

2 The workability parameters of Self-Compacting Concrete

The SCC parameters are defined in EN 12350 [2], among which the most popular 6
parameters considered in this paper are: slump-flow test (SF), T500, V-funel, L-box,
J-ring and SR. To evaluate the concrete mixture storage time¶V HIIHFW to SCC com-
pressive strength (the R28 parameter), concrete mixture samples were cast every 30
minutes to form concrete samples, then maintained under standard conditions and
tested with a spindle compressor after 28 days. The 6&& PL[WXUH¶V workability pa-
rameters decrease with retention time. To evaluate this, we conducted various tests in
a free fall mixer. After mixing, we measured the parameters to get initial values. Lat-
er, the mixture was rotated in a slow mixer tank, sampled every 30 minutes to get the
same set parameters. The testing temperature condition was varied from 11oC to 43oC,
humidity varied in 30-85%. Totally, 35 tests were conducted.
The examples of results of the experiments are shown in the Table 1 below.

Table 1. Examples of measured parameters of SCC

Sampling condition Sampling time (minutes)

Parameters
Tenv TSCC RH% 0 30 60 90 120
SF (mm) 650 680 670 640 635
T500 (s) 3.1 3.66 3.7 4.72 4.73
Vfunel (s) 11.72 10.78 11.0 12.00 13.0
23.4 oC 25.6 oC 81%
Lbox 0.89 0.90 0.87 0.86 0.85
Jring (mm) 9.0 7.0 9.1 10 11
R28 (MPa) 58.6 57.1 53.2 52.1 46.4
SF (mm) 645 650 650 640 580
T500 (s) 3.68 3.45 3.9 5.2 6.9
Vfunel (s) 11.4 10.5 11.2 13.3 19.4
31 oC 31.3 oC 75%
Lbox 0.94 0.93 0.9 0.89 0.86
Jring (mm) 8.0 6.0 8.0 12 16
R28 (MPa) 61.0 62.3 60.0 53.0 46.0
SF (mm) 650 640 570 470 400
T500 (s) 3.4 2.75 4.8 6.7 7.4
Vfunel (s) 9.2 11.9 17.1 34.2 40.3
43.1 oC 40.5 oC 43%
Lbox 0.88 0.87 0.67 0.5 0.3
Jring (mm) 7 9.7 13 15 20
R28 (MPa) 58.0 63.0 55.0 47.0 40

As it can be deducted from the measurement results, the parameters of SCC depend
nonlinearly on the resting time, the environment temperature and the concrete tem-
perature itself [3,4]. The need of a mathematical model predicting the parameters
from the input conditions is very high because thanks to the generalization ability of
Application of Neural Network to predict the workability... 1163

the model, we can predict the parameters for the input conditions, which were not
presented in the learning data samples [5]. In this paper, we proposed to use the clas-
sical multi-layer perceptron (MLP) as the non-linear estimator.

3 The MLP and its application in nonlinear modeling

The MLP (MultiLayer Perceptron) is a feed forward structure with cascaded layers
[6]. The network with 1 input layer, 1 output layer and 1 hidden layer is the most
popular in applications. In Fig. 1 an example of MLP network with only 1 hidden
layer is presented. This network is used widely in nonlinear mapping problem [6].
When a structure with a given number of hidden neurons is trained to fit a learning
data samples set, the weights connecting neurons are adapted. If we denoted the set
containing p pairs of input-output ^xi , di ` where i 1,..., p; xi  £ N ; di  £ K then
the parameters of MLP are adapted to minimize the error function:

1 p 2
E ¦ MLP  xi   di
2i 1
o min (1)

Fig. 1. An example of MLP with one hidden layer

We used the classical Levenberg ± Marquadrt algorithm for training the MLPs [6,
7]. The number of hidden neurons is selected with the trial and error method. We
started from a network with only 1 hidden neuron) and generated randomly 50 net-
works with that number of hidden neurons. All the networks were trained with the
data sets, the network with the lowest error will be selected. If the best error was still
high, the number of hidden neurons is increased by 1. In order to avoid overlearning
effect [6], we selected the network with lowest possible number of hidden neurons
(i.e. the simplest possible network), which can still approximate the data. Simulations
showed that we need about 5 hidden neurons to achieve satisfactory results. The train-
ing process was performed using programs with 0DWODE¶V Neural Network Toolbox.
1164 C. H. Nguyen et al.

4 The numerical results

For 6 parameters, one MLP network was created for each parameter. MLP networks
were trained with following conditions. Number of networks was 6, each had 3 inputs
(x1 ± environment temperature, x2 ± concrete temperature, x3 ± time), 1 output (one of
the six parameters), 1 hidden layer with the number of neurons tried from 1 to 5.
Number of data samples: 35 pair of data for each case, which makes total data sam-
ples of 6˜ 35 210  data samples.
The achieved results for each parameter are listed below. In Fig. 2 to 4 are the re-
sults for the SF, T500, V-funel, L-box, J-ring and R28 parameters respectively.

Fig. 2. The approximation results for SF and T500 parameters

Fig. 3. The approximation results for V-funel and Lbox parameters

Fig. 4. The approximation results for J-ring and R28 parameters

Application of Neural Network to predict the workability... 1165

As we can see in Fig. 2, 3, 4 above, MLP models can approximate accurately

the SCC parameters with very small difference between the expected d (the yellow
brown line) and the estimated y network (the light blue line). Figure 5 presents the
correlations between the actual output from MLP and the desired destination values
for Lbox and R28 parameters. It can be seen that the achieved results have very high
correlation.

Fig. 5. Examples of correlation between MLP outputs and the destination values for Lbox and
R28 parameters

The performance was further assessed with following measures: MAE (Mean Ab-
solute Error), MRE (Mean Relative Error), Max AE (Max Absolute Error) and the
correlation coefficient between the outputs from MLP and the actual destination val-
ues. The numerical errors are collected in the Table 2.

Table 2. Approximation errors for the models

Parameters MAE MRE (%) MaxAE Correlation
SF 2.12 0.33 11.30 0.98
T500 0.075 1.76 0.42 0.99
Vfunel 0.13 1.08 0.73 0.99
Lbox 0.003 0.33 0.021 0.98
Jring 0.26 2.70 0.93 0.99
R28 0.49 0.91 1.68 0.98

The training was done with 35 data sets for each parameter for the SCC. Results in
Table 2 showed that MAE, MRE (%) and Max AE have low value and correlation
coefficient is close to 1. This proves that MLP models with single hidden layer have
good performance, thus are appropriate for SCC parameters determination. This is
similar with results of other researches [9, 10]. On the other hand, in order to avoid
the over-fitting effect when using neural networks for approximation, the approach to
use the simplest possible network to perform the learning task was used. This ap-
proach along with other such as earlier stopping, network regulating... is well-known
[6, 9] and was selected due the limited number of samples we were able to experi-
mentally create. Our tests showed that networks with 4 hidden neurons could not
perform the training task well (errors were unacceptably bigger) and networks with 5
1166 C. H. Nguyen et al.

hidden neurons started to be able to learn well. And this number of hidden neurons
was selected for further application. Higher number of neurons did not help achieve
better learning results but could potentially cause the over-fitting effects [6].
In comparing with [8], it used ANN Feed-Forward Back Propagation including two
hidden layers with 6 neurons in the 1st layer and 7 neurons in the 2nd one to predict the
slump of concrete (8 inputs) have high accuracy rate (coefficient R2 of 0.99848). Our
results with 1 hidden layer of 5 neurons has almost similar accuracy rate but it is im-
portant to note that the two hidden layers test are much more complicated compared
with single hidden layer one.

5 Conclusions

An application of the MLP to estimate the workability parameters of the self-

compacting concrete depending on the environment temperature, the concrete tem-
perature and the humidity was proposed and tested with practical real data. The nu-
merical results showed that a high accuracy can be achieved for predicting 6 parame-
ters of the SCC based on input factors as environment temperature, the concrete tem-
perature and the retention time.
The method can be extended for estimation with different concrete components
and also can form the reverse problem, in which we can find the input mixing compo-
nents to achieve a given working parameter of SCC. From these results, the SCC
mixing units can make plan on their SCC delivery to construction sites to ensure
quality of a new construction project.

References
1. Powers, T. C.: The properties of fresh concrete (1968).
2. Standard, B. BS EN 12350 Series, BSI Standard Publications: London, UK (2000).
3. Sobhani, J., Najimi, M., Pourkhorshidi, A. R., Parhizkar, T.: Prediction of the compressive
strength of no-slump concrete: A comparative study of regression, neural network and
ANFIS models, Construction and Building Materials, 24 (5): 709-718 (2010).
4. Tattersall, G. H.: Workability and quality control of concrete, CRC Press, (2014).
5. Yeh, I., Chen, J.-W.: Modeling Workability of Concrete Using Design Of Experiments
And Artificial Neural Networks, 20 (2): 153-162 (2005).
6. Haykin, S.: Neural Networks and Learning Machines, 3 rd ed,. Pearson Prentice Hall,
(2009)
7. Tran, L. H.: Neural Networks and Their Applications in Signal Processing, Hanoi
University of Science and Technology Publisher, Viet Nam, (2014).
8. Agrawal, V., Sharma, A.: Prediction of slump in concrete using artificial neural networks,
World Academy of Science, Engineering and Technology, 45: 25-32 (2010).
9. Zin, R. M., Majid, M. Z. A., Putra, C. W. F. C. W., Mohammed, A. H.: Neural Network
model for design constructability assessment, Jurnal Teknologi, 40 (1): 27-40 (2004).
10. Ogunlana, S., Bhokha, S.: Application of Artificial Neural Network (ANN) to Forecast
Construction Cost of Buildings at the Pre-design Stage, Engineering, Construction and
Architectural Management, 6 (2): 133-144 (1999).
Development of Artificial Neural Networks for Prediction
of Compression Coefficient of Soft Soil

Binh Thai Pham1*, Manh Duc Nguyen2, Hai-Bang Ly1*, Tuan Anh Pham1, Vu Ho-
ang1, Hiep Van Le1, Tien-Thinh Le3*, Hung Quang Nguyen4; and Gia Linh Bui1
1 University of Transport Technology, 54 Trieu Khuc, Thanh Xuan, Hanoi, 100000, Vietnam
2
University of Transport and Communications, 3 Cau Giay, Dong Da, Hanoi, Vietnam
3
NTT Hi-Tech Institute, Nguyen Tat Thanh University, Ho Chi Minh City, 700000, Vietnam
4
Thai Nguyen Transportation Construction Consultant Joint Stock Company, Thai Nguyen
City, 24000, Vietnam
* Corresponding authors: Binh Thai Pham (binhpt@utt.edu.vn);
Hai-Bang Ly (banglh@utt.edu.vn); Tien-Thinh Le
(tienthinhle.vn@gmail.com)

Abstract. Compression coefficient (Cc) of soft soil is an important parameter in

solving many geotechnical problems. In this study, the main of objective is to
develop an Artificial Neural Networks (ANN) for prediction of the Cc of soft
soil. A total of 189 soft soil samples collected from the Ninh Binh ± Hai Phong
national highway project were used to carry out the laboratory tests for deter-
mining the parameters for modelling of which thirteen factors (depth of sample,
clay content, moisture content, bulk density, dry density, specific gravity, void
ratio, porosity, degree of saturation, liquid limit, plastic limit, plasticity index,
liquidity index) were considered as input variables and the Cc was considered
as a output variable for prediction models. This data was divided into two parts
of training (70%) and testing (30%) datasets for building and validating the
models, respectively. To validate the performance of the ANN, various methods
named Mean absolute error (MAE), root mean square error (RMSE), squared
correlation coefficient (R2) were used. The results show that the ANN are prom-
ising method for prediction of the Cc of soft soil. This study might help ge-
otechnical engineers to reduce the cost of implement of laboratory tests and the
time for construction.

Keywords: Artificial Intelligence; Artificial Neural Networks, Compression

Coefficient, Soft Soil, Vietnam.

1 Introduction

Compression coefficient of soft soil (Cc) is an important parameter in solving many

geotechnical problems [1]. Basically, the Cc is obtained from the e-ORJ௘SFXUYHRIWKH
consolidation test whereas e is void ratio and p is consolidation pressure [2]. Howev-
er, this experiment might take long time and expensive cost. Therefore, many re-
searches have been carried out to build relationship between the Cc and other simple

© Springer Nature Singapore Pte Ltd. 2020 1167

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_187
1168 B. T. Pham et al.

physical properties of soil. Koppula [3] built the empirical relationships between the
Cc and liquid limit or void ratio. Aplin et al. [4] studied the relationship of the Cc of
mudstones with detailed lithology. Park and Koumoto [5] also constructed the rela-
tionship between the Cc and porosity of remolded clay. In general, these studies have
provided the empirical equations for determination of the Cc. However, they consid-
ered only several factors (one or two) in relationship with the Cc. With many factors
(larger than three), this approach is more complex and might not be applicable.
In recent years, Artificial Intelligence (AI) approaches, which can deal with diffi-
cult problems with complexity of data, have been discovered potentially in solving
many real-world problems including the prediction of properties of soil. Shahin et al.
[6] proved the potential of the AI approaches in geotechnical engineering. Shahin et
al. [7] studied and showed the recent advances and future challenges for applications
of the AI in geotechnical engineering. In general, these AI studies show that the AI
approaches like ANN is a great tool in prediction of the soil properties. Therefore, the
main objective of this study is to develop the ANN to predict the Cc of soft soil. A
total of 189 samples were collected from the construction site of Ninh Binh ± Hai
Phong coastal national road project, and used to determine the data for modeling.
Various methods named Mean absolute error (MAE), root mean square error
(RMSE), squared correlation coefficient (R2) were used to validate the predictive
capability of the ANN.

2 Methods used

2.1 Artificial Neural Networks

Artificial Neural Networks (ANN) is one of the most popular AI techniques, which is
based on the biological neural systems of human brain [8]. It is well-known as a back
model which is more powerful in the case of unknow relationship among variables
[9]. Main feature of the ANN is that it has an independent statistical distribution with
self-learning and interdependent memory. In general, the ANN consists of a number
of neurons (nodes) which construct its structure with three layers including input lay-
ers, hidden layers and output layers [10]. In this study, the sigmoid activation func-
tion, which is a common ANN function [11], was used in one hidden layer to connect
the input variables with output parameter. In addition, we used neural network
toolbox in Matlab with backpropagation (BPN) algorithm based on Levenberg-
Marquardt [8].

2.2 Mean absolute error, root mean square error, and squared correlation
coefficient
Mean absolute error (MAE), root mean square error (RMSE), squared correlation
coefficient (R2), which are popular methods for validation of the statistical models,
were selected to validate the performance of the ANN in predicting the Cc in the pre-
sent study. MAE calculates the difference between the actual and predicted values
whereas RMSE calculates the average square root of squared difference between the
Development of Artificial Neural Networks... 1169

predicted and actual values [12]. R2 evaluates the correlation between predicted and
actual values [13]. In general, high value of R2 show better accuracy of the ANN and
verse vice. In contrast, low RMSE and MAE values show better accuracy of the
ANN.

3 Data used

In this study, a total of 189 samples were collected from the construction site of Ninh
Binh ± Hai Phong coastal national road project. These samples were then used to
determine the input and output parameters for the modeling. More specifically, input
parameters include depth of sample (m), clay content (%), moisture content (%), bulk
density (g/cm3), dry density (g/cm3), specific gravity, void ratio, porosity (%), degree
of saturation (%), liquid limit (%), plastic limit (%), plasticity index (%), liquidity
index whereas output parameter is the Cc (cm2/s x 10^-3). Detail description of these
parameters is present in Das and Sobhan [2]. In modeling, data of 189 samples was
divided into two parts. One part includes 70% of data (132 samples) which was then
used to generate the training dataset for building the model whereas another part con-
sists of 30% remaining data (57 samples) which was then used to create the testing
dataset for validating the model.

4 Results and discussion

Validation of the model has been achieved as shown in Fig. 1 and Fig. 2.

Fig. 1. Comparison of compression strength of soft soil in case of: (a) predicted ANN vs. ex-
perimental values, (b) error mean and (c) standard deviation of the training part
1170 B. T. Pham et al.

It can be clearly observed that the experimental data and the predicted values obtained
from the training data of ANN algorithms is in excellence agreement with the experi-
mentally collected one (Fig. 1a). The corresponding errors between target and pre-
dicted values are plotted in Fig 1b, along with the histogram of error (Fig. 1c). For the
training part, the values of errors were calculated as: MAE=6.61E-5, RMSE=0.0257,
mean error=0.0023 and standard deviation St.D=0.0257. The corresponding testing
data was also highlighted in Fig. 2. Similar observation is monitored for comparison
between output values and target data, with satisfactory error values, for instances,
MAE=1.5E-3, RMSE=0.0381, mean error=4.6E-3 and St.D=0.0382. It is universally
concluded that such the ANN method has the effectiveness and good rate of success
in finding optimal solutions.

Fig. 2. Comparison of compression strength of soft soil in case of: (a) predicted ANN vs. ex-
perimental values, (b) error mean and (c) standard deviation of the testing part

From a statistical point of view, the performance of the ANN algorithm needs to be
evaluated. As mentioned in the simulation procedure, 132 over 189 collected data
were randomly selected in order to construct the ANN algorithm in the training part.
Therefore, the performance of such model might be influenced by the choice of se-
lected data in the training part. Monte Carlo simulation [14] was selected as an effi-
cient numerical approach to quantify the uncertainties of the combination of input
parameters in the training part to the predicted results. Therefore, a total number of
10,000 numerical simulations were carried out. The R2 values of all these simulations
were plotted in Fig 3a and the corresponding histogram were plotted in Fig 3b. It is
observed that the proposed ANN model gave very satisfactory R 2 values within the
range from R2=0.85 to 0.95. The most frequent R2 obtained over 10,000 simulations
Development of Artificial Neural Networks... 1171

was R2=0.901 with a frequency of 536. Besides, a gamma function has also been ap-
plied to the R2 values and the maximum gamma set was 0.893 at a frequency of 453.
In general, the Cc of soft soil can be predicted using ANN algorithm within an
excellence error (average R2=0.901). It thus reasonably stated that such model is a
very promising AI tool for the prediction of the Cc, thus could provide useful
information for further investigation.

Fig. 3. (a) Scatter plot of R2 values for 10.000 simulations for testing part, (b) the correspond-
ing histogram of R2 values

5 Conclusions

Compression coefficient of soft soil (Cc) is an important parameter in the field of

geotechnical engineering. This study constructed and analyzed the accuracy of the
ANN for the prediction of such mechanical parameter. A total of 189 soft soil samples
collected along the Ninh Binh ± Hai Phong national highway and used for AI model-
ing of which thirteen factors (depth of sample, clay content, moisture content, bulk
density, dry density, specific gravity, void ratio, porosity, degree of saturation, liquid
limit, plastic limit, plasticity index, liquidity index) were considered as input variables
and the Cc was considered as output variable. Various methods namely MAE, RMSE,
and R2 were used to evaluate the performance of the ANN algorithm. The results
showed that the ANN are promising method for prediction of the Cc of soft soil with
excellence coefficient of determination (average R2=0.901) between the predicted and
target data. However, the correlation analysis among parameters used should be car-
ried out to remove the highly inter-correlated input factors for improving the ANN
performance. This study paves the way for geotechnical engineers to reduce the cost
of laboratory experiments as well as time and man power saving.

Acknowledgment: This research was carried out under the Transport Ministry
SURMHFW QDPHG ³$SSOLFDWLRQ RI DGYDQFHG DUWLILFLDO LQWHlligence methods of industry
revolution 4.0 in prediction of geo-environment in Hai Phong ± Ninh Binh coastal
1172 B. T. Pham et al.

URDG SURMHFW´ SURMHFW QXPEHU '7  7KH DXWKRUV WKDQN WR WKH 8QLYHUVLW\ RI
Transport Technology for the support.

References

[1] 6KDUPD +DUL ' DQG 'H $QLUEDQ ³0XQLFLSDO 6ROLG :DVWH /DQGILOO 6HWWOHPHQW 3RVW
FORVXUH3HUVSHFWLYHV´Journal of Geotechnical and Geoenvironmental Engineering, vol.
133, no. 6, pp. 619±629, Jun. 2007.
[2] B. M. Das and K. Sobhan, Principles of Geotechnical Engineering. Cengage Learning,
2013.
[3] 6 ' .RSSXOD ³6WDWLVWLFDO (VWLPDWLRQ RI &RPSUHVVLRQ ,QGH[´ GTJ, vol. 4, no. 2, pp.
68±73, Jun. 1981.
[4] A. C. ApOLQ<<DQJDQG6+DQVHQ³$VVHVVPHQWRIȕWKHFRPSUHVVLRQFRHIILFLHQWRI
PXGVWRQHVDQGLWVUHODWLRQVKLSZLWKGHWDLOHGOLWKRORJ\´Marine and Petroleum Geology,
vol. 12, no. 8, pp. 955±963, Jan. 1995.
[5] 3DUN-+DQG.RXPRWR7³1HZ&RPSUHVVLRQ,QGH[(TXDWLRQ´ Journal of Geotech-
nical and Geoenvironmental Engineering, vol. 130, no. 2, pp. 223±226, Feb. 2004.
[6] M. A. Shahin, M. B. Jaksa, and H. R. Maier, Artificial Neural Network Applications in
Geotechnical Engineering. .
[7] ³5HFHQW DGYDQFHV DQG Iuture challenges for artificial neural systems in geotechnical
HQJLQHHULQJ DSSOLFDWLRQV´ >2QOLQH@ $YDLODEOH
https://dl.acm.org/citation.cfm?id=2170264. [Accessed: 30-Mar-2019].
[8] B. Samanta, K. R. Al-Balushi, and S. A. Al-$UDLPL ³$UWLILFLDO QHXUDO QHtworks and
VXSSRUWYHFWRUPDFKLQHVZLWKJHQHWLFDOJRULWKPIRUEHDULQJIDXOWGHWHFWLRQ´Engineering
Applications of Artificial Intelligence, vol. 16, no. 7, pp. 657±665, Oct. 2003.
[9] M. B. Kia, S. Pirasteh, B. Pradhan, A. R. Mahmud, W. N. A. Sulaiman, and A. Moradi,
³$QDUWLILFLDOQHXUDOQHWZRUNPRGHOIRUIORRGVLPXODWLRQXVLQJ*,6-RKRU5LYHU%DVLQ
0DOD\VLD´Environ Earth Sci, vol. 67, no. 1, pp. 251±264, Sep. 2012.
[10] / 0DJQLHU DQG ) +DJKLJKDW ³0XOWLREMHFWLYH RSWLPL]DWLRQ RI EXLOGLQJ GHVLJQ XVLQg
7516<6VLPXODWLRQVJHQHWLFDOJRULWKPDQG$UWLILFLDO1HXUDO1HWZRUN´ Building and
Environment, vol. 45, no. 3, pp. 739±746, Mar. 2010.
[11] ³3')3HUIRUPDQFH$QDO\VLVRI9DULRXV$FWLYDWLRQ)XQFWLRQVLQ*HQHUDOL]HG0/3
Architectures of Neural NetwoUNV´ ResearchGate. [Online]. Available:
https://www.researchgate.net/publication/228813985_Performance_Analysis_of_Variou
s_Activation_Functions_in_Generalized_MLP_Architectures_of_Neural_Networks.
[Accessed: 30-Mar-2019].
[12] G. Montavon et al. ³0DFKLQH learning of molecular electronic properties in chemical
FRPSRXQGVSDFH´New J. Phys., vol. 15, no. 9, p. 095003, Sep. 2013.
[13] G. Schüürmann, R.-8(EHUW-&KHQ%:DQJDQG5.KQH³([WHUQDO9DOLGDWLRQDQG
Prediction Employing the Predictive Squared Correlation Coefficient ² Test Set Activi-
W\ 0HDQ YV 7UDLQLQJ 6HW $FWLYLW\ 0HDQ´ -Oct-2008. [Online]. Available:
https://pubs.acs.org/doi/abs/10.1021/ci800253u. [Accessed: 30-Mar-2019].
[14] R. Y. Rubinstein and D. P. Kroese, Simulation and the Monte Carlo Method, 3 edition.
Hoboken, New Jersey: Wiley, 2016.
 Real-time detection of asymmetric surface deformation
and field stress in concrete-filled circular steel tubes via
multi-vision method

Yunchao Tang1,2, Kuangyu Huang3, Lijuan Li2,*, Xiangjun Zou3, Wenxian Feng2
1
College of Urban and Rural Construction, Zhongkai University of Agriculture and Engineer-
ing, Guangzhou 510225, China
2
School of Civil and Transportation Engineering, Guangdong University of Technology,
Guangzhou 510006, China
3
College of Engineering, South China Agricultural University, Guangzhou 510642, China

Abstract. Aiming at the characteristics of full-field asymmetric surface defor-

mation on the surface of concrete-filled steel tubular columns, a mark-free mul-
ti-vision measurement method is proposed, and the vision system structure,
curved surface measurement principle, image acquisition, preprocessing algo-
rithm and 3D point cloud reconstruction algorithm are illustrated. An image
stitching algorithm is proposed to calculate the deformation radius of the
asymmetric surface and the radius of curvature and its height of the largest de-
formation section. In the seismic test, the stereo vision system developed by the
authors is used to automatically collect and process the three-dimensional de-
formation and field strain of the specimens. Besides, the three-dimensional
point cloud reconstruction algorithm is proposed to realize the full-field three-
dimensional deformation image reconstruction. The comparison results indicate
that the multi-ocular vision measurement method can measure the deformation
process of the asymmetric surface and the exact value of the full field strain, its
accuracy meets the engineering requirement. The proposed method can evaluate
the asymmetric deformation of the structure subjected to impact loads or vibra-
tion and provide real-time decision data for structural health monitoring.

Keywords: Computer vision, Non-contact measurement, Image reconstruction.

1 Introduction

Against these backgrounds, a multi-ocular stereo-vision system is proposed for dy-

namic asymmetric curved deformation tracking during the low cycle loading test. A
novel asymmetric deformation detection method is used for three-dimension recon-
struction. The full field true strains are obtained via point cloud registration.

© Springer Nature Singapore Pte Ltd. 2020 1173

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_188
1174 Y. Tang et al.

2 Methodology

Four-ocular visual detection process includes image acquisition, image pre-

processing, stereoscopic matching, point cloud computation and denoising, and image
stitching. The asymmetric deformation parameter and full-field strain are obtained via
image stitching. We use Visual Studio 2017 and OpenCV development software to
run the image processing algorithm and complete the automatic measurement. The
four-ocular visual detection and its image processing flow are shown in Fig. 1.

Fig.1. Four-ocular visual inspection and its image processing flow

Real-time detection of asymmetric surface deformation... 1175

2.1 Curvature radius fitting

The deformed edge of the steel tube is an irregular curve, as shown in Figure 2. The
curved surface of the deformation area is composed of n cross sections, and the diam-
eter of the i-th cross-section surface is:
GL  U UL   UL   (1)
To solve the radius of curvature of the arc U , the three-dimensional point cloud is
calculated by the circle fitting algorithm. Then the parameters (diameter and height)
of i-th cross-section surface in the deformation region of the steel tube are deter-
mined.
The nonlinear least-square algorithm is used to fit the point cloud top-view map.
The sum of squares of residuals is obtained, and the model coefficients corresponding
to the minimum values are taken. The equation for the target circle determined by the
top view of the resulting three-dimensional point cloud is:
( x  a ) 2  ( y  b) 2 U 2 (2)
Where a, b is constant. In combination with the depth direction and the position of
a specimen, the 3D point cloud that is reversely calculated:
y f ( x) b  U 2  ( x  a ) 2 (3)
According to the residual square sum minimum criterion, the objective function S
is:
n
min S (a, b, r ) ¦[b  U 2  ( x  a)2  yi ]2 (4)
i 1

A new optimization iterative process is performed by performing a first-order Tay-

lor expansion on the objective function and replacing the original function with its
OLQHDU SDUWLDO DSSUR[LPDWLRQ 7KH ILWWHG FLUFOH FRHIILFLHQW YHFWRU RI WKH VWHHO WXEH¶V
7
convex horizontal projection is set as E ª¬D E U º¼ . The corresponding iterative
optimization model is:
n š š š
min S ( E k 1 ) ¦[ y ( E i k )  zi ( E k ) E k 1 ]2 (5)
i 1

   
Where \ L EЦN is the constant of the (k+1) iteration, ]L EЦN is the constant of iter-
ation gradient.

d1-ȡ
d22-ȡȡ
dn-ȡ

Fig.2. 3D Reconstruction
1176 Y. Tang et al.

The nonlinear least-square problem is converted into a linear least squares question
by Gauss-Newton method, and the optimal solution U is obtained by linear theory.
Due to the asymmetry of the convex deformation in steel tubes, two sets of the binoc-
ular vision point cloud are stitched via circle fitting algorithm, and the diameter of the
steel tube in the cross-section is obtained by equation (1), as shown in Fig. .

dȡ
ɏͳ  ɏʹ 

(a) (b) (c)

(a)Point cloud top-view acquired by binocular camera A (b) Point cloud top-view acquired by
binocular camera B (c) Point cloud stitching
Fig.3. WŽŝŶƚĐůŽƵĚƐƚŝƚĐŚŝŶŐĚŝĂŐƌĂŵ

2.2 Asymmetric surface derivation and calculation

For the full-field surface deformation with load time t, we take the left and right edges
of Figure 2 as an example to illustrate the coordinates of the asymmetric edge points
and obtain the parameters of the three-dimensional coordinates of the convex surface
deformation. The parameters are calculated by the radius of curvature and height, and
the projection of the radius on the X-axis in the horizontal plane can also be solved by
the radius of curvature. The i-th section height h of the convex surface of the three-
dimensional reconstruction is ݄௜ .
If the deformation at the same height of the curved section is asymmetrical, the two
curvature radius values are not equal.
The coordinate points of the i-th point on the left and right edges of the section are
݂௜௟ ƒ†݂௜௥ corresponding coordinates are ݂௜௟ ሺߩ௜௟ ǡ ݄௜௟ ሻƒ†݂௜௥ ሺߩ௜௥ ǡ ݄௜௥ ሻ.
Since the left and right edge lines on the same section deformation height h are re-
spectively composed of consecutive n points, we can conclude that:
݂ଵ௟ ሺߩଵ௟ ǡ ݄ଵ௟ ሻǡ ݂ଵ௥ ሺߩଵ௥ ǡ ݄ଵ௥ ሻǡ ݂ଶ௟ ሺߩଶ௟ ǡ ݄ଶ௟ ሻǡ ݂ଶ௥ ሺߩଶ௥ ǡ ݄ଶ௥ ሻ ««, ݂௜௟ ሺߩ௜௟ ǡ ݄௜௟ ሻǡ ݂௜௥ ሺߩ௜௥ ǡ ݄௜௥ ሻ ,
««, ݂௡௟ ሺߩ௡௟ ǡ ݄௡௟ ሻǡ ݂௡௥ ሺߩ௡௥ ǡ ݄௡௥ ሻ
In fact, ݄௜௟ ൌ ݄௜௥ . If the deformation at the same height ݄௜ of the curved section is
symmetrical, ߩ௜௟ ൌ ߩ௜௥ .
At the height of the deformation zone, the curvature diameter of the largest defor-
mation section ݀௠௠௔௫ on the surface can always be found, and the maximum defor-
mation height is ݄௠ , then the coordinate point of the maximum deformation parame-
ter can be calculated: ݂௠௟ ሺߩ௠௟ ǡ ݄௠ ሻǡ ݂௠௥ ሺߩ௠௥ ǡ ݄௠ ሻ.
Real-time detection of asymmetric surface deformation... 1177

At this point, we can calculate the arbitrary section and height deformation point of
the asymmetric surface at a certain time t of the full field deformation, and calculate
the maximum deformation parameter and its height.

3 Test and result analysis

3.1 Comparison of visual and laser measurement

The convex curve diameter fitting results of about 183 sets of images of the four-eye
visual tracking are analyzed. As shown in Fig. , at the height of 47 mm, the full field
curvature diameter of the visual and laser measurement is compared. The average
error standard deviation was 1.23%, and the maximum absolute value average error
was 2.82%.

Fig.4. Variation tendency in diameter of the convex deformation of G1 specimen

4 Conclusions

In this paper, the full-field surface deformation of large-field concrete-filled steel

tubular column is tracked and measured by the target-less four-ocular stereo vision
system. The mathematical model of four-eye visual point cloud matching is con-
structed. The visual inspection and image processing software is developed to realize
automatic calibration and dynamic surface tracking measurement.
The comparison between vision and laser measurement shows that the average er-
ror standard deviation of the two sets of specimens is 1.23%, and the maximum abso-
lute value of the average error is 2.82%. The three-dimensional asymmetric surface is
reconstructed by the circle fitting algorithm, which realizes dynamic real-time track-
ing of structural deformation and full field strain calculation and provides an automat-
ic non-destructive non-contact measurement method for structural damage identifica-
tion and health monitoring.
1178 Y. Tang et al.

References
1. Kong X, Jian L.: Vision-based fatigue crack detection of steel structures using video fea-
ture tracking. Computerဨaided Civil & Infrastructure Engineering 33, 783-99 (2018).
2. Kim H., Kim H.: 3D reconstruction of a concrete mixer truck for training object detectors.
Automation in Construction 88, 23-30 (2018).
3. Feng D., Feng M.: Computer vision for SHM of civil infrastructure: From dynamic re-
sponse measurement to damage detection±A review. Engineering Structures 156, 105-17
(2018).
4. Feng D., Feng M.: Visionဨbased multipoint displacement measurement for structural
health monitoring. Structural Control and Health Monitoring 23, 876-90 (2018).
5. Pan B.: Digital image correlation for surface deformation measurement: historical devel-
opments, recent advances and future goals. Measurement Science and Technology 29,
082001 (2016).
6. Pan B., Tian L., Song X.: Real-time, non-contact and targetless measurement of vertical
deflection of bridges using off-axis digital image correlation. Ndt & E International 79, 73-
80 (2016).
7. Celik O., Dong C., Catbas F: A computer vision approach for the load time history estima-
tion of lively individuals and crowds. Computers & Structures 200, 32-52 (2018).
8. +XĖDG\ R., Hagara M.: A new procedure of modal parameter estimation for high-speed
digital image correlation. Mechanical Systems and Signal Processing 93, 66-79 (2017).
9. Shan B., Ren N., Xue Z. Stereovision-based surface deformation detection of offshore
wind turbine model under ship impact. Measurement 131, 605-14 (2019).
10. Chen Y., Wei J., Huang H., Jin W., Yu Q.: Application of 3D-DIC to characterize the ef-
fect of aggregate size and volume on non-uniform shrinkage strain distribution in concrete.
Cement and Concrete Composites 86, 178-89 (2018).
11. Khuc T., Catbas FN.: Computer vision-based displacement and vibration monitoring with-
out using physical target on structures. Structure and Infrastructure Engineering 13, 505-16
(2017).
12. Lee J., Lee KC., Cho S., Sim SH.: Computer vision-based structural displacement meas-
urement robust to light-induced image degradation for in-service bridges. Sensors 17, 2317
(2017).
13. Yoon H., Elanwar H., Choi H., GolparvarဨFard M., Spencer Jr BF.: Targetဨfree approach
for visionဨbased structural system identification using consumerဨgrade cameras. Structural
Control and Health Monitoring 23, 1405-16 (2016).
14. Ferrer B., Acevedo P., Espinosa J., Mas D.: Targetless image-based method for measuring
displacements and strains on concrete surfaces with a consumer camera. Construction &
Building Materials 75, 213-9 (2015).
15. Sun L., Abolhasannejad V., Gao L., Li Y.: Non-contact optical sensing of asphalt mixture
deformation using 3D stereo vision. Measurement 85, 100-17 (2016).
16. Ma Z., Liu S.: A review of 3D reconstruction techniques in civil engineering and their ap-
plications. Advanced Engineering Informatics 37, 163-74 (2018).
17. Wang N., Zhao X., Zhao P.: Automatic damage detection of historic masonry buildings
based on mobile deep learning. Automation in Construction 103, 53-66 (2019).
 A Possibility of AI Application on Mode-choice
Prediction of Transport Users in Hanoi

Truong Thi My Thanh1, Hai-Bang Ly1, and Binh Thai Pham1

1
University of Transport Technology, 54 Trieu Khuc, Thanh Xuan, Hanoi, Vietnam
thanhttm@utt.edu.vn

Abstract. Mode choice is a significant analysis of travel demand modelling.

The selection of transport mode highly depends on the travel behavior of
transport users, for instance, travel distance, trip cost, trip purpose, household
income and so on. This study aims at investigating the possibility of applying
Artificial Intelligent (AI) method to predict the mode choice through travel be-
havior survey data with a focus on Hanoi city - Vietnam. Firstly, travel inter-
view survey was conducted with the involvement of 311 transport users at dif-
ferent land-use types. The study secondly applies the Ensemble Decision Trees
(EDT) method to predict the mode-choice of transport users. Finally, the rec-
ommendation for a possibility of AI application on travel mode-choice is also
proposed. The results of this study might beneficial for transport planners and
transport authorities. The application of AI on parking demand forecast also
contributes for the big data application on transport demand modeling.

Keywords: Travel Behavior, Mode Choice, Machine learning; Ensemble Deci-

sion Trees

1 Introduction

Rapid development of motorcycle traffic in Asian developing countries is a unique

phenomenon. During late 1980s-early 2000s, motorcycle population exponentially
grew in Asian countries, for instances, China 25% per year, Vietnam 15% and India
11%. As a result, Asian cities experienced rapidly increasing and large shares of mo-
torcycle trips, remarkably 80-90% in Hanoi and Ho Chi Minh City, 60% in Jakarta
and around 30% in Taipei and Bangkok [1]. In Vietnam, motorcycles and cars are the
two main modes of transport in terms of absolute volume and contribution to cargo
transport in the whole country, especially in urban areas and economically developed
areas. Motorcycles are by far the dominant mode.
From international studies, there are many factors influencing travel mode choice.
The impact aspects relate to socio-economic characteristics of travelers, including
gender, occupation, income, trip cost, travel distance [2], travel mode and departure
time [3], the quality of walkway [4], and car ownership rate [5]. The influences might
come from the spatial configuration of land use, urban structure, and transport struc-
tures [6]. Other influences relate to public transport service, for instance, the accessi-
bility to the bus station (walking distance and the quality of the walkway) [7], provi-

© Springer Nature Singapore Pte Ltd. 2020 1179

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_189
1180 T. T. M. Thanh et al.

sion of transit system, bus fare, and the quality of bus service. Additionally, parking
pricing is also used as an effective instrument for traffic management.
In transport field, discrete choice analysis is traditionally used to analyze and pre-
dict travel decision-making. It is a probabilistic model to predict probabilities of dif-
ferent effects with binomial model or multinomial model. Therefore, discrete choice
analysis is utilized to predict the probability that an alternative is chosen, given cer-
tain values of input variables, which collected through travel interview surveys. How-
ever, there are still gaps regarding the uncertainties in conducting travel data surveys
and modeling a non-linear correlation between travel alternatives and influenced fac-
tors. Those gaps cannot be solved using traditional techniques. Recently, methods
from the field of Artificial Intelligence (AI) are a promising alternative to statistical
approaches for modeling travel decisions.

2 Decision Trees Algorithm

Ensemble Decision Trees (EDT) is a non-parametric supervised learning method

which has been applied in solving many real-world problems such as soft classifica-
tion [8], business failure prediction [9], natural hazard prediction [10]. It has several
advantages such as (i) it does not require prior assumptions related with the distribu-
tion of the input data, (ii) it can handle complex nonlinear relationship between input
and output variables, (iii) it can also deal with both numeric and categorical inputs,
and (iv) it is easy to interpret as the classification tree structure is explicit [11].
In this study, we have used the E'7ZLWKVSOLWFULWHULRQ*LQL¶VGLYHUVLW\LQGH[DQG
four maximum number of splits to predict travel mode-choice. Several statistical cri-
teria namely accuracy and Root Mean Square Error (RMSE) [12] were used for vali-
dating the performance of the EDT.

3 Data Preparation

$VXUYH\RQSDUNLQJXVHU¶VEHKDYLRUZDVFRQducted in June 2016 in three major zones

in Hanoi, including Old Quarter Area, Developed Area, and New Development Area.
Old Quarter Area is characterized by ancient buildings. It is the location of the busiest
shopping streets and tourist attractions. Developed Area locates inside the Ring Road
II and having the highest population density. There is mixed land use area, with the
combination of Government authorities, office buildings and shopping streets. New
Development Area places inside the area of Ring Road II and Ring Road III with
modern condominiums, office buildings and universities.
A total of 311 people was randomly selected at the buildings of different land use
types. They were directly interviewed to analyze their travel behavior changes under
the impacts of some factors including travel distance, trip purpose, household income,
and parking charge. All transport users used individual vehicles. The respondents
were given two options of mode-choice, including (1) still using individual mode or
(2) changing to public transport.
A Possibility of AI Application on Mode-choice... 1181

Among 311 transport users to be interviewed, there are 24 car users, 242 motorcy-
cle users, and 45 bicycle users. Their socioeconomic characteristics are presented in
Fig.1. The survey revealed that among 311 interviewees, most of the car users used
their vehicles for working purposes with 75% and for shopping purpose with 21%.
Meanwhile, 65% of the motorcyclists used their vehicle for work and 29% used it for
shopping trips. As for bicycle users, 70% used their vehicle for shopping, 15% for
going to school and 11% for going to work.

Fig.1. Socioeconomic characteristics of the samples (N=311)

(Exchange rate in 2018: 1 USD = 22,680 VND)

4 Results and Discussion

The objective of this work lies in the construction and validation of an artificial intel-
ligence model based on decision trees algorithm for the prediction of travel mode-
choice in transport sector. Survey data are collected with the final aim at investigating
travel behavior under certain situation. Based on such data, an AI-based numerical
tool is constructed to predict the travel mode-choice of transport users. These re-
sponses have an important effect on travel demand, especially in case of Hanoi city.
The collected travel behavior data are divided into 2 parts: the training part (70% of
data or 218 samples) is used to construct the prediction tool, whereas the testing one
(30% of data or 93 samples) is used for validation of model. In the present study,
input parameters for the AI numerical tool are: I1, I2, I3, I4, I5, I6, I7, I8, I9, I10,
indicating the gender, age, study, occupation, total monthly house hold income, trip
mode, trip purpose, trip length, trip cost and parking fee, respectively. The output (O)
refers to the mode choice of travel users, discretized in binary classification as 2
choices: using current transport mode (1) or shifting to others (0).
Firstly, the accuracy of the constructed AI model is performed via a confusion ma-
trix. The latter visualizes the performance of a given algorithm, where each row rep-
resents the predicted response of the model and each column represents the target
survey data. Confusion matrix between target and predicted values is shown in Fig 2.
It can be clearly observed that the predicted values obtained from the training and the
testing data of the EDT algorithm is in excellence agreement with the survey data.
1182 T. T. M. Thanh et al.

Such hypothesis is validated in both training part as well as validation/testing one. For
the training part, correct and false prediction accuracies are 98.2% and 1.8%, respec-
tively, whereas for the testing part, these values are 95.7% and 4.3%. Considering the
training part, the correct response in case of shifting to another transport mode is 12
(92.3%) and the false one is 1 (7.7%), whereas that in the case of staying with current
transport mode is 202 (98.5%) and 3 (1.5%), respectively. For the testing part, the
accuracy of the model is satisfactory with 95.7% correct responses, 4 in the case of
shifting to another transport mode and 85 for the other. These results clearly show that
the EDT model is practicable method for predicting the mode choice of travel users.

Fig. 2. Target and prediction output values in confusion matrix form of mode choice
with 0 = change and 1=keep mode: (left) training EDT, (right) testing EDT
Errors between survey data and prediction output are plotted in Fig 3. The values
of error criteria for the training part are RMSE=0.1355 and MAE=0.0183. Regarding
the testing part, these values are RMSE=0.2074 and MAE=0.0430. Histogram of the
errors are also plotted for the training part (Fig 3c) as well as the testing part (Fig 3d).
The average error of the training part is Err Mean=-0.0092 and the corresponding
standard deviation is St.D=0.1355, while these values for the testing part are Err
Mean=-0.0215, St.d=0.2074.
In a statistical and mathematical point of view, the performance of the proposed AI
algorithm needs to be evaluated. As mentioned above, 218 over 311 collected data
were randomly selected to construct the AI algorithm. Therefore, the accuracy of such
model might be influenced by the choice of selected training data. Monte Carlo simu-
lations were selected as an efficient numerical approach to quantify the uncertainties
of the choice of input parameters to the predicted results. A total number of 10,000
numerical simulations were carried out. The histogram of the accuracy and RMSE
errors are plotted in Fig 4. In general, the mode choice of transport users can be pre-
dicted using EDT algorithm within an excellence accuracy (96% to 97%). It thus
reasonably stated that such model is a very promising AI tool for the prediction of the
choice of transport users, thus could provide useful information for further investiga-
tion. Similar observations were also monitored for RMSE, as lowest value of RMSE
indicates highest AI performance.
A Possibility of AI Application on Mode-choice... 1183

Fig 3. Error between target and prediction values: (a) Error training EDT, (b) Error
testing EDT, (c) Histogram training EDT and (d) Histogram testing EDT

Fig 4. Histogram of R and RMSE for 10000 simulations: (a) R for testing EDT, (b)
RMSE for testing EDT

5 Conclusions

In this study, an AI approach namely EDT has been proposed for prediction of mode-
choice of transport users in Hanoi. A comprehensive understanding of the characteris-
tics of travel demand and the possible impacts of trip time, travel distance, trip cost,
household income, and parking pricing on the mode choice of transport users in Ha-
noi was analyze to construct the database for constructing the prediction model using
1184 T. T. M. Thanh et al.

EDT. Various criteria namely Accuracy and RMSE were used to validate the predic-
tive capability of the model. The results show that the EDT is a promising approach
for prediction of mode-choice of transport users in Hanoi with accuracy of 96% to
97% and RMSE of 0.145 to 0.181.
In short, the results of this study would be useful for transport planners and
transport authorities to formulate effective transport management strategies to manage
urban transport in developing countries. Further study should be conducted on the
modal shift under individual measures, and also in combination with other manage-
ment measures such as time-restriction and location restriction parking measures.

References
[1] S. Morichi and S. R. Acharya, Transport Development in Asian Megacities, no. April.
Berlin, Heidelberg: Springer Berlin Heidelberg, 2013.
[2] % %URZQ ³0RGDO FKRLFH ORFDWLRQGHPDQG DQGLQFRPH´ J. Urban Econ., vol. 20,
no. 2, pp. 128±139, 1986.
[3] 6<+HDQG-7K¡JHUVHQ³7KHLPSDFWRIDWWLWXGHVDQGSHUFHSWLRQVRQWUDYHOPRGH
choice and car ownership in a Chinese megacity: The case of GuaQJ]KRX´Res. Transp.
Econ., pp. 1±11, 2017.
[4] 6 3DUN ³'HILQLQJ 0HDVXULQJ DQG (YDOXDWLQJ 3DWK :DONDELOLW\ DQG 7HVWLQJ ,WV
,PSDFWVRQ7UDQVLW8VHUV¶0RGH&KRLFHDQG:DONLQJ'LVWDQFHWRWKH6WDWLRQ´
[5] X. Li, C. Shao, L. Yang, D. de la Hoz, anG$0RQ]yQ³0RGHOLQJWKH,QIOXHQFHRI
6RFLRHFRQRPLF &KDUDFWHULVWLFV DQG /DQG 8VH RQ &RPPXWH 0RGH &KRLFH LQ 0DGULG´ LQ
2009 Sixth International Conference on Fuzzy Systems and Knowledge Discovery, 2009,
vol. 2, pp. 567±571.
[6] .6DVDNL³,QFRPHFODVVPRGDOFKRLFHDQGXUEDQVSDWLDOVWUXFWXUH´J. Urban Econ.,
vol. 27, no. 3, pp. 322±343, 1990.
[7] 1 /LPWDQDNRRO 0 'LMVW DQG 7 6FKZDQHQ ³7KH LQIOXHQFH RI VRFLRHFRQRPLF
characteristics, land use and travel time considerations on mode choice for medium- and
longer-GLVWDQFHWULSV´J. Transp. Geogr., vol. 14, no. 5, pp. 327±341, 2006.
[8] 0 ;X 3 :DWDQDFKDWXUDSRUQ 3 . 9DUVKQH\ DQG 0 . $URUD ³'HFLVLRQ WUHH
UHJUHVVLRQIRUVRIWFODVVLILFDWLRQRIUHPRWHVHQVLQJGDWD´Remote Sens. Environ., 2005.
[9] $*HSS..XPDUDQG6%KDWWDFKDU\D³%XVLQHVVIDLOXUHSUHGLFWLRQXVLQJGHFLVLRQ
WUHHV´J. Forecast., 2010.
[10] A. Jaafari et al. ³0HWD RSWLPL]DWLRQ RI DQ DGDSWLYH QHXUR-fuzzy inference system
with grey wolf optimizer and biogeography-based optimization algorithms for spatial
SUHGLFWLRQRIODQGVOLGHVXVFHSWLELOLW\´Catena, 2019.
[11] 0 $ )ULHGO DQG & ( %URGOH\ ³'HFLVLRQ WUHH FODVVLILFDWLRQ RI ODQG FRYHU IURP
UHPRWHO\VHQVHGGDWD´Remote Sens. Environ., 2002.
[12] H. I. Erdal and O. Karakurt³$GYDQFLQJPRQWKO\VWUHDPIORZSUHGLFWLRQDFFXUDF\RI
&$57PRGHOVXVLQJHQVHPEOHOHDUQLQJSDUDGLJPV´J. Hydrol., 2013.
 Machine learning based tool for identifying errors in
CAD to GIS converted data

Mohamed Badhrudeen1, Nalin Naranjo1, Ali Mohavedi1, and Sybil Derrible1

1
Complex and Sustainable Urban Networks (CSUN) Lab, University of Illinois at Chicago, IL
60607 USA

Abstract. Recent advances in Geographic Information System (GIS) has made

the storage, manipulation, and analysis of spatial data easier than ever. As a re-
sult, many public and private agencies have switched from the traditional Com-
puter Aided Design (CAD) format to GIS for storing information about their in-
frastructure. The existing data stored in CAD therefore needs to be converted to
GIS, and this process brings about at least two problems. First, GIS requires ge-
ographical coordinates that CAD data do not have, and the accurate projection
and alignment of the infrastructure is not straightforward. Second, the original
CAD data often possess errors such as overlapping lines, split lines, presence of
text, and human errors, in addition to errors introduced during the conversion
process. The goal of this study is to develop and apply a tool that can identify
these errors and correct them automatically without human intervention. This
study focuses specifically on the identification of the errors using decision tree
learning. This tool can significantly reduce the time spent identifying errors on
the GIS data obtained after the conversion. Finally, as a case study, this tool is
applied to the wastewater system of the University of Illinois at Chicago campus.

Keywords: GIS, CAD, Underground Infrastructure, Decision Tree, Machine

Learning, Topological Error

1 Introduction

Across the globe, many public and private agencies collect and use data related to in-
frastructure such as roads, water pipes, and wastewater conduits [1], often with the aim
to design smarter, more sustainable, and more resilient infrastructure systems [2, 3].
These agencies tend to be independent of one another, and the collected data is rarely
stored in a common database²with the notable exception of transport infrastructure,
for which impressive databases were created, such as openstreetmap. Moreover, im-
portant infrastructure data is often missing or contain errors that limit the ability to use
the data for practical purposes. In parallel, the adoption of Geographic Information Sys-
tems (GIS) technology has been gaining popularity, partly to overcome this problem.
Further, the ease of working with GIS data and all the advantages it brings with it has
made public and private agencies put more effort into storing infrastructure data in GIS
format. Nonetheless, data on existing infrastructure²especially for underground infra-
structure²are usually stored in Computer Aided Design (CAD) format.

© Springer Nature Singapore Pte Ltd. 2020 1185

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_190
1186 M. Badhrudeen et al.

The conversion process from CAD to GIS format is not straightforward and usually
does not yield accurate GIS data. In fact, the converted GIS files are generally fraught
with inaccuracies²mainly topological²in addition to the inherent inaccuracies asso-
ciated with the CAD data. Furthermore, in order to integrate the existing data with new
data, geolocation information shared among infrastructure systems must be identical.
CAD data generally does not possess any geolocation information, so it must be added
to the converted data manually. In order to build an accurate database, inaccuracies
present in the original data or generated during the conversion process should be cor-
rected before added to the GIS database. These corrections are usually carried out man-
ually; however, if the volume of the information converted is large, it is difficult to do
so manually. Further, the person who is responsible for managing GIS data in an or-
ganization may not have sufficient background about the infrastructure being managed,
further complicating the problem. One way of tackling this problem is to leverage the
capabilities of data science and machine learning techniques that have proven to be
powerful and applicable to engineering problems [4, 5]. In the context of this work,
machine learning can be used to identify potential errors and inaccuracies in GIS, which
then can then be corrected either manually or automatically.
In the GIS literature, the application of machine learning remains limited [6, 7]. GIS
is an umbrella term that encapsulates most physical systems associated with their geo-
location element and is often used both in practice and in research [8]. Past research
studies applied machine learning algorithms on information such as landslide suscepti-
bility assessment [9, 10], urban flood susceptibility assessment [11, 12], spatial model-
ing using rules and pattern extracted from the dataset [7], groundwater mapping [6],
and so on. These studies used algorithms such as decision tree learning, random forest,
k-nearest neighbor, and support vector machine. Most existing studies, however, used
raster-based GIS data for their analysis as opposed to vector-based GIS data. Generally,
the identification of errors can be posed as a classification problem (i.e., a piece of
infrastructure possesses an error or not), and the information extracted from the GIS
data can be used to train the machine learning model for error identification.
The main goal of this study is to use machine learning to identify potential errors
resulting from the CAD to GIS conversion process, along with errors in original CAD
data. To that end, this study uses wastewater network data from the University of Illi-
nois at Chicago campus for error identification, and these data are converted from CAD
to vector-based GIS data. Though the necessary information can be extracted from the
CAD data before conversion to identify errors, it will not account for the errors intro-
duced during the conversion process. Therefore, all the information is extracted from
the converted GIS data for the subsequent process.

2 Methodology

2.1 Data

The data used here represent the wastewater system of the University of Illinois at Chi-
cago (UIC) campus; original CAD data from the campus administration was used and
partly synthesized for practical purposes in this work. A map of the system in CAD is
Machine learning based tool for identifying errors... 1187

shown in figure 1. The CAD data contain information on the location of underground
sewer pipes, catch basins (mostly at street intersections), drainage pipes (connecting
catch basins to the sewer pipes), nearby buildings (connecting buildings to sewer pipes),
and manholes. In CAD, these are simply drawings with lines representing the pipes,
and circles representing the manholes and catch basins. The features of interest here are
manholes, catch basins, sewer pipes, drainage pipes, and buildings.

Fig. 1. CAD map of the UIC campus wastewater system

To be able to use machine learning, we need to have a training and a testing dataset.
By studying the data manually, we observed the presence of three types of errors. Error
type 1 occurs after the CAD to GIS conversion. In the CAD drawings, the manholes
and catch basins are represented as circles, and the sewer pipes are not connected to the
center but to the edge of the circles. In the conversion process, the circles are replaced
with point features and the sewer pipes are converted to lines, which creates a gap be-
tween the two features (i.e., the sewer pipes are not connected to the manholes and
catch basins). These types of errors are labeled ³error type 1.´ Error types 2 and 3 were
usually present in the CAD drawings directly, when the drawings were first created.
For example, we found that when two sewer pipes are connected, a manhole is placed
on top of the connection for access. However, at some connections the manholes are
not placed exactly on top of the pipes, leading to ³error type 2,´ or the manholes are
incorrectly placed elsewhere, leading to ³error type 3.´ Based on these findings, we
created a new variable in the attribute table in GIS indicating 1 if the feature (either
manholes or pipes) possesses one of the three error types, and 0 otherwise (i.e., no er-
ror). This new variable is used as the dependent variable to train and test the machine
learning algorithm. It should be noted that this study only focuses on identifying if a
feature has an error or not, identifying the error type is RXWRIWKLVVWXG\¶V scope.
From the GIS data, we further extracted the following five variables, all bound in [0,
f[, that serve as independent variables; they are:
1188 M. Badhrudeen et al.

x Number of connections to a manhole ± M1

x Number of connections to a sewer pipe ± S1
x Number of connections to a drainage pipe ± D1
x Number of connections to a catch basin ± C1
x Number of connections to a building ± B1

2.2 Decision Tree

The machine learning algorithm used to identify the errors is decision tree learning.
Decision tree learning was selected because of its rule-based nature that aligns with the
typology of errors present in the data. Moreover, the rules of a decision tree are inter-
pretable and provide logical statements, which make decision tree learning a good fit
for this problem. Each decision tree has three types of nodes: root node (topmost node),
internal node, and leaf node (end node). Decision tree algorithms use simple rules to
split up a root, go through internal nodes, and finally end up with the leaves [13]. We
note that ensemble methods that use multiple trees (e.g., random forest, gradient boost-
ing) are also available, but they should not be needed here because of the deterministic
nature of the problem (i.e., one tree should be able to capture all rules needed to identify
the errors present).
Decision tree learning generally finds the rules based on the impurity of the data
after the split. In this study, and as often used in classification problem, we use the
³Gini impurity index´>14] to measure the impurity after the split on a node. The Gini
impurity is calculated by the following formula:
௠
‫݅݊݅ܩ‬ሺ‫ܦ‬ሻ ൌ ͳ െ σ௜ୀଵ ‫݌‬௜ଶ (1)

where, D is a data partition, m is the set of classes (i.e., error / no error), and pi is the
probability of belonging in a tuple in D with class Ci, which is calculated by |Ci, D|/|D|.
For each split, the decision tree algorithm calculates the Gini index of all the varia-
bles in the data partition D. The variable that provides the lowest Gini index is selected
for the split. The process of splitting continues until the current Gini index is less than
any new split with the remaining variables. In the decision tree, we can also understand
how important each decision variable is for predicting the target value using the concept
of Feature Importance (FI). FI gives us a way WRDVVHVVWKHGHFLVLRQYDULDEOHV¶overall
contribution. FI LVPHDVXUHGE\FDOFXODWLQJWKHGLIIHUHQFHEHWZHHQWKHPRGHO¶VSUHGLF
tion before and after permuting the feature.

3 Results

After the decision variables M1, S1, D1, C1, B1 are extracted from the GIS data, we
trained the decision tree algorithm using 70 % of the data with the Python library scikit-
learn [15]. Finally, we used the trained decision tree model on the remaining 30 % of
the data to check the accuracy. We found that the model achieved an accuracy of 92%.
Moreover, a 6-fold cross validation analysis gave us an accuracy of 92% r 1%. The
final decision tree structure is shown in figure 2.
Machine learning based tool for identifying errors... 1189

Fig. 2. Decision Tree structure

Fig. 3. Feature importance analysis

,QDGGLWLRQWRWKHPRGHO¶VSHUIRUPDQFHZHcalculated the FI of each variable; see

figure 3. From the figure, we can see that number of connections to sewer pipes (S1)
contributed more to the accuracy among all variables. The second most important de-
cision variable is number of connections to catch basins (C1). Further, it is interesting
to see the information about the number of connections to buildings and manholes²
B1 and M1²have a significantly lower contribution in the model.
In conclusion, we converted CAD drawings of the UIC campus wastewater system
into GIS format and found three types of errors. We then trained a decision tree using
70% of the data to identify the errors. Finally, we used the trained model on the 30% of
the data, which yielded an overall accuracy of 92%. Future work should focus on im-
proving the current accuracy (potentially by creating new independent variables) and
1190 M. Badhrudeen et al.

on developing a model (possibly using machine learning) to automatically identify the

error types (i.e., 1, 2, or 3), and fix those errors detected that will result in clean and
workable GIS data.

References
1. Balasubramani, B.S., Belingheri, O., Boria, E. S., Cruz, I., Derrible, S., Siciliano, M.D.:
GUIDES: Geospatial Urban Infrastructure Data Engineering Solutions. In: 25 th ACM
SIGSPATIAL International Conference on Advances in Geographic Information Systems,
Article 90, ACM, California, USA (2017).
2. Mohareb, E., Derrible, S., PeiraviDQ),QWHUVHFWLRQVRI-DQH-DFRE¶s Conditions for Diver-
sity and Low-Carbon Urban Systems: A Look at Four Global Cities. Journal of Urban Plan-
ning and Development. 142(2), (2016).
3. Derrible, S.: An Approach to Designing Sustainable Urban Infrastructure. MRS Energy and
Sustainability 5 (E15), (2018).
4. Ahmed, N., Derrible, S., Cabezas, H.: Fisher Information to Assess Stability in the Perfor-
mance of Public Transportation Systems. Royal Society Open Science 4(4), 160920 (2017).
5. Ahmed, N., Derrible, S.: An Information Theory Based Robustness Analysis of Energy Mix
in US States. Energy Policy 120, 167-174 (2018).
6. Naghibi, S. A., Pourghasemi, H. R., Dixon, B.: GIS-Based Groundwater Potential Mapping
using Boosted Regression Tree, Classification and Regression Tree, and Random Forest
Machine Learning Models in Iran. Environmental Monitoring and Assessment 188(44), 1-
27 (2016).
7. Qi, P., Zhao, C., Feng, Z.: GIS- and Machine Learning ± Based Modeling of the Potential
Distribution of Broadleaved Deciduous Forest in the Chinese Loess Plateau. GIScience and
Remote Sensing 47(1), 99-114 (2013).
8. Kermanshah, A., Derrible, S.: Robustness of Road System to Extreme Flooding using ele-
ments of GIS, Travel Demand and Network Science. Natural Hazards 86 (1), 151-164
(2017).
9. Yeon, Y. K., Han, J.G., Ryu, K. H.: Landslide Susceptibility Mapping in Injae, Korea, using
a Decision Tree. Engineering Geology 16 (3-4), 274-283 (2010).
10. Neuhauser, B., Terhorst, B.: /DQGVOLGH6XVFHSWLELOLW\$VVHVVPHQWXVLQJ³:HLJKWV-of- Evi-
GHQFH´$SSOLHGWRD6WXG\$UHDDWWKH-XUDVVLF(VFDUSPHQW(SW-Germany). Geomorphology
86(1-2), 12-24 (2007).
11. Abebe, Y., Kabir, G., Tesfamariam, S.: Assessing Urban Areas Vulnerability to Pluvial
Flooding using GIS applications and Bayesian Belief Network Model. Journal of Cleaner
Production 174, 1629-1641 (2018).
12. Sadat, Y.K., Arabsheibani, R., Karimipour, F., Nasseri, M.: A New Approach to Flood Sus-
ceptibility Assessment in Data-Scarce and Ungauged Regions based on GIS-Based Hybrid
Multi Criteria Decision-Making Method. Journal of Hydrology 572, 17-31 (2019).
13. Han, J., Pei, J. Kamber, M.: Data Mining: Concepts and Techniques. 3 rd edition. Elseveir,
USA (2011).
14. Raileanu, L.E., Stoffel, K.: Theoretical Comparison Between the Gini index and the Infor-
mation Gain Criteria. Annals of Mathematics and Artificial Intelligence 41, 77-93 (2004).
15. Scikit Learn: Decision Tree, https://scikit-learn.org/stable/modules/gener-
ated/sklearn.tree.DecisionTreeClassifier.html last accessed 2019/03/26.
Development of 48-hour Precipitation Forecasting Model
using Nonlinear Autoregressive Neural Network

Tien-Thinh Le1*, Binh Thai Pham2, Hai-Bang Ly2; Ataollah Shirzadi3; and Lu Minh
Le4
1
NTT Hi-Tech Institute, Nguyen Tat Thanh University, Ho Chi Minh City, 700000, Vietnam
tienthinhle.vn@gmail.com
2
University of Transport Technology, Trieu Khuc, Thanh Xuan, Hanoi 100000, Vietnam
binhpt@utt.edu.vn; banglh@utt.edu.vn
3
Department of Rangeland and Watershed Management, Faculty of Natural Resources, Uni-
versity of Kurdistan, Sanandaj, Iran, atashirzadi@gmail.com
atashirzadi@gmail.com
4
Faculty of Engineering, Vietnam National University of Agriculture, Gia Lam, Hanoi 100000,
Vietnam
lmlu@vnua.edu.vn

Abstract. Rainfall intensity has a significant impact on urban drainage infra-

structures and the precipitation forecast therefore remains essential in urban ar-
eas. In this study, a prediction model using Nonlinear Autoregressive Neural
Networks (NANN) was proposed to forecast 48-hour-ahead the rainfall intensi-
ty. The proposed NANN model, which is based on a precipitation data of five-
year time series, was constructed and validated using various parameters such
as Coefficient of Determination (R2), Root Mean Squared Error (RMSE) and
Mean Absolute Error (MAE). The results exhibited a high statistical correlation
between the outputs of NANN model and the measured data for 48 hour ahead
prediction, i.e. R2=0.8998, RMSE=3.2909 and MAE=1.8672. This indicates
that the developed model is very promising for precipitation forecasting and
could contribute to improve the urban drainage systems.

Keywords: Nonlinear Autoregressive Neural Networks, Precipitation, Time Se-

ries, Forecasting.

1 Introduction

Infrastructures in urban areas are dependent on rain frequency and intensity. Ex-
treme precipitation intensity could induce disruption in transport network, which is
the key of social and economic activities of the urban zone [1]. The damage to the
drainage system could be even more significant in the case of high intensity and pro-
longed precipitation. In 2014, a flood happened in Genoa city, Italia, due to heavy
rainfall [2]. After the flood, a study of University of Genoa demonstrated that the
current flood warning system of the city needed to be improved. The consequences of
urban floods to buildings and city infrastructures are also remarkable. For instance, in
[3], Merz et al. estimated a high level of flood damage to buildings based on data

© Springer Nature Singapore Pte Ltd. 2020 1191

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_191
1192 T.-T. Le et al.

from nine flood events of urban areas in Germany. In the case of underground net-
works, precipitation intensity is highly critical and could be a relevant factor of many
undesirable problems. In these circumstances, the rainfall in urban areas remains so
challenging and receives a huge attention from researchers from all around the world,
especially in terms of forecast.
In the last decade, Artificial Intelligent (AI) techniques have been widely proposed
for forecasting various meteorological variables such as solar radiation, wind speed
and direction or temperature [4]. The methods exposed a high capability that is not
found in traditional models to predict such atmospheric parameters, which are highly
stochastic in nature [5]. As an example, an AI technique was successfully applied for
precipitation forecasting in the Sao Paulo region of Brazil [6]. A prediction model of
daily rainfall time series was also proposed using data-driven technique for Daning
and Zhenshui river basins in China [7]. However, rain is one of the most stochastic
nonlinear phenomena of the atmospheric processes [5]. The dependency of rain inten-
sity on other meteorological parameters and also on location and time are reported in
many studies [5]. Therefore, the main purpose of this research is to develop an effi-
cient AI forecast model in order to fully characterize the nonlinear behavior of precip-
itation time series, especially for urban areas. With this aim, a Nonlinear Autoregres-
sive Neural Network was trained and validated using data acquired from different
meteorological stations. The developed model was finally applied for a 48-hour fore-
cast of the precipitation.

2 Material and method

2.1 Methodology

In this research, the methodology was based on two main steps such as (i) the col-
lection of meteorological data and (ii) the construction and validation of a neural net-
work model for forecasting the precipitation intensity.
As shown in the previous studies in the literature, neural network models exhibited
a strong capability in in predicting the time series data. For instance, Panapakidis et
al. [8] has successfully applied an artificial neural network model for the prediction of
electricity price in a context of energy management. In another attempt, Doucoure et
al. [9] has combined an artificial neural network model with the Wavelet decomposi-
tion in order to reduce the complexity in a wind forecast problem.
In this study, the Nonlinear Autoregressive Neural Networks (NANN) model was
chosen for investigating the time series data of precipitation. The reasons of the
choice are the followings: firstly, the NANN model is simple to implement because of
its one-dimensional space [10]; secondly, the NANN model is robust in tracking high
order fluctuations in time series data [11, 12]. Finally, it is easily applicable because
of its suitability for forecasting any rainfall time series without the requirement of
other meteorological variables or location. Indeed, regarding to the meteorological
data in real condition, not every station has the same radars nor the same number of
meteorological variables [5, 13]. The description of the selected NANN model is
presented in the next section.
Development of 48-hour Precipitation Forecasting... 1193

2.2 Method used: NANN model

The main idea of the proposed NANN model is to forecast future values of a time
series based on its history [14]. Mathematically, the NANN model is expressed by the
following equation [15] (see Fig 1):
\W )  \W   \W  « \W  G  (1)
where y(t) is the value of the considered time series y at time t, d is the time delay
and F is the unknown prediction function. \W   \W  « \W  G are the d-
historical values of the time series. In most of the cases, the prediction function F is in
nonlinear form because of nonlinear behavior of time-varying variables, particularly
in terms of meteorological processes [16].

Fig 1. NANN model used in this study.

In this research, the weight and bias values of the neural network were updated ac-
cording to the common learning Levenberg-Marquardt technique [17] in order to
optimize the prediction function F. It should be noticed that the number of hidden
layers and neurons per layer largely depend on the specific problem. For that reason,
they are flexible. Increasing the number of hidden layers and neurons per layer might
make the network more complicated, especially costly in computational training time.
In this study, we selected one hidden layer for the NANN model and there was ten
neurons in this layer. Such architecture was chosen with respect to both achieved
performance of the model and consuming time. Finally, a sigmoid activation function
was applied to calculate the weights of inputs in the NANN model while mean
squared error was the cost criterion of the minimization problem.
Various statistical validation criteria such as Coefficient of Determination (R2),
Root Mean Squared Error (RMSE) and Mean Absolute Error (MAE) [18] were used
in this work in order to investigate the performance of the developed model. The de-
tails of the formulation of R2, RMSE and MAE could be found in the literature [18].
2.3 Data used

In this study, the time series precipitation data (in mm) were collected from a rain-
fall gauge located at Latitude 20.763m and Longitude 105.312m (Cao Phong district,
Hoa Binh province, Vietnam), for a period between 2007 and 2011. The precipitation
data were divided in two parts, 1278 first values (approximately 70%) were used for
training the AI model and 548 last values (approximately 30%) were served as the
testing part. This ratio of 70-30 was chosen as suggested by Salcedo-Sanz et al. [19]
and Sharma and Manish [20] in order to ensure the efficiency in the training and test-
ing phases of the AI model.
1194 T.-T. Le et al.

Hoa Binh city is the capital of Hoa Binh province, located 76 kilometres from Ha-
noi, the capital of Vietnam, and 5 kilometres from the Da river. Hoa Binh is also one
of the most prone provinces in Vietnam for floods because of its typical climate and
location. Consequently, Hoa Binh is still faced with various problems associated with
expanding urbanization, particularly in terms of drainage systems. Currently, Hoa
Binh city has an urban area of 133.34 km2 with a population density of 1418/km2. In
recent years, various master plans were established in order to develop the infrastruc-
tures for the city. Therefore, it is necessary to increase the performance of weather
forecast systems in order to prevent undesirable problems such as flood.

3 Results and discussion

Data from the training part were first used to train the constructed NANN model.
After training, the network was next applied for forecasting the precipitation values
and compared with data from the testing part. Statistical parameters such as R2,
RMSE, MAE, mean and standard deviation of error were also highlighted in order to
explore the performance of the NANN model in each step. All values of the men-
tioned above parameters were listed in Table 1. Fig 2 presents the output of the
NANN model versus the measured precipitation data while Fig 3a presents the distri-
bution of error for both training and testing parts. As indicated in Table 1, for the
training part, R2=0.8981, RMSE=2.4496, MAE=1.3675, and for the testing part,
R2=0.8998, RMSE=3.2909, MAE=1.8672. The numerical model also exhibited a high
capacity in capturing nonlinear information inside the precipitation time-series data
(see also Fig 3b).

Fig 2. Graphs of predicted versus target precipitation values for: (a) the training
part and (b) the testing part. Values of R2 are also highlighted.
Fig 3a presents the histogram of error E=(Predicted-Target) with a resolution of 5
mm. It can be seen that both training and testing part yield an error distribution cen-
tered at zero with a high concentration (i.e. there were almost 80% of population at
zero bin of the histogram). Indeed, the mean and standard deviation values of error
are Emean=0.0200 and -0.4259, Estd=2.4505 and 3.2662, for training and testing part
respectively as indicated in Table 1.
Development of 48-hour Precipitation Forecasting... 1195

Fig 3b shows the measured and predicted precipitation in a time series form. As
indicated in the previous section, the remaining 30% values in the time series were
used as testing data. It means that almost a year and a half precipitation values were
predicted using NANN model with an error mean of -0.4259 mm and a standard devi-
ation of 3.2662 mm. It could be also noticed in Fig 3 that information in both rainy
and no rainy days was fully tracked, even with a high order of stochastic fluctuations.

(b)

Fig 3. Prediction capability of the NANN model: (a) distribution of error between
predicted and target values and (b) time series representation of testing dataset.
Table 1. Summary of prediction capability.
Criteria Training part Testing part
R2 0.8981 0.8998
RMSE 2.4496 3.2909
MAE 1.3675 1.8672
Emean 0.0200 -0.4259
Estd 2.4505 3.2662
4 Conclusion
As reported in the literature, rain is one of the most random phenomenon among all
atmospheric processes. Rainfall exhibits a high order of nonlinearity due to its fluc-
tuation based on various meteorological parameters. In this paper, a Nonlinear Auto-
regressive Neural Network (NANN) model was developed in order to: (i) capture the
nonlinear information inside precipitation time series and (ii) use such nonlinear in-
formation for forecasting rainfall in the future (i.e. 48-hour ahead) for urban areas.
Precipitation data were collected from meteorological stations for a five-year period.
First three years and a haft precipitation data (approximately 70%) were used to train
the NANN model while last year and a haft data (approximately 30%) were used to
test the model. Various statistical criteria such as R2, RMSE, MAE, mean and stand-
ard deviation of error, were proposed in order to quantify the performance of the de-
veloped model. From testing configuration, the results were R2=0.8998,
RMSE=3.2909, MAE=1.8672, Emean=-0.4259 and Estd= 3.2662. This indicated that the
proposed NANN model exhibited a high performance in forecasting rainfall with a
lead-time of 48h. The developed model could provide efficient information for im-
proving the drainage systems in an urban area. However, a comparison between the
developed NANN model and traditional models should be addressed in the future
works.
1196 T.-T. Le et al.

References
1. Pregnolato, M., Ford, A., Robson, C., Glenis, V., Barr, S., Dawson, R.: Assessing urban strategies for
reducing the impacts of extreme weather on infrastructure networks. R Soc Open Sci. 3, (2016).
doi:10.1098/rsos.160023
2. Faccini, F., Luino, F., Paliaga, G., Sacchini, A., Turconi, L., de Jong, C.: Role of rainfall intensity and
urban sprawl in the 2014 flash flood in Genoa City, Bisagno catchment (Liguria, Italy). Applied Ge-
ography. 98, 224±241 (2018). doi:10.1016/j.apgeog.2018.07.022
3. Merz, B., Kreibich, H., Thieken, A., Schmidtke, R.: Estimation uncertainty of direct monetary flood
damage to buildings. Natural Hazards and Earth System Sciences. 4, 153±163 (2004).
doi:https://doi.org/10.5194/nhess-4-153-2004
4. Mba, L., Meukam, P., Kemajou, A.: Application of artificial neural network for predicting hourly
indoor air temperature and relative humidity in modern building in humid region. Energy and Build-
ings. 121, 32±42 (2016). doi:10.1016/j.enbuild.2016.03.046
5. Villarini, G., Seo, B.-C., Serinaldi, F., Krajewski, W.F.: Spatial and temporal modeling of radar
rainfall uncertainties. Atmospheric Research. 135±136, 91±101 (2014).
doi:10.1016/j.atmosres.2013.09.007
6. Valverde Ramírez, M.C., de Campos Velho, H.F., Ferreira, N.J.: Artificial neural network technique
for rainfall forecasting applied to the São Paulo region. Journal of Hydrology. 301, 146±162 (2005).
doi:10.1016/j.jhydrol.2004.06.028
7. Wu, C.L., Chau, K.W., Fan, C.: Prediction of rainfall time series using modular artificial neural
networks coupled with data-preprocessing techniques. Journal of Hydrology. 389, 146±167 (2010)
8. Panapakidis, I.P., Dagoumas, A.S.: Day-ahead electricity price forecasting via the application of
artificial neural network based models. Applied Energy. 172, 132±151 (2016).
doi:10.1016/j.apenergy.2016.03.089
9. Doucoure, B., Agbossou, K., Cardenas, A.: Time series prediction using artificial wavelet neural
network and multi-resolution analysis: Application to wind speed data. Renewable Energy. 92, 202±
211 (2016). doi:10.1016/j.renene.2016.02.003
10. Islam, M.P., Morimoto, T.: Non-linear autoregressive neural network approach for inside air tempera-
ture prediction of a pillar cooler. International Journal of Green Energy. 14, 141±149 (2017).
doi:10.1080/15435075.2016.1251925
11. Pereira, F.H., Bezerra, F.E., Junior, S., Santos, J., Chabu, I., Souza, G.F.M. de, Micerino, F., Nabeta,
S.I.: Nonlinear Autoregressive Neural Network Models for Prediction of Transformer Oil-Dissolved
Gas Concentrations. Energies. 11, 1691 (2018). doi:10.3390/en11071691
12. Potdar, K., Kinnerkar, R.: A non-linear autoregressive neural network model for forecasting Indian
index of industrial production. In: 2017 IEEE Region 10 Symposium (TENSYMP). pp. 1±5 (2017)
13. Hung, N.Q., Babel, M.S., Weesakul, S., Tripathi, N.K.: An artificial neural network model for rainfall
forecasting in Bangkok, Thailand. Hydrology and Earth System Sciences. 13, 1413±1425 (2009)
14. Billings, S.A.: Nonlinear System Identification: NARMAX Methods in the Time, Frequency, and
Spatio-Temporal Domains. John Wiley & Sons (2013)
15. Boussaada, Z., Curea, O., Remaci, A., Camblong, H., Mrabet Bellaaj, N.: A Nonlinear Autoregres-
sive Exogenous (NARX) Neural Network Model for the Prediction of the Daily Direct Solar Radia-
tion. Energies. 11, 620 (2018). doi:10.3390/en11030620
16. Duncan, A.P., Chen, A.S., Keedwell, E.C., Djordjevic, S., Savic, D.A.: Urban flood prediction in
real-time from weather radar and rainfall data using artificial neural networks. Weather Radar and
Hydrology. 351, 568±573 (2012)
17. Marquardt, D.: An Algorithm for Least-Squares Estimation of Nonlinear Parameters. Journal of the
Society for Industrial and Applied Mathematics. 11, 431±441 (1963). doi:10.1137/0111030
18. Willmott, C.J., Matsuura, K.: Advantages of the mean absolute error (MAE) over the root mean
square error (RMSE) in assessing average model performance. Climate Research. 30, 79±82 (2005).
doi:10.3354/cr030079
19. Salcedo-Sanz, S., Deo, R.C., Carro-Calvo, L., Saavedra-Moreno, B.: Monthly prediction of air tem-
perature in Australia and New Zealand with machine learning algorithms. Theoretical and applied
climatology. 125, 13±25 (2016)
20. Sharma, A., Goyal, M.K.: Bayesian network model for monthly rainfall forecast. In: 2015 IEEE
International Conference on Research in Computational Intelligence and Communication Networks
(ICRCICN). pp. 241±246. IEEE (2015)
 Part VII
Building Information Modeling
 Building Information Modeling Based Optimization Of
Steel Single-Plate Shear Connections Using Differential
Evolution Algorithm

Tran-Hieu Nguyen1 and Anh-Tuan Vu1

1 National University of Civil Engineering, 55 Giai Phong Road, Hanoi, Vietnam
hieunt2@nuce.edu.vn

Abstract. The conventional method that steel connections are still manually de-
signed one by one is time-consuming. In this paper, a building information
modeling based framework is proposed for automated generating steel connec-
tions. The differential evolution algorithm is used to seek the optimal solution
of connection where the design variables are bolt diameter, bolts arrangement
as well as plate dimensions. The objective function is the minimum cost of steel
connection while the design constraints adopt the specification AISI/AISC 360.
The optimized steel connection is then automatically created in the BIM model.
A toolkit is developed to implement the proposed framework on Tekla Struc-
tures through open application programming interface (oAPI). The toolkit is
applied for single-plate shear connections, which are the most popular connec-
tion type in steel buildings. An example of a steel frame is conducted to demon-
strate the effectiveness of the proposed framework. The result of the example
shows that the proposed framework can shorten the design time.

Keywords: building information modeling, optimization, steel connection, sin-

gle-plate shear connection, differential evolution.

1 Introduction

The steel structure design is a complex process including some steps as structural
layout, structural analysis, member design, connection design, detailing, drawing and
cost estimation. Although the popular structural analysis programs such as CSI
SAP2000, Bentley StaadPro or Autodesk Robot Structural Analysis Professional are
often integrated the member design function, the capability to design steel connec-
tions of those programs is still limited. Steel connections are designed either by
spreadsheets or by software. Using spreadsheets requires much human effort to man-
ually input design data as section properties, design forces, bolt size, bolts arrange-
ment, etc. This task takes a long time and sometimes leads to mistakes. Some steel
connection design programs such as RAM Connection or RISA Connection allow
importing data from the finite element analysis model but still requires to customize
many parameters in the program interface. In both ways, steel connections shall be
designed one by one and consequently, leading to time-consuming. Therefore, full

© Springer Nature Singapore Pte Ltd. 2020 1199

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_192
1200 T.-H. Nguyen and A.-T. Vu

automation and optimization are needed in order to obtain a steel connection design
with the minimum cost in a faster manner.
In recent years, the building information modeling (BIM) technology has become
the trend of the architecture, engineering and construction (AEC) industry due to its
advantages. Firstly, BIM represents a building in a 3D format with high accuracy for
better visualization [1]. It helps avoid clashes and design errors. Moreover, the BIM
model is a storage of information that can be easily extracted along the project life
cycle. Because of its benefits, BIM technology is changing the conventional design
process and it will soon replace Computer Aided Design (CAD) technology. Howev-
er, a BIM model with a high level of detail requires a lot of time to build. The ap-
proach to solve this problem is to make modeling process more efficient by creating
tools to automate repetitive tasks.
It can be noted that while the creation of steel connections represents only a portion
of the full modeling process, it is the most time-consuming task. For example, accord-
ing to Sweco Finland, a complex steel structure with 929 members and 696 joints
typically takes about 4,000 hours to design and 50% of time spends in connection
modeling [2]. A tool to automatically generate steel connections in the BIM model
allows to significantly reduce the modeling time. This gives the motivation for the
current study.
In this paper, a framework is developed to simultaneously optimize and automated
generate steel connections in the BIM model. Because the BIM model contains not
only geometry information but also other related information, the data that is needed
for calculation of steel connections such as material properties, end-conditions and
spatial relationships of building elements can be directly extracted from BIM models,
thereby reducing the needs for manual handling, and as a result, avoiding human mis-
takes and being time-saving. The proposed framework is firstly applied to single-plate
shear connections which are the most popular connection type in steel buildings. Oth-
er connection types could be implemented similarly but are out of the scope of the
present paper.
This paper is organized as follows. The optimal design of single-plate shear con-
nections is presented in Section 2. Section 3 proposes the BIM-based framework for
automated generation of optimal connection in the model. An example of a steel
frame is conducted to demonstrate the applicability of the proposed framework in the
practical design. Summary and future work are discussed in Section 4.

2 Optimization of single-plate shear connections

2.1 Optimization problem definition

The steel single-plate shear connection is widely used in steel structures. According to
AISC Manual [3], the single-plate shear connection is classified into the conventional
configuration (Fig. 1a) and the extended configuration (Fig. 1b), in which the extend-
ed single-plate connection is preferred in the practice due to their ease of fabrication
and erection. In some cases, the steel plate is lengthened vertically to prevent the
Building Information Modeling Based Optimization... 1201

buckling of the plate or the fracture of the weld. This kind of connection, called the
full depth stiffened extended single-plate connection (Fig. 1c), is used in this study.

a a
(a) (b) e (c) e

Fig. 1. Single-plate connections: (a) conventional, (b) extended, (c) full depth stiffened.

Six parameters of the single-plate connection are considered as design variables in the
optimization problem including bolt diameter, number of bolt columns, number of
bolt rows, bolt pitch, bolt gage and plate thickness. The design pool of variables is
prepared as described in Table 1. Parameters such as bolt grade, weld electrode, the
edge distance are considered as design constants.

Table 1. Design pool of design variables

Design variable Notation Design pool

Bolt diameter db [M16; M20; M22; M24; M27]
Number of bolt rows nr [2; 3; 4; 5; 6; 7; 8; 9; 10; 11; 12]
Number of bolt columns nc [1; 2; 3; 4]
Bolt pitch p 40y120mm
Bolt gage g 40y120mm
Single-plate thickness tp [5; 6; 8; 10; 12; 14; 16; 18; 20; 22; 25; 28; 32]
The cost function is defined as follows:

C Wp u C p  Wb u Cb (1)

where Wp is the weight of the single-plate [kg]; Cp is the price of the structural steel
plate [USD/kg]; Wb is the weight of bolts [kg] and Cb is the price of bolts [USD/kg].
For the extended single-plate connections, it is necessary to consider the effects of
the bending moment due to the eccentricity of the shear force. The design procedure
of unstiffened extended single-plate shear connections is proposed in [4]. Mechanical-
ly, the full depth stiffened extended single-plate connection is much stronger than the
others. Therefore, it could be conservatively applied this procedure to the full depth
stiffened extended single-plate connection. The design constraint is such that the con-
nection must be satisfied six limit states as follows:
1. Checking of the bolt shear strength (AISC360 [5] - Sect. J3.6) and the bolt hole
bearing strength (AISC360 - Sect. J3.10) taking into account the eccentricity e.
1202 T.-H. Nguyen and A.-T. Vu

2. Checking of the ductility of the plate. This condition ensures that the plate must
yield without rupturing the bolts or the welds.
3. Checking of the plate shear yielding strength, shear rupture strength (AISC360 -
Sect. J4.2) and block shear strength (AISC360 - Sect. J4.3).
4. Checking of the plate flexural strength based on the von-Mises criterion [3].
5. Check the plate for buckling as a double coped beam [6].
6. Sizing the weld leg size.

2.2 Optimization algorithm

The differential evolution (DE) algorithm introduced by K. Price and R. Storn [7] is
selected to solve this optimization problem. Compared to other evolutionary algo-
rithms, DE is much more simple and straightforward to implement [8]. The Pseudo
code of the algorithm is presented in Fig. 2.

Fig. 2. Pseudo code of the DE algorithm

30
Cost (USD)

25

20

15
0 10 20 30 40 50
Generation

Fig. 3. Full history of the optimization process

A connection between a supported beam W12X35 and a supporting girder W18X65 is

selected as a numerical example to illustrate the capability of DE. Materials used in
this connection are as follows: beam grade ASTM A992, plate grade ASTM A572
Building Information Modeling Based Optimization... 1203

Gr.50, bolt grade A325ST/N and weld electrode E70XX. The connection is designed
with the shear force that equals to 40% of the shear strength of the supported beam.
The main parameters of DE are taken as NP=60; CR=0.8; F=0.7 and
MAXGEN=50. The full history of the optimization process during 50 generations is
plotted in Fig. 3. It could be observed that the optimal solution is found in the 33 th
generation. The configuration of the optimal connection is: db=M20; nr=04; nc=02;
p=65mm; g=55mm and tp=10mm (Fig. 5b).

3 Proposed BIM-based framework

BIM Data Extraction Design Code Detailing DE Optimization
Requirements Requirements
Geometry Data
Optimal
DE optimization
connection
operator
Material Data configuration

BIM model
x Bolt diameter;
Support Conditions; x Bolt arrangement;
End Reactions x Plate thickness

Connection design data

(Bolt grade, Electrode, Steel Grade, Design pool)

Update BIM Model

Fig. 4. The proposed BIM-based framework

The proposed BIM-based framework is composed of two modules: BIM-based data

extraction module and DE-based optimization module as presented in Fig. 4. The
BIM software used in this study is Trimble Tekla Structures (TS). TS provides a use-
ful WRROFDOOHG³FRPSRQHQW´WRTXLFNO\FRQQHFWSDUWVLQWKHPRGHO ,QWKLVVWXG\WKH
FRPSRQHQWQXPEHU³)XOO'HSWK6´is used to create the full depth extended sin-
gle-plate connection. In addition, the properties of a comSRQHQWFDQEHVDYHGLQ³DW
WULEXWH´DQGUHXVHd in other projects.
A .NET toolkit is developed to implement the proposed framework. The toolkit
firstly connects to the TS model through open application programming interface
(oAPI) and detects all possible connection positions. All needed information such as
the geometry data, the material data, the support conditions as well as the end reac-
tions is then extracted from the BIM model. Based on the received data, the DE algo-
rithm is used to seek the optimal configuration of connections. The optimal configura-
tions are VDYHGLQDWWULEXWHVQDPHG³%%B$B%M´LQZKLFK$LVWKHSURILOHRIWKH
supported beam; B is the profile of the supporting girder. Finally, the connections are
automatically created in the corresponding positions in the BIM model. The devel-
oped toolkit is implemented on a steel structure frame as illustrated in Fig. 5a. It can
be seen that 389 connections were optimized and created in 67s. This example illus-
trates the effectiveness of the proposed framework. The detail of a typical connection
created by this toolkit is presented in Fig. 5b.
1204 T.-H. Nguyen and A.-T. Vu

(a) (b)

Fig. 5. (a) Output of the developed toolkit, (b) Detail of a typical connection

4 Conclusion

This paper presents a BIM-based framework for optimization and automated genera-
tion of steel connections. The framework considers the bolt diameter, the bolt ar-
rangement and the plate thickness as variables and solves the optimization problem
using the differential evolution algorithm. A toolkit is developed to implement the
proposed framework in the BIM software Tekla Structures. The illustrated example
shows that the framework greatly reduces the design time and archives economic
benefits. However, the toolkit is just developed to the single-plate shear connection.
Extending the toolkit for remaining connection types such as end plate connection,
double angle shear connection, etc. will be considered in the future work.

References
1. C. Eastman, P. Teicholz, R. Sacks, K. Liston: BIM Handbook: A Guide to Building Infor-
mation Modeling for Owners, Managers, Designers, Engineers and Contractors. 2nd edn.
Wiley, New Jersey, USA (2008).
2. KIRA-digi Homepage, http://www.kiradigi.fi/en/news/sweco-boosts-construction-
productivity-and-quality-with-ai.html, last accessed 2019/05/27
3. AISC: Steel Construction Manual. 15th Edition. American Institute of Steel Construction,
Chicago, Illinois (2016).
4. L. Muir, C. Hewitt: Design of Unstiffened Extended Single-Plate Shear Connection. AISC
Engineering Journal Second Quarter, 67±79 (2009).
5. AISC: AISI/AISC 360-16 Specification for Structural Steel Buildings. American Institute
of Steel Construction, Chicago, Illinois (2016).
6. L. Muir, W. Thornton: A Direct Method for Obtaining the Plate Buckling Coefficient for
Double Coped Beams. AISC Engineering Journal Third Quarter, 133±134 (2004).
7. K. Price, R. Storn, J. Lampinen: Differential evolution: a practical approach to global op-
timization. 1st Edition. Springer, Berlin, Germany (2005).
8. J. Vesterstrom, R. Thomsen: A comparative study of differential evolution, particle swarm
optimization, and evolutionary algorithms on numerical benchmark problems. Proceedings
of the 2004 Congress on Evolutionary Computation. IEEE, Portland, USA (2004).
 BIM application for the design consultant on the irriga-
tion and hydropower projects in Vietnam

Van Toan TRAN1, Huu Hue NGUYEN1, Tuan Long CHU1, Lam Tuan MAI1, Van
Do NGUYEN1, Dinh Tai NGUYEN1 and The Minh Ngoc VU2
1
Thuy Loi University, 175 Tay Son, Dong Da, Ha Noi, Vietnam
2
Vietnam Hydraulic Engineering Consultants No1
tranvantoan@tlu.edu.vn

Abstract. In the recent years, the construction industry has inherited great
achievements from the rapid development of science and technology to improve
labor productivity, increase the scientific quality, reduce risks and increase the
efficiency of construction investment. Especially BIM applications are becom-
ing an inevitable trend of the construction industry in the country thanks to the
advantages of science, technology, economics and benefits in construction
management. Vietnam is a developing country, so the construction industry has
been and will continue to play a very important role in the country. Therefore,
the application of BIM for the construction industry is also indispensable to re-
duce the risk and increase the efficiency of construction investment. However,
the BIM application for the construction of irrigation and hydropower projects
is still limited by the peculiarity and difficulty of human resource understanding
on BIM technology. This paper presents some research results on application of
BIM technology for the design consultant on the irrigation and hydropower pro-
jects in Vietnam.

Keywords: BIM, design consultant, irrigation project, hydropower project.

1 Introduction

The definite BIM had existed since 1970 decade [1]. 7KHWHUP³%XLOGLQJLQIRUPDWLRQ
PRGHO´ ILUVW DSSHDUHd at 1992 in the documents of G.A van Nederveem and
F.P.Tolman [2]. However, tKH WHUP ³%XLOGLQJ LQIRUPDWLRQ PRGHO´ DQG ³%XLOGLQJ
LQIRUPDWLRQPRGHOLQJ´ KDVQRWEHHQ widely used until 10 years later, when in 2002,
Autodesk ODXQFKHGDERRNZLWKWKHWLWOH³%XLOGLQJLQIRUPDWLRQPRGHOLQJ´ DQGRWKHU
software vendors also begun to confirm their interest in the this area. Based on contri-
bution from Autodesk, Bentlry Systems and Graphisoft, addition observation with the
industry, in 2003 Jerry Laiserin has helped popularize and standardize this term as a
FRPPRQQDPHIRU³WKH'LJWLWDOVLPXODWLRQRIWKHSURFHVVRIEXLOGLQJDFRQVWUXFWLRQ´
[3]. In Vietnam, the Prime Minister of the Socialist Republic has issued a decision of
2500 which requires the application of BIM technology in construction management
activities since 2018 [4].
Application of BIM technology has many advantages to design consultants. First, the
projects will be simulated visual 3D which will facilitate for presentation, evaluation,
selection effective design solution. Second, BIM application will contribute to in-
© Springer Nature Singapore Pte Ltd. 2020 1205
C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_193
1206 V.T.Tran et al.

creasing productivity and quality of design documents, easily adjusting design solu-
tions and limiting errors in the design process. Third, the volume components and 2D
drawing profiles of the project will be extracted from automatic or semi-automatic 3D
models that help to estimate construction costs quickly and accurately. Fourth, struc-
tural analysis, energy analysis and architectural analysis ... become easier when ex-
ploiting the available 3D model. Fifth, BIM application for consulting design compa-
nies will gradually create a working group style, contributing to establishing a profes-
sional and modern working environment to easily integrate with the world. Final, the
BIM application will have to exchange data through cloud computing technology to
help geographically different design teams easily exchange data and store data secure-
ly.
In Vietnam, Application of BIM technology for civil construction and transportation
consultancy has been popular. However, the application of BIM to consulting on con-
struction of irrigation and hydropower projects is still limited by the specificity, lim-
ited understanding of human resources and limited government guidance. Moreover,
in order to apply BIM in design consulting, consultants need to know how to use var-
ious 3D modeling software and know how to combine BIM management software to
create total design products well suited.

2 Application process of BIM

2.1 The current status of BIM application for irrigation project in Vietnam

Currently, the government has requested a number of BIM application on the irriga-
tion and hydropower construction projects such as: Ngoi Gianh reservoir at Phu Tho
province, Tra Khuc reservoir at Quang Ngai province, Canh Tang reservoir at Hoa
Binh province, Ban Lai reservoir at Lang Son province, etc [5]. However, they are
not applied in the design phase but during the project implementation phase. There-
fore, the application of staples is mainly to re-examine the design documents (2D
drawings) and create the construction completion model to serve the management and
maintenance of the works later. This leads to not only not optimizing the design work
for design consultants but also increasing the cost of consulting. And of course the
investors is still undesirable implementing BIM applications for projects that they
manage. So far, only one irrigation reservoir is implementing the application of BIM
as Canh Tang reservoir at Hoa Binh province.
2.2 The content needs to be implemented by design consultants when applying
BIM technology

The main types of works in irrigation and hydropower projects are earth dams, gravity
concrete dams, culverts, spillways, canal systems, hydropower plants, pumping sta-
tions, siphons, pressure pipes, tunnels, ... They often lie on terrain with different ele-
vations and intersect with different geological layers. Design consultants also need to
perform basic tasks in accordance with the construction law but the design product
must be a 3D model. The 3D models of each construction structure must be divided
into sections according to the construction norms issued by the Ministry of Construc-
BIM Adoption in Construction Projects Funded... 1207

tion and be created according to strict procedures to ensure the next steps can be ex-
ploited. Then, if there is any work to adjust the design solution, automatically change
the information of the design files attached as shown Fig. 3 and Fig. 4. A project that
has applied BIM in the design step will create premise for BIM application for the
next steps in the process of implementing the investment.
2.3 Process of BIM
The design consultant on the irrigation and hydropower projects in Vietnam which
have own characteristic, so let that is useful, we need to apply BIM for design from
the beginning project. In Vietnam, according to the current construction law [6], irri-
gation and hydropower projects often undergo 3 basic design steps as follows:
x Step 1: Design concept;
x Step 2: Technical design;
x Step 3: Technical design of construction drawings.
Design concept
When applying BIM technology to the preliminary design, the design teams can fol-
low the procedure in Fig. 1. The designer will exploit topographical survey data in 3D
model from GIS (Bing maps, Google Earth). After that, the designers proceed the
alignment of the works, architectural design, choosing the type of structure and de-
signing the most basic parameters of the project in 3D model (Infraworks, Civil 3D,
Revit, Tekla). From there, they perform operations on the 3D model to output the
main building volume and drawings automatically. Next, they set up the method and
the order of construction of each work item. After that, they calculate the cost of con-
struction. Furthermore, in this step conflict checking is carried out to correct the pro-
file design (Navisworks, Navigator). At the same time, they also illustrate the struc-
tural and architectural solutions by the 3D movies for reporting (Infraworks, Lumion,
3DS Max). Finally, they will summarize and report the solutions to the investor so
that the investor can choose a reasonable solution and issue approval decision.
Technical design

Fig. 1. The BIM application process for prelim- Fig. 2. The BIM application process for
inary design technical design

When the investor has chosen the solution, in order to be able to technical design for
the construction, we need to investigate the actual terrain. When the survey data is
correct, the design solution will be updated (Civil 3D) (see Fig. 2). After that, the
1208 V.T.Tran et al.

project 3D model will be designed in detail, followed by design of construction meth-

od (Civil 3D, Revit, Tekla). Once completed, the drawing file and amount of material
of construction are automatic publish and make rate of progress (Navisworks,
Synchro, Navigator, Vico control). Furthermore, in this step conflict checking is car-
ried out to correct the profile design (Navisworks, Navigator). The estimation is done,
this work indicates that the investor knows exactly how much money he will spend
when implementing the construction plan. The work results from ³6XUYH\GDWDIURP
*,6´ update to ³&RVW(stimate´ will be synthesized and reported in 3D film there by
bringing the most intuitive look for investors (Infraworks, Lumion, 3DS Max). Final,
there is a suitable choice.
Technical design of construction drawings
In this step, the project will be designed with construction methods and construction
sites more detailed (Civil 3D, Revit, Tekla, Infraworks, Navisworks, Navigator).
2.4 Specific characteristics of the design of irrigation and hydropower
projects

In terms of structures, culverts, canal systems, hydropower plants, pumping stations,

siphons, pressure pipes, tunnels, ... are not much different from civil and transport
structures. However, dams and spillways have many different characteristics. There-
fore, Therefore, the author will focus on describing the content of applying BIM tech-
nology in the design consultant for them.
Dam design

Fig. 3. Application of BIM technology for dam design consultancy

Dams are often divided into different blocks according to the type chosen or construc-
tion method. The dam foundation is often placed across the riverbed, on good geolog-
ical layers. Therefore, before dam construction, the dam foundation pit must be cut
through different geological layers. In order to accurately simulate the 3D shape of
the dam foundation, the 3D terrain simulation must be done first. Then the simulation
BIM Adoption in Construction Projects Funded... 1209

of 3D geology is very difficult. In addition, the dam foundation pit may also have to
be partially excavated depending on the solution of divert water flow over construc-
tion time. Next, the dam is also partially constructed depending on the solution of
divert water flow over construction time. Final, automation of volumes output and 2D
drawings can be done accurately.
Application of BIM for dam design consultancy is carried out according to the proce-
dure shown in Fig. 1 and Fig. 2. Accordingly, the contents to be performed in turn
are: simulate 3D terrain, 3D geology; calculate dam top elevation; select dam type,
divide 3D dam blocks; select dam cross section parameters; 3D dam details, seepage
analysis, stable analysis, 3D foundation pit; design 3D measurement and monitoring
system, design 3D electromechanical systems (if necessary), select 3D solution of
divert water flow; select 3D solutions and construction sequence; simulate 3D water
compartment and drainage works; export volumes and 2D drawings; set 3D construc-
tion progress, arrange 3D construction sites; estimate costs; illustrate project by film
and 3D images. After each design step, the parameters of the dam cross section can be
adjusted to be reasonable. The design consulting works are shown very visually (see
Fig. 3).
Spillways design

Fig. 4. Application of BIM technology for spillway design consultancy

Spillways can be one or more blocks according to the type chosen. The spillway
foundation is often on good geological layers. Therefore, before spillway construc-
tion, the spillway foundation pit must be cut through different geological layers. In
order to accurately simulate the 3D shape of the spillway foundation, the 3D terrain
and 3D geology simulation must be often done when designing dams. Next, the spill-
way is also partially constructed depending on the solution of divert water flow over
construction time. Final, automation of volumes output and 2D drawings can be done
accurately.
Application of BIM for spillway design consultancy is carried out according to the
procedure shown in Fig. 1 and Fig. 2. Accordingly, the contents to be performed in
1210 V.T.Tran et al.

turn are: select spillway type, calculate spillway threshold elevation, divide 3D spill-
way blocks (if necessary); select spillway cross section parameters; 3D spillway de-
tails, seepage analysis, stable analysis, design 3D mechanical parts and select equip-
ment (if necessary), 3D foundation pit; design 3D measurement and monitoring sys-
tem, design 3D electromechanical systems (if necessary), select 3D solutions and
construction sequence; simulate 3D water compartment and drainage works (if neces-
sary); export volumes and 2D drawings; set construction progress, arrange 3D con-
struction sites; estimate costs; illustrate project by film and 3D images. After each
design step, the parameters of the spillway cross section can be adjusted to be reason-
able. The design consulting works are also shown adjusted accordingly (see Fig. 4).

3 Conclusion

Research on application of BIM technology for the design consultant on the irrigation
and hydropower projects in Vietnam, the author found that:
BIM for design consultants on the construction projects has its own peculiarity, but
generally there are also many advantages to easily apply.
Design consultancy with BIM contributes to increasing the visualization and com-
pleteness of the project profile by the 3D model representation. The exploitation of
the 3D model designed to facilitate the next steps will significantly increase the
productivity and quality of design documents.
In order to apply BIM widely for the design consultant on the irrigation and hydro-
power projects in Vietnam, the government should issue legal guidelines and applica-
tion roadmap which are visually clear.

References

1. Simon Ruffle, Architectural design exposed: from computer-aided-drawing to computer-

aided-design, Environments and Planning B: Planning and Design, Los Angeles, USA, 1985.
2. G.A. van Nederveen and F.P. Tolman, Modelling multiple views on buildings, Automation
in Construction, Delft, The Netherlands, 1992.
3. Chuck Eastman, Paul Tiecholz, Rafael Sacks, Kathleen Liston, BIM Handbook: A Guide to
Building Information Modeling for Owners, Managers, Designers, Engineers and
Contractors (Second Edition), John Wiley and Sons, Inc., Hoboken, New Jersey, 2011.
4. 3ULPH 0LQLVWHU RI WKH 6RFLDOLVW 5HSXEOLF RI 9LHW 1DP 'HFLVLRQ 1R 4Ĉ-TTg on
Approving the Scheme on application of construction information model (BIM) in
construction and operation management activities, Hanoi: Government of the Socialist
Republic of Viet Nam, 2016.
5. Ministry of Agriculture & Rural Development, Profile of survey and design of the Ban Lai,
Ngoi Gianh, Tra Khuc, Canh Tang reservoir projects, Hanoi, 2016-2018.
6. Vietnam National Assembly, The Construction Law of Vietnam No 50/2014/QH13, Hanoi:
Government of the Socialist Republic of Viet Nam, 2014.
 BIM Adoption in Construction Projects Funded
with State-managed Capital in Vietnam:
Legal Issues and Proposed Solutions

Thuy-Ninh Dao 1,2, The-Quan Nguyen 2, Po-Han Chen 1

1
Department of Civil Engineering, National Taiwan University, Taiwan
2
Faculty of Construction Economics and Management, National University of Civ-
il Engineering, Vietnam
d06521023@ntu.edu.tw*, quannt@nuce.edu.vn, pohanchen@ntu.edu.tw

Abstract. It is claimed that Building Information Modelling (BIM) has been

adopted in the construction industry in Vietnam since early 2000. However, in
practice, this proves not to be the case, particularly in the context of construc-
tion projects funded with state-managed capital. Considering that such funding
accounts for the largest market share in construction funded projects in Vi-
etnam, further investigation is warranted. A preliminary survey reveals that the
owners and other stakeholders of these projects are cautious of BIM adoption,
primarily due to avoiding legal issues that may arise in such projects. In this
context, this paper aims to explore the potential legal issues of BIM adoption
and propose solutions for the stakeholders. The results will aid in avoiding the
issues outlined, which can aid in promoting BIM adoption in Vietnam. Litera-
ture shows that the legal issues are not only applicable to Vietnam, but to other
countries. To identify the legal issues which arise from BIM adoption in con-
struction projects globally, a through literature review is carried out, where then
a survey is performed with stakeholders of construction projects funded with
state-managed capital in Vietnam. This then provides the opportunity to verify
if the literature findings apply to Vietnam, as well as to discover new issues in a
Vietnamese context. According to the survey results, the biggest legal issues for
BIM adoption in construction projects funded with state-managed capital in Vi-
etnam are; (i) no expenses system for BIM implementation available, so there is
no source of funding for paying consultants/contractors for BIM; (ii) rights and
obligations of stakeholders in BIM projects are not defined clearly in the rele-
vant regulations, so there is a high risk of claims and disputes arising; (iii) no
digital submission system available, thereby BIM models are not accepted, but
hardcopy printouts used, leading to significant work of reviewing and verifica-
tion of as-built documents. Solutions for these issues are generated and validat-
ed with focus groups, which lead to the suggestions of regulation alterations as
well as revisions of contract forms/articles applied in construction projects
funded with state managed capital in Vietnam
Keywords: Building information modeling; BIM Vietnam; construction contract.

© Springer Nature Singapore Pte Ltd. 2020 1211

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_194
1212 T.-N. Dao et al.

1 Introduction

The documented advantages and subsequent widespread acceptance that BIM brings
is widely acknowledged. Over the past decade, numerous construction projects have
been completed using BIM in Finland, Sweden, Norway, Germany, France, Singa-
pore, Australia, among others. Such examples demonstrate the advantages of BIM for
more sustainable projects when compared with non-BIM adopted projects [1]. The
BIM adoption benefits are universally accepted; thus, further justifying developing
countries, and in particular, state-funded projects to also adopted BIM.
Vietnam is a developing country that is proactively trying to apply technical appli-
cations and processes into the architecture, engineering and construction (AEC) in-
dustry. In Vietnam, BIM has been adopted in the construction industry since early
2000, but it is still not applied widely. This is particularly the case in construction
projects funded with state-managed capital, which accounts for the largest market
share of construction projects in Vietnam. In comparison with neighboring nations
and other similarly developed countries, it appears the acceptance levels are lower
than first anticipated. This is further compounded by the lack of standards and guide-
lines to further aid this development and realization within the Vietnamese market.
Other BIM adoption and implementation barriers in Vietnam are: (i) no govern-
ment published policies for applying BIM in Vietnam; and (ii) where BIM is referred
to, such references are scant , but not widely applied. From a stakeholders perspec-
tive, the barriers include a need for well-define business process models, for practical
information integration strategies for tools that are used by the industry and computa-
ble model-based digital design data. This is the organizational culture in Vietnam,
where the Government policy enacted takes precedence, whereby other authorities
and construction project stakeholders will follow.
In Vietnam, all construction projects funded from capital derived from loans,
bonds and funds are managed by the state in investment, community safety, national
defense and security and efficiency. The investor shall take responsibility for manag-
ing the project as prescribed in the Vietnam construction legal system [2]. Hence,
having a BIM regulation system for construction stakeholders and authorities can
adopt and implementation BIM in construction projects is essential. Therefore, the
scope of this study focuses on state-funded projects only.
It is acknowledged that new software brings widespread benefits, but also disad-
vantages. BIM adoption is no exception. This is particularly the case where the role
and integration of each participant in the BIM model is complicated. This factor also
relates to the laws implemented in datacentres, where there are major concerns
surrounding the security of the data, especially in the case of sensitive projects [4].
BIM presents a new paradigm within the AEC industry, one which encourages the
integration of information exchange between all stakeholders in a project. It is also
considered a barrier for owners and construction firms applying BIM. This therefore
leads to the following; how to increase the level of BIM implementation in the Viet-
BIM Adoption in Construction Projects Funded... 1213

namese construction industry and subsequently, how to systematically overcome the

challenges in such adoption.
Founded on a literate review, issues and the barriers in BIM adoption are identified
and documented. Susequently, a survey is conducted investigating these issues in the
context of the Vietnamese law system and the AEC industry as a whole. By investi-
gating the the reason for such slow progressive changes in BIM adoption in Vietnam,
it is possible to then address and provide solutions as to how to gain momentum in
developing BIM implementation in Vietnam. The results generated are validated us-
ing focus groups, resulting in suggestions for regulation alterations, in addition to
revisions in contract forms/articles applied in construction projects funded with state-
managed capital in Vietnam. It is envisaged that any progress in Vietnam will need to
base on such Government regulations. [2]

2 Methodology

The research aim is to identify factors promoting BIM adoption and implementation
in Vietnamese construction projects funded by state-managed capital. Focus is on the
current situation and proposals on how to promote BIM adoption in Vietnam.
Considering world-wide barriers to BIM adoption and more particularly, in Vi-
etnam, a detailed desk-based investigation via Web of Science, Science Direct,
Google Scholar, and Scopus publications between 2008 and 2018 is conducted. The
papers identified are screened and checked for quality and eligibility, to elicit the
information related to the legal issue, BIM adoption barriers, and subsequent govern-
ment policies. Among 56 papers identified, including journal papers and international
conference proceedings, the issues that consider the Vietnamese legal system and
social culture are identified. The subsequent factors are then used to create a ques-
tionnaire survey for circulation, to gain a further and more detailed insight.
Subsequent to this, the BIM adoption barriers and legal issues are surveyed, with
suggested solutions generated and validated using focus groups. The results lead to
the suggestions of regulation alterations, as well as revisions of contract forms/articles
applied in construction projects funded with state-managed capital in Vietnam.
The research methodology is summarized in the figure below:

Fig. 1. Research framework

 1214 T.-N. Dao et al.

3 Literature review

Of the 56 publications identified, these were analyzed. The articles were identified
based on referencing the legal barrier in BIM adoption in the construction industry.
The lack of BIM-related aspects and tangible evidence of application in the construc-
tion industry and authorities, lead to the failure of BIM adoption and implementation.
Figure 2 shows the publications by year published, focusing on the last decade.

Fig. 2. Journal papers and conference proceedings published by year

As highlighted, one of the reasons that OHDGWRVWDNHKROGHUV¶DQ[LHW\LV³WKHXVHRI

BIM in a project raises important contractual issues relating to project responsibilities
and risks, contractual indemnities, copyright, and the use of documents that are not
addressed by the standard industry cRQWUDFWIRUPV´[3]. In many cases, the contractors,
and the contracting authorities do not have enough adequate specialize personnel or
the legal competences to create a contract that would fully ensure their interests. Fur-
ther to this, the BIM legal issues also referenced in previous studies are listed in Table
1 below:

Table 1. Legal issues in BIM adoption and implementation

Number of papers that
BIM legal issues
mentioned
Contract issues 18
Intellectual property rights, copy rights 25
Responsible of technical risk, model mistakes 23
Security and access rights 9
Lack of software compatibility 5
Lack of BIM standards, consensus on protection of information in
9
conversion and interoperability
Unclear legal liability 8
Lack of contract forms to clearly mandate the BIM practices and
7
address legal concerns.
Data inconsistency and compatibility 20
BIM Adoption in Construction Projects Funded... 1215

The legal issues above are also considered in the barriers of BIM adoption and im-
plementation. Some of these factors have the potential to lead to disputes among
stakeholders.

4 Survey results

This study covers the barriers that lead to the slow adoption and implementation of
BIM in the Vietnames construction industry, and in particular, on state-funded pro-
jects. After considering a systematic approach of BIM implementation in other coun-
tries, a questionnaire survey was conducted, the respondents are people who work in
nation organizations such as Ministry of Construction, Construction companies, BIM
consultant firms, among others. The questionnaire-based survey was issued to 115
recipients, all of which completed and returned the survey. The results are collated
and evaluated to address the research objectives. Fig. 3 and Fig. 4 below show the
subsequent data:

Fig. 3. Percentage of respondents Fig. 4. Experiment years percentage of

respondents

The majority of respondents highly agree that the crucial factors for the reduced
label of BIM adoption on state-funded construction project in Vietnam are listed in
table below:

Table 2. The respond of legal issues surrounding BIM

No. Legal issues surrounding BIM Agree (%)

1 Lack of BIM regulations 91
No correlation of the systems leading to the difficulty of the authori-
2 94.7
ties to inspect / supervise
3 Lack of consistency and participation of stakeholders 95.7
Lack of contract forms to clearly mandate the BIM practices and
4 89.3
address legal concerns
Liability exposures to design errors, non-compliant design, transi-
5 82.2
tion errors, loss of data or data misuse
6 Lack of State BIM standards 89.2
7 Model ownership and Intellectual Property Rights 74.2
8 ,QIULQJHPHQWRIDQRWKHU¶V,QWHOOHFWXDO3URSHUW\5LJKWV 75.9
1216 T.-N. Dao et al.

9 Roles involving coordinating maintaining and controlling the model 73.3

10 Rights of owners to change the design 74.1
11 Unclear rights and responsibilities 73.2

In comparison with BIM adoption legal issues that collected from previous publi-
cations, Vietnam construction industry has similar barriers. Besides, there is a clear
necessity for having BIM management training course for government officials, to
further address these issues. Based on actuality of the construction industry in Vi-
etnam, most of the respondents recommend that having the concrete regulations in
BIM implementation into state construction projects is necessary. Following is the
important role of software purchase, digital submission system budget and BIM man-
agement training course for government officials.

From the answers responds, the top solutions for the actual BIM adoption barriers
in Vietnam construction market are: the necessary of having the concrete regulation in
BIM implementation in to State construction projects; the budget for software pur-
chase, digital submission system and the BIM management training course for gov-
ernment officials in particular.

5 Discussion and conclusion

The study from countries that have successfully implemented BIM highlight that,
government play an important role in leading the process of applying BIM, giving
strategy, roadmap and setting goals for the construction industry. Thereby then, the
BIM contract forms need to be suggested to BIM participants, who then can clearly
understand their own rights and responsibilities during BIM project in order to avoid
the risks that may happen (Table1). In the Vietnam construction industry, its prerequi-
site is that the state, the government and relevant ministries need to enact specific
regulation to obligatory construction projects adopting BIM. In addition, BIM regula-
tions, BIM standards, and BIM contract forms are required to clearly mandate, rights
and responsibilities of BIM participants. This will thereby mitigate and potentially
remove, the barriers of BIM adoption, particularly in the context of state-funded con-
struction projects in Vietnam.

References
1. Arayici, Y.; Khosrowshahi, F. Roadmap for implementation of BIM in the UK
construction industry. Eng, Const and Arch Man 2012, 19, 610±635.
2. VIETNAM DECREE ON CONSTRUCTION PROJECT MANAGEMENT.
3. Porwal, A.; Hewage, K.N. Building Information Modeling (BIM) partnering
framework for public construction projects. Automation in Construction 2013,
31, 204±214.
 BIM-based innovative bridge maintenance system using
augmented reality technology

NgocSon Dang1[0000-0002-3341-3034], ChangSu Shim2

1 Ph.D. student, ChungAng University, Seoul, South Korea
2
Professor, ChungAng University, Seoul, South Korea

Abstract. A smooth transportation system plays a pivotal role in the development

of a country since it enables rapid connect among industrial zones or territories.
Especially bridges, which commonly has around 50 years’ service life, needs to
be paid more attention to maintenance to keep it in a good service state. Nowa-
days most bridge has its own bridge maintenance system (BMS) but still bridge
engineer is challenged by difficulty in term of store the damage records and repair
history, as well as assess the current behavior of the structural system. This paper
proposed an innovative BMS which using a schematic BIM-based information
management system, collaboration with an automate inspection task using aug-
mented reality (AR) device. According to preventive maintenance strategy, a data
schema for BIM model was investigated. An integrated digital model is created
to store, manipulate and share the inspection data and maintenance history. On
another hand, the onsite inspection task is timely performed. The state-of-the-art
lies on the versatile capacity in term of real-time data manipulate of AR device.
Besides capturing feature, a chain of algorithms based on computer vision is em-
bedded into the AR device, aims to enhance the precision and performance of
inspection task. Right after, the technical damage report is fed-back to the man-
agement system and assessment model is discussed. A pilot application to an ex-
isting cable-stayed bridge is introduced, which is well applied for a year and show
a good potential for bridge maintenance.

Keywords: BIM, BMS, AR, Preventive Maintenance, Image Processing.

1 Introduction

Building information modeling (BIM) for the existing asset was paid special attention
in the last decades. It is highlighted some main challenges in term of BIM model crea-
tion or data capturing, as well as manipulating and updating information, or finally as-
sessing and controlling the uncertainty data (Volk et al. 204). Especially the developed
countries are facing complexity in term of infrastructure management. It is no longer
the kind of direct trouble where the engineer can easily solve onsite, it is a continuous
and systematic network, which affects each other. Therefore various BMS were sug-
gested (Shim et al. 2017; 2019; Dang et al. 2018), however, developing an automated
system has always been a difficult job for engineers. The first challenge is the process

© Springer Nature Singapore Pte Ltd. 2020 1217

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_195
1218 N. S. Dang and C. S. Shim

of creating and managing the information system for the bridge. Bridges are diverse,
the information systems might significantly change for different bridge or type of
bridge. From the perspective of an automated system, if the input/output data judgment
and responsibilities are not strictly controlled, the risk of propagating errors increases.
System performance degradation also needs to be considered when several parameters
and geometric constraints are included (Lee et al., 2006). On another hand, the inspec-
tion workflow also needs to be enhanced. Several automatic workflows based on the
image processing technology have been proposed (Matsumoto et al., 2012; Nagrale et
al., 2013). In which structural damage can be detected in real-time by applied machine
vision concept. Among all inspection objectives, crack detection is the most important
and is increasingly attracting significant attention (Pereira et al., 2015; Mohan et al.,
2017). The inspection data can be directly uploaded onto the server after capturing
through a mobile device or an unmanned aerial vehicle (UAV), or another weara-
ble/augmented reality (AR) device. At the same time, the engineer can analysis to de-
termine the crack profile, width, length, as well as propagation direction, significantly
supports the decision-making process from a remote office. The following figure pro-
poses the configuration for the BMS using two modules: a remote information manage-
ment system, and an automatic site inspection system.

Fig. 1. BMS based on information management system and site inspection system

2 Information management system (IMS)

2.1 Information requirement (EIR)

For maintenance purpose, the information system is designed at field-verified level, it

means the real conditions of aged bridge should be updated eternally. But currently, the
maintenance demand of an in-service aged bridge always have far outpaced the availa-
ble resources, therefore a comprehensive but minimalist approach to bridge manage-
ment is required. Referring to the preventive maintenance aspect, maintenance strategy
BIM-based innovative bridge maintenance system... 1219

should be implemented as soon as right after the handover of the bridge. The general
procedure for maintenance work is a closed loop of interactive processes and continu-
ously repeated throughout the service life of the bridge. It includes inspecting, moni-
toring, and performing appropriate repair or rehabilitation work and upgrading the feed-
back to the database. Damage classification needs to be considered before the creation
of IMS, aims to save the time and resource of obtaining the necessary information -
which needs to be paid much attention later.
2.2 Data-schema for bridge maintenance

The common data environment (CDE) for a bridge stores all information related to the
physical features of that bridge, which can be gathered from design and construction
phase, and upgraded from inspection and repair records. BIM concept is the kind of
alignment-based and object-oriented authoring, the 3D information model is separately
defined and eventually assembled together by specific ID and bridge alignment. Each
structural element is categorized into an inventory system according to either role or
service in entire bridge system. Thereafter, specific element ID is defined, and the BIM-
oriented information model can be generated by adding the corresponding information
into specified object ID.
In the CDE, two main characteristics: "attribute" and "archive" are the backbone of
the information system. The individual attribute feature consists of all physical infor-
mation such as geometric and material property, orientation and role, quantity/cost es-
timating. This piece of data is the kind of “active information” that means it put into
use for every purpose or project stage. The individual archive data, on the other hand,
is the kind of inactive or superseded information which consists all related information
following the EIR's aspects. For example, inspection plan/manual, damage/repair rec-
ord etc. which is continuously obtained during the bridge lifecycle, from design and
construction, to O&M stage, and specified in a list. Depending on the level of develop-
ment (LOD) of information requirement, only needed information is adequately pro-
vided.

Figure 2. Data schema for the IMS

1220 N. S. Dang and C. S. Shim

2.3 Modeling authoring

It is required high interoperability such as linking the metadata, including all attribute
and archive. In this regards, an integrated model is suggested to use. The integrated
model embedded all archive information of any structural member into the 3D model,
following the ID system accordingly. BIM-based 3D model is generated through para-
metric modeling concept. For digital mock-up (DMU), an integrative work between a
CAD program and a programing language was launched by the aided of application
programming interface (API). The common API can ease the work of programming
graphical components and facilitate the integration of new features into existing appli-
cations through the advantages of open-source language. Moreover, an API is able to
assist otherwise distinct applications with sharing data, which can help to integrate and
enhance the functionalities of the applications.

Figure 3. BIM-based integrated model

3 Site inspection system

3.1 Inspection workflow
Inspection works no longer based on the visual observation and manually describe
the damage of the structure due to the inconsistent error can be occurred. Recently the
hand-off technical toolkits or AR devices help the engineer to improve the inspection
performance. Depend on the performance of the AR device, the main tasks of inspec-
tion works should be done piecemeal. In this regards, the device not only plays a role
as an input/output device but also it can pop-up the 3D model in the real environment,
to overlap the virtual model with the actual asset. Start with scanning the QR code, so-
called "target", which attached on the existing structure, the device automatic accesses
the 3D information model to find the corresponding structural element and pop-up it
into the real environment. The virtual model is predefined to match the location and
orientation of the attached "target" on the structure ideally. But due to the limited ca-
pacity of the AR device in term of holographic computing, the model is often popped-
up with reasonable tolerance, that's why the manual adjustment is required for a perfect
overlapping of the federated model. Fortunately, surface detection feature is a very
BIM-based innovative bridge maintenance system... 1221

basic function of the AR device, that means if the "target" is attached perpendicular
with the floor (e.g. attach on a wall), then the virtual model always matches the vertical
alignment. Therefore, the overlapping procedure is manually handled by the planar-
transform operation such as movement or rotation.
Once the overlapping task is perfectly fixed, the virtual model might be shifted to
the transparent mode in order to highlight the damage on the surface of the structure. A
chain of computer vision -based algorithms is embedded into the AR device, for exam-
ple, crack detection. The AR device starts with performing a threshold algorithm to
create a binary image. This work aims to simplify or partition the damage photo into
sets of analyzable pixels (each pixel has only two possible values: black or white).
Thereafter, an appropriate edge detection algorithm is applied to identifying the pixel
which is sharply changed in brightness, in other word, discontinuities. Those pixels are
organized into a set of curved line segments, termed edges. Herein, the distinction be-
tween the crack and the edge of the structure should be taken in to account by applying
different edge detection method (edge of the structure is defined as a boundary between
two uniform regions by comparing the intensity of neighboring pixels). By exclusion
algorithm the edge of structure from the edge detection algorithm in general, the crack
damage is highlighted.

3.2 Real-time assessment

Figure 4. Application frame work for AR device

Note that despite an AR device can cover the entire process of inspection task, but the
device performance (or capacity) needs to be taken into account. Therefore, the data
processing is suggested to perform on SQL server, aims to speed up the process of the
whole system. Damage capture in digital image format is immediately uploaded onto
the SQL server. Image tracing conversion, as so-called raster-to-vector conversion or
vectorization, is applied. In this scene, each pixel (in raster graphic) is mapped into the
vector graphic environment, the pixels’ coordinates can be represented by a square-grid
coordinate system. Depending on the different R-G-B values of each pixel, points are
generated. The boundary line for the entire mapped points thereafter forms the shape
of the crack. By using interpolation functions, it is easy to recreate all related infor-
mation for the cracks, such as the crack profile, total length, and maximum width of
1222 N. S. Dang and C. S. Shim

cracks, etc. Technical damage report is generated without any discontinuity or defi-
ciency, and finally, be updated to the inspection system and synchronized with the in-
formation management system at the main office. Decision-making team can work col-
laboratively with the inspection team at the same time, significantly improve the per-
formance of operation and maintenance job.

4 Conclusion
A full application framework for a real-time bridge maintenance system is proposed.
The state-of-the-art lies in its versatile capacity in term of data manipulation, with just
a couple of wearable-devices used by a couple of engineers, inspection work can be
quickly and flexibly performed. Parts of this research work including an application of
BMS is applying into an existing cable-supported bridge in Korea and shows good po-
tential for performance-enhancing of maintenance work. Furthermore, depending on
device ability in term of connection, further development of AR application should be
considered. Hopefully by the next generation of AR device (Microsoft Hololens 2,
which has been released this year), with the higher speed and performance of the pro-
cessor, data processing and even machine learning can be totally implemented with just
one device.

Acknowledgements
This research was supported by a grant “Development of life-cycle engineering technique and
construction method for global competitiveness upgrade of cable bridges (16SCIP-B119960-01)”
from Smart Civil Infrastructure Research Program funded by MOLIT and KAIA.

References
1. Volk R., Stengel J., Schultmann F.: Building Information Modeling (BIM) for existing buildings.
Literature review and future needs Automation in Construction, 38:109-127 (2014).
2. Shim, C.S., Kang, H.R., Dang, N.S., Lee, D.K: Development of BIM-based bridge maintenance
system for cable-stayed bridges. Smart Structures and Systems, 20(6): 697-708 (2017).
3. Dang, N.S., Shim, C.S.: BIM authoring for an image-based bridge maintenance system of existing
cable-supported bridges. IOP Conference Series: Earth and Environmental Science, 143(1)
(2018).
4. Dang, N.S., Kang, H.R., Lon, S., Shim, C.S.: 3D digital twin models for bridge maintenance,
Proceedings of 10th International Conference on Short and Medium Span Bridges, Quebec city,
Quebec, Canada (2018).
5. Shim, C.S., Dang, N.S., Lon, S., Jeon, C.H.: Development of a bridge maintenance system for
PSC bridges using 3D digital twin model, Structure and Infrastructure Engineering. (Accept on
26-Feb-2019).
6. Lee, G., Sacks, R. & Eastman, C. M.: Specifying parametric building object behavior (BOB) for
a building information modeling system. Automation in construction, 15(6): 758-776 (2006).
7. Matsumoto, M., Mitani, K., Sugimoto, M., Hashimoto, K., Miller, R.: Innovative bridge assess-
ment methods using image processing and infrared thermography technology. “Innovative Infra-
structures - Toward Human Urbanism, 18th Congress of IABSE”, Seoul, IABSE, 18: No. 13,
pp.1181-1188 (2012).
8. Nagrale, M. S. K., Bagde, M.S.T.: Application of image processing for development of automated
inspection system. Int. J. Comput. Eng. Res, 3(3) (2013).
9. Pereira, F. C. & Pereira, C.E.: Embedded image processing systems for automatic recognition of
cracks using UAVs. IFAC-PapersOnLine, 48(10): 16-21 (2015).
10. Mohan, A., Poobal, S.: Crack detection using image processing: A critical review and analysis.
Alexandria Engineering Journal (2017).
 Perspectives on BIM Profession of BIM Specialists and
non-BIM Specialists: Case Study in Vietnam

Ngoc Quyet Le1, Michael Er2, Shankar Sankaran3 and Ngoc Binh Ta4
1,2,3
University of Technology Sydney, NSW 2007, Australia
4
Institute of Construction Economics, Hanoi, Vietnam

Abstract. The adoption of Building Information Modelling (BIM) has been re-
cently increasing within the Architecture, Engineering and Construction (AEC)
industry. The current approach on adoption of BIM by the Vietnamese decision
makers (e.g. government agencies and senior industry leaders) is primarily con-
cerned with improving the adoption rate measured by the speed of diffusion and
the number of adopters at basic implementation level such as 3D functions. This
paper explores a different perspective on BIM adoption in Vietnam which has
been neglected by proposing that the efforts of the decision makers should shift
into regulatory supports and diffusion networks facilitating higher levels of BIM
implementation such as 4D construction scheduling, 5D cost estimating etc. to
confirm their long-term commitment to advanced BIM practices. Twenty-nine
participants including BIM specialists and non-BIM specialists were selected
from seven AEC organizations. Semi-structured interviews were employed for
data collection. Key findings revealed general perceptions of the BIM profession
such as “job insecurity”, “depleted motivation”, “BIM as supporting roles” and
“BIM as new skill sets”. Recommendations for programs supporting BIM adop-
tion are also discussed.

Keywords: BIM Adoption, BIM Profession, BIM Specialists, non-BIM Spe-

cialists, Qualitative Case Study, Developing Countries, Vietnam

1 Introduction: Review of BIM Adoption in Vietnam

The AEC industry by its nature of complexity (e.g. multidisciplinary working environ-
ment) has suffered from high fragmentation, poor productivity, cost and time overruns
as well as enduring conflicts and disputes. BIM as an integrated information system,
therefore, promises to address these problems. However, the use of BIM has not had
the expected impact on the construction industry in most developing countries as com-
pared to developed countries even though it has been introduced a decade ago [1]. Par-
ticularly in Vietnam, BIM has been reportedly being in use since 2009 [2] but 3D co-
ordination and 2D drawing extraction still are the most used applications among AEC
stakeholders [3]. Recent BIM studies in Vietnam showed that 3D designs alone, despite
their wide range usage, has not significantly changed the AEC industry from traditional
2D CAD projects (i.e. linear platform) to BIM-based projects (i.e. collaborative plat-
form) as expected [2], [3]. Diffusion of Innovation Theory (DOIT) has been widely

© Springer Nature Singapore Pte Ltd. 2020 1223

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_196
1224 N. Q. Le et al.

used to interpret why an individual either decides to make full use of an innovation as
the best course of action available (adoption), or choose not to adopt it (rejection) [4].
However, the phenomenon of “retaining the adoption of innovation [BIM] but being
limited into small scale (e.g. pilot projects) and low level of implementation (e.g. 3D
designs) for a long time” found in Vietnam [2], [3] is relatively unexplored in DOIT.
With the aim of contributing to DOIT and its relevance to BIM adoption, this paper
focuses on the perspectives on BIM profession of BIM specialists and non-BIM spe-
cialists to understand why Vietnamese AEC professionals hold BIM back to be applied
to 3D designs. Further, the Government mandatory submission of 3D models for all
first-class projects and public projects by 2020 [5]; and the organizational strategy fo-
cusing on extensive in-house equipment of 3D tools are likely not adequate for chang-
ing the sluggish adoption in Vietnam [2]. This is because BIM model itself is not a goal,
but rather a mean by which project goals may be achieved [6]. The truly successful
BIM is related to sociology (e.g. communication and collaboration) rather than BIM
technology update. Therefore, diffusion networks (e.g. professional network for BIM)
and regulatory supports (e.g. the national recognition of BIM professionals) could be
more important to ensure industry engagement with BIM at national level. This issue
will be discussed as a recommendation for policy makers and senior industry managers.

2 Methodology

Qualitative case study was employed as it allows for multiple facets of the phenomenon
to be revealed and understood; particularly when there is less research on BIM adoption
regarding Vietnamese context [7]. Respondents were identified through a snowball
sampling technique which is based on referrals from initial respondents to generate ad-
ditional respondents. This technique could help the researchers to collect primary data
in a time and cost-effective manner. Respondents’ profiles are coded to protect their
anonymity (see Table 1). The findings will be discussed using DOIT (See Section 4).

Table 1. Respondents’ profiles

Company Respondent Position BIM role
R1D1; R2D1 Director Board Non- BIM Specialist
D1 (Design) R3D1; R4D1; R5D1 Design Leader BIM Specialists
R6D1; R7D1; R8D1 Drafter BIM Specialists
D2 (Design) R1D2 Director Board Non- BIM Specialist
D3 (Design) R1D3 BIM Manager BIM Specialists
D4 (Design) R1D4 Director Board Non- BIM Specialist
R1C1 BIM Manager BIM Specialist
C1 (Contractor) R2C1; R3C1; R4C1 Site Manager Non- BIM Specialist
R5-10C1 BIM coordinator BIM Specialist
R11-15C1 Site Engineer Non- BIM Specialist
O1 (Owner) R1O1 BIM Manager BIM Specialist
G1 (Government Agency) R1G1; R2G1 Change Agent Non- BIM Specialist
Total: 07 organizations with 29 respondents (15 BIM specialists and 14 non-BIM specialists)
Perspectives on BIM Profession of BIM Specialists... 1225

3 Findings

3.1 Theme 1: Self-recognition of BIM profession in AEC firms

Job Insecurity. Firstly, BIM specialists indicated their welfare concern arising from
the copyright liability of BIM software. “Copyright liability can make BIM team stop
working until all related legal issues are resolved.”- R3D1; R4D1; R1D3. Most of re-
spondents admitted illegal BIM programs have been used in their workplace, thus, rais-
ing the risk of financial and reputational damage. “Penalties for infringement could
damage organizational finance and affect employees’ wages.”- R3D1; R4D1; R1D3. Sec-
ondly, majority of BIM specialists acknowledged that their products are deficient and,
thus, perceived inadequate to assure their BIM roles within the company, not to men-
tion construction industry. “We can’t guarantee our position in company with imperfect
BIM products.”- R3D1; R4D1; R1C1; R5-10C1. This is because purchasing BIM authoring
packages can represent a significant overhead for construction companies; whereas a
small number of licensed software is not adequate for the large project’s demands.
Depleted Motivation. Most of BIM specialists conceded their poor morale when con-
stantly working in a trial or pilot environment. This mechanism does not offer an op-
portunity for BIM specialists, especially young staffs, to acquire practical experiences
– making them feel unenthusiastic and anxious about their skill erosion. “Every day
working around pilot projects and trial experiments frustrate us.”- R1C1; R5-10C1. Ad-
ditionally, BIM specialists found their career path disadvantaged in comparison with
peers using AutoCAD. Particularly, in the same period of time working, AutoCAD us-
ers could accumulate sufficient experiences from real projects- qualified by the number
and size of completed projects, to gain promotion or achieve professional certificates.
Conversely, BIM adopters only carry out pilot projects (i.e. being inadequately recog-
nized in professional profiles) and the scope of work is limited in large-long duration
project (i.e. being slowly updated in professional profiles). “Our career path is ambig-
uous and lag behind CAD users.”- R1C1; R5-10C1. Finally, the feeling of isolation at an
emerging BIM profession would possibly discourage BIM adopters from pursuing
higher level of BIM implementation. “We feel lonely in new BIM profession as the
connection to BIM fellows and experts is lacking. “- R3D1; R4D1; R5D1; R1C1; R5-10C1.
The lack of “diffusion networks” is likely the main reason. BIM specialists stated that
it is impossible for them to raise questions of BIM applications or report a software
crash to any official support (e.g. BIM knowledge network) because the administrators
could track the software licenses back and find their violation. Moreover, it appears that
young BIM staffs get little guidance from senior managers regarding proper BIM im-
plementation in daily tasks. This is because older people proficient in AEC speciality
tend to lack in BIM technical skills whereas young people with better technology com-
petence are short of discipline specific skills [3].
3.2 Theme 2: Social-recognition of BIM profession in AEC business
BIM profession as a supporting role. Most non-BIM specialists at high level man-
agement described BIM as a supporter for project deliverables. For example, “3D rep-
resentation becomes popular in visual marketing strategy and technical meeting with
stakeholders.”- R1D2; R1D4; R1O1. This is because decision makers (e.g. directors and
1226 N. Q. Le et al.

owners) tend to heavily rely upon the effectiveness of problem solving process exer-
cised by senior managers’ experiences, particularly in planning and rough estimating,
rather than BIM technologies. Regarding non-BIM specialists at lower level manage-
ment (e.g. site engineers), BIM is perceived as simply 3D visual representation to assist
them with buildability issues. “3D models help site staffs with understanding complex
MEP intersection settings.”- R11-15C1. Site staffs expressed that their distrust of using
BIM models for entire construction process is due to the corrupted data transmission
and downtime problem. “We can’t rely on the inaccuracy 3D models for our consecu-
tive scheduling analysis” and “It’s impossible to create 4D scheduling models as the
speed of BIM tools were not full able to catch up the site progress.”- R11-15C1. In addi-
tion, site staffs cannot fully engage in developing BIM models as their tasks are under
the control of site managers. Information update for as-built models, hence, is likely
either interrupted or inconsistent as the new decisions of personnel rotation and arrange-
ment within and across projects are practiced frequently by site managers.
BIM as emerging skills for AEC professionals. The data disclosed that BIM has been
still not recognized as a mainstream profession such as architecture and engineering.
Currently, BIM is perceived as a new skillset (i.e. computer technology enhancement)
necessary for AEC professionals rather than a new construction domain. This is because
BIM philosophy of fully collaboration with all stakeholders relies on trust, collabora-
tion and transparency which are all new in compared to the conventional approach. The
Government agents asserted that their current intentions are to increase the awareness
of BIM among industry by emphasizing the need for BIM education associated with
universities and companies. “Considering BIM as a new discipline means the Govern-
ment must revise most building codes and standards to support it- that are really ex-
hausted tasks”- R1G1; R2G1. Further, BIM is viewed as not much of relevance to con-
struction business. Neither general contractors nor owners are satisfied with additional
payment for BIM services. Contractors supposed that BIM skill sets help designers to
increase the speed and accuracy of the designs but these benefits are not directly rele-
vant to their bottom line. “We still make profits with traditional methods, just rework
and raise extra-claims”- R2C1; R3C1; R4C1. For owners, their focus is seen as making
quick profits by selling buildings faster. They may have to “build the units at lower
cost to sell quickly rather than raise the standard of construction”- R1O1. In other
words, BIM is perceived as a tool (or skill set) to help project members (e.g. contrac-
tors) achieve their goals [8] but has been slow to change the owners’ business models.

4 Discussion

The perception of job insecurity is influenced by the contextual conditions in Vietnam

where BIM specialists feel unsecured about their roles. First, the copyright liability of
BIM software could suspend BIM team’s activities. The low morale and sensitivity to
copyright violation could be explained by the Asian “collectivist culture” which leads
to a weak individual assumption of responsibility- meaning “the rightness of a law de-
creases when more people violate it” [9]. Second, BIM specialists do not feel satisfied
Perspectives on BIM Profession of BIM Specialists... 1227

and happy with their malfunctioned BIM products as they insufficiently meet the ex-
pectations of other social members (e.g. peers and senior managers). This feeling
matches the concept “structural equivalence” in social system of DOIT [4]- that is, the
degree of equality in network position could influence the adoption of innovations [10].
The perception of depleted motivation coincides with the concept “interpersonal
communication” of DOIT [4]– meaning that the effects of peer comparison and the lack
of knowledge sharing network are critical barriers to higher BIM practices. Whereas,
the perceptions of BIM as only a supporting role and BIM as new skill sets all come
from the concept “perceived relative advantage of an innovation” in DOIT [4]- imply-
ing that although Vietnamese contractors and owners are actively engaged on gaining
all possible outcomes and benefits out of BIM, they do not see any significant growth
with just 3D. To these stakeholders, 3D model [even in BIM platform] is only as good
as a paper-based blueprint, just that it is easier to interpret and explain.
Given the above perceptions on BIM profession, it is argued that at the individual
level (e.g. single users and in-house usages), the diffusion and adoption of “basic” BIM
technologies could be critical; whereas, at the industrial/project level (e.g. multi-disci-
plines and stakeholders involved), regulatory supports and diffusion networks could be
more important. Policy makers should facilitate true collaboration [by regulatory sup-
ports] between disparate project stakeholders who are often reluctant to share innova-
tive solutions due to contractual relationships and intellectual property issues. Copy-
right issue should be taken into account to enhance the development of local BIM soft-
ware companies for more affordable BIM solutions. Also, the official recognition of
BIM as a mainstream AEC profession is necessary to reassure the roles of BIM special-
ists into companies and projects- helping BIM specialists with the more confident ca-
reer path. Since contractors are eager to manage their onsite tasks, 4D BIM scheduling
for optimal resource uses and manageable time-frame is recommended. Whereas, for
owners, 5D BIM estimating for tracking cash-flow is on their top concern. Therefore,
to ensure the digital transformation of the AEC industry happens, the long-term vision
and commitment of senior industry officials to support higher BIM practices are nec-
essary. Companies in the construction sector should develop a BIM knowledge network
with professional institutions and university academics to be able to maintain additional
channels of advice and support. Also, intra-organization BIM network could be devel-
oped with the leadership of key BIM diffusion drivers (e.g. top general contractors and
developers) to feed updated BIM practice knowledge back into BIM projects.

5 Conclusion

This research identified key themes that cover common understandings of stakeholders
regarding BIM profession relevant to their job/business. While the themes (job insecu-
rity and depleted motivation) show the negative attitudes of BIM specialists towards
the lack of “trust” environment for participative safety, the themes (BIM as supporting
roles and BIM as emerging tools) present the “wait and see” attitudes of non-BIM spe-
cialists due to the lack of reliable sources for seeking higher levels of BIM solutions
specific to their situations. The recommendations, hence, tend to shift the focus to the
1228 N. Q. Le et al.

“network building” associated with regulatory supports rather than the “simply” man-
datory submission of architectural models, and “basic” BIM education framework.
The recommendations may seem contradictory to the common diffusion approach
(top-down processing) of generalizing basic [3D] BIM principles using “propaganda”
from the Government-led curricula. Despite having certain benefits which are obvi-
ously realized and easily imitated (thus, raising awareness and faster diffusion); 3D
BIM is still seen as “incremental” and “low-tech” innovation application compared to
the full BIM exploitation. Further, industrial parties in Vietnam have come a long way
in the last decade-evident by some advanced companies now advise on BIM implemen-
tation (radical practices) back to the “BIM lagging behind” Government; and the fact
that the national BIM regulations, industrial standards and academic courses have not
kept pace with innovative BIM practices. The industry innovation, hence, does not re-
quire to be generally educated but to be re-aligned (e.g. orientation, connection, and
facilitation) for self-evolution. Building the strong peer-network sponsored by the Gov-
ernment and driven by industry leaders (bottom-up processing) appears to be the prom-
inent solution and high-priority task in the case of resource shortage, low-tech infra-
structure, and overwhelmed by policy changes- not only BIM standards/protocols but
also relevant policies supporting sustainable BIM development such as ownership and
intellectual property, contractual agreement, authenticity, product liabilities risks, in-
centives, etc. Also, the draft BIM legislation is easily put on the networks to get public
feedback/comment for policy-revision and “already” mindset-preparation for change.

References
1. Mehran, D.: Exploring the adoption of BIM in the UAE construction industry for AEC firms.
Procedia Engineering 145, 1110 – 1118 (2016).
2. Le, N., Er, M., Sankaran, S.: The implementation of Building Information Modelling (BIM)
in construction industry: Case studies in Vietnam. International Journal of Engineering and
Technology 10 (4), 335–340 (2018).
3. Le, N., Er, M., Sankaran, S.: Building Information Modeling (BIM) adoption and
implementation: Interaction between BIM users and non-BIM users. In: 4th Australia and
South-East Asia Structural Engineering and Construction Conference, pp. 1– 6. ISEC Press,
USA (2018).
4. Rogers, M.: Diffusion of innovation. 5th edn. Free Press, NewYork (2003).
5. Ismail, N., Chiozzi , M., Drogemuller, R.: An overview of BIM uptake in Asian developing
countries. In: 3rd International Conference on Construction and Building Engineering, pp.
1–7. AIP Publishing, USA (2017).
6. Walasek, D., Barszcz, A.: Analysis of the adoption rate of Building Information Modeling
[BIM] and its return on investment [ROI]. Procedia Engineering 172, 1227–1234 (2017).
7. Yin, R.: Case study research: Design and methods. 3rd edn. Sage Publications, USA (2003).
8. Sistani, N., Rezaei, A.: Building Information Modeling; a new business trend in the
construction industry of Iran. In: Achour, N. (eds.) CIB WORLD BUILDING CONGRESS
2016, vol.5, pp. 856 – 867. Tampere University of Technology Publishing, Finland (2016).
9. Wang, F.: Software piracy and ethical decision making behavior of Chinese consumers.
Journal of Comparative International Management 8 (2), 43–56 (2005).
10. Valente, T.: Social network thresholds in the diffusion of innovation. Social Networks 18,
69– 89 (1996).
 The contextual influence on Building Information
Modelling implementation: A cross-case analysis of
infrastructure projects in Vietnam and Norway

Nam Bui

Oslo Metropolitan University, Oslo, Norway

nam.bui@oslomet.no

Abstract. This paper compares Building Information Modelling (BIM) imple-

mentation in two infrastructure projects in two different countries to understand
the contextual influence on innovation adoption in the construction industry.
These two countries are Norway and Vietnam, which have different contexts in
term of technology advance. In Norway, Bane NOR - the Norwegian railway
authority decided to use BIM in the InterCity project which connects Oslo and
nearby cities. InterCity is the first Norwegian large-scale railway project that uti-
lizes BIM. The situation is similar in the Thu Thiem 2 bridge project in Vietnam.
The design package of Thu Thiem 2 Bridge is a BIM pilot project of the Viet-
namese government. Using qualitative interpretive case studies, this paper re-
veals the importance of the BIM champions in the implementation and different
consequences derived from the project organization, technological context, and
the focus of BIM use. The insights obtained from the comparison might be useful
for increasing the BIM uptake and overcoming the barriers of traditional require-
ments for innovation. This is not only applicable to the Norwegian and Vietnam-
ese infrastructure sectors but also for other contexts, which have limited BIM
implementation experience.

Keywords: BIM implementation, cross-case analysis, developing country, in-

stitutional theory.

1 Introduction

Construction productivity has been almost steady since 1995 despite various improve-
ment efforts. Based on a sample of 41 countries representing 96% of the global gross
domestic product (GDP), Barbosa et al. [1] reported that the construction labor produc-
tivity only grows one percent over the past two decades. This growth lags behind the
average annual rate of 3.6% in the manufacturing industry, or 2.8% of the world economy.
Increase in productivity might raise the construction growth rate and support the global
economy. In fact, the construction industry accounts for 13% of the world GDP. Barbosa
et al. [1] reported seven construction areas that can boost productivity. One of those areas
is implementing technology and innovation, particularly digital technology [1].

© Springer Nature Singapore Pte Ltd. 2020 1229

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_197
1230 N. Bui

Construction practitioners propose Building Information Modelling (BIM) as a main

technology for digitalization. However, the uptake of BIM varies among countries. De-
veloping countries seem slower than developed ones in BIM implementation. To sup-
port BIM uptake in developing countries, particularly in the infrastructure domain, it is
ZRUWKWRH[SORUH³What are contextual influences on BIM implementation?´
This paper addresses the research question by conducting a cross-case analysis be-
tween two infrastructure projects in Norway and Vietnam. In brief, the following con-
tent will explain BIM implementation experiences from a Norwegian railway project
and a Vietnamese bridge project through the institutional theory lens. The insights
which emerged from the comparison might be essential for BIM implementation in
other projects which share similar contexts.

2 BIM implementation in infrastructure

BIM, an application of digital technology in construction [2], provides new tools to

improve project performance. The definition of BIM has various versions. In this paper,
BIM is understood as a process enabled by digital technology [3]. The benefits of BIM
seem limited to building projects. In fact, construction practitioners suppose that BIM
technology is unlikely to create the same benefits in infrastructure because this sector
has substantial differences from building [4]. While buildings generally have vertical
deployments, infrastructure projects often expand along central lines. That makes con-
struction practitioners refer to BIM uses in infrastructure and building DV³KRUL]RQWDO´
DQG ³YHUWLFDO´ %,0 UHVSHFWLYHO\ ,Q WKH 86 %,0 XSWDNH LQ infrastructure lags three
years behind that of building [5]. Since BIM has proven benefits in buildings, infra-
structure projects might gain similar advantages through proper implementation.
Norway has seen two main directions to develop BIM technology, namely conven-
tional and open approaches. With the conventional approach, companies use their own
resources to develop BIM applications internally. With the open approach, construction
practitioners build a common foundation to exploit this technology. An example of the
open approach is openBIM including open standards for data exchange in the built envi-
ronment [6]. In Norway, both public and private sectors have been actively involved in
BIM development. From the public sector, Statsbygg, the Norwegian government agency
for public facility management and development, issued the BIM guidance version 1 in
2008 and requested IFC (openBIM) use from 2010 [3, 7]. In 2015, Statens Vegvesen
published a BIM handbook for transportation projects. From the private sector, some soft-
ware vendors such as JotneIT, Vianova started offering BIM tools in the early 1990s.
In Vietnam, Tran et al. [8] reported BIM uses in major construction companies such
as Hoa Binh Corporation, Contrexim, VNCC, Polisius Vietnam, etc. Furthermore, the
government agencies in construction actively guide BIM adoption on the national scale.
BIM Steering Committee, established in March 2017, holds the leading role in this adop-
tion effort. In April 2018, the Ministry of Construction announced twenty BIM pilot pro-
jects in connection with the BIM adoption plan. These pilot projects come from both
building and infrastructure domains and have funding from both the public and private
sectors. The Vietnamese government aims to apply BIM countrywide from 2021 [9].
The contextual influence on Building Information... 1231

3 Theoretical lens

This paper adopts institutional theory [10] as guidance for data collection. This theory
identifies pressures which are at play when institutionalized processes go through
change. Particularly, the focus lays on the institutional effects, which derive from coer-
cive, normative, and mimetic pressures. Coercive pressures are about explicit regula-
tory processes which organizations have to follow. Normative pressures relate to social
obligations such as professionalization, following rules and norms in a discipline. Mi-
metic pressures make an organization imitate practices of other organizations which
seem more successful [10]. Institutional theory also mentions actors within the organi-
zation who have special interest and resources, namely institutional entrepreneurs, who
support the implementation of new practices, especially in the beginning stages [11].
Since institutional theory can help to explain innovation adoption in organizations [12],
this theory is a useful lens to understand BIM implementation in the selected studied
cases.

4 Method

This paper presents a comparative analysis between the design of the Dovrebanen rail-
way, a part of InterCity project in Norway, and the Thu Thiem 2 Bridge project (TT2)
in Ho Chi Minh City, Vietnam. The case studies were selected based on four criteria:
1. Have BIM application as an objective
2. Are pilot projects in their domains
3. Have different contexts
4. Have commonalities
In Norway, the railway is behind other domains in BIM implementation. While the
building, bridge, and road projects have BIM guidance, there is no BIM instruction for
railway projects. Therefore, the BIM implementation in InterCity aims not only to take
the advantages from the technology but also to build guidance for future railway pro-
jects. The InterCity project team won the Autodesk 2016 AEC Excellence Awards for
successfully applying BIM to facilitate communication across 120 internal and external
stakeholders. This achievement suggests that InterCity is an interesting case for this
study.
In Vietnam, TT2 may be the first effort to use BIM in the infrastructure domain.
There was no official BIM guidance from the government or the industry, even from
other construction domains in Vietnam at the time of data collection. The TT2 consult-
ant team consisted of WSP, a foreign consultant from Finland, and TEDI South, a local
engineering company. After design completion, the TT2 design won Tekla BIM Award
2017 for successfully integrating all disciplines in the project.
The interview guide is semi-structured with seventeen open questions based on in-
stitutional theory. With the focus on institutional effects, these open questions aim to
capture how coercive, normative, and mimetic pressures influenced BIM-based work
in the case studies. The data collection task was carried out from February to July 2017.
There were ten interviews with the InterCity project team and eleven with the TT2
1232 N. Bui

team. Each interview lasted about one hour. The two project teams had experienced
managers. All 11 managers had more than 10 years of experience in construction. While
the Norwegian and Finish practitioners have used BIM since the early days of their
career, TT2 was the first BIM project of the Vietnamese engineers. The BIM managers
in these two projects had more than 20 years of BIM related experience. Two previous
reports about these case studies covered more detail of the informants [13, 14]. In brief,
70% of interviewees are managers. Besides the focus on institutional effects, the data
also includes backgrounds and roles of the interviewees in the case studies. In general,
these two case studies have contextual differences but also share three common obsta-
cles. These obstacles are limited BIM experience from the client, no specific BIM guid-
ance, and paper-based approval process.

5 Findings

This section presents the common obstacles and differences found in the two studied
projects. The highlights of differences on problem-solving based on institutional lens
will provide insights on contextual influences on BIM implementation.
Table 1. Response to the obstacles on BIM implementation
InterCity Thu Thiem 2 Bridge
Limited BIM experi- - Decide to use BIM in In- - Organize BIM workshops
ence from the client terCity project c for client and relevant ad-
- Hire experienced BIM ministration agencies m
manager m - Sign contract with experi-
enced Finnish consultant c,m
No specific BIM - Form a reference group m - Adopt the Finnish BIM ex-
guidance - Prepare BIM handbook c ecution plan m
- Organize internal training
workshop m
Paper-based approval - Use the 3D model as an il- - Check the design based on
process lustration tool n the 3D model n
c: coercive pressure, m: mimetic pressure, n: normative pressure

5.1 Contextual differences

Norway and Vietnam have inherent differences in economic, investment on research

and development, culture. This section limits the focus on the differences in technology
adoption that the data reveals. From the data, the InterCity project members used BIM
in previous projects. They were familiar with the Norwegian BIM handbooks in build-
ing, bridge and road domains. The InterCity consultants could use any BIM tools that
offered free viewers. This was the open approach to BIM which supports the client to
control the data better. The main tools for designing and collaboration were Revit, Bent-
ley, and Trimble. Each consultant prepared their work packages and submitted with
dwg format. The InterCity BIM manager controlled this collaboration process.
The contextual influence on Building Information... 1233

In contrast, the Vietnamese local consultant had limited BIM experience and followed
the Finish partner regarding BIM implementation in TT2. The BIM tool for modelling
was Tekla Structures. Separately, the local consultant designed the approach bridge, while
the Finish consultants developed the main span before merged them together. The Finish
BIM manager led the collaboration. Vietnam had no official BIM handbook or guidance
when TT2 began. The TT2 project team adapted the Finish guideline to implement BIM
in the project.

5.2 How to overcome the obstacles

Regarding the Norwegian project, the InterCity leaders decided to mandate BIM use in
the project. Then, the client team hired a BIM manager with more than 20 years of BIM
implementation experience. In this way, the InterCity leaders not only formally requested
BIM use but also provided a reference person for the implementation. A consultant BIM
coordinator commented that ³WKHUHwas little progress before we had the BIM manager
[in the client team]´. In fact, the client BIM manager led a reference group with repre-
sentatives from sub-project teams. This group shared experience in ongoing projects and
built a handbook, which would become a unified guide for future railway projects in Nor-
way. In addition, she facilitated the integrated concurrent engineering (ICE) meetings,
which gathered necessary decision makers and engineers to solve specific problems.
These ICE actors collaborated through BIM models. BIM models also helped to illustrate
the design to relevant construction administration. The BIM model not only increased the
design quality but also reduced misunderstandings during the approval with paper draw-
ings.
In contrast, the TT2 client signed a consultant contract with WSP Finland. At the same
time of preparing the design, the foreign consultant organized various workshops and
training to showcase procedures and benefits of BIM implementation to the client, the
local partner, and relevant administrations. ³:HDUHVLPSO\ZRUNLQJDQGVWXG\LQJDWWKH
VDPHWLPH´VDLGWKH77FOLHQWWHFKQLFDOPDQDJHU. The BIM manager from the Finnish
consultant held the leading position in organizing the workshops and training events re-
lated to the project. For the BIM guidance, the project team adopted and adjusted the
Finnish BIM execution plan (BEP) based on the local regulations. In addition, the local
partner learned how to apply BEP through internal training led by the foreign partner. The
TT2 BEP would become an input for the Ministry of construction to build national BIM
guidance later. Also, TT2 teams used the BIM model to check their design in 2D draw-
ings. By using the 3D models in parallel with the conventional drawings, the two project
teams demonstrated how BIM technology could improve the design quality.

6 Discussion and conclusion

In the two studied cases, the client determination in using BIM was the starting point for
the implementation. Through formal requests for BIM in consultant contracts, the clients
exercised coercive pressures on innovation adoption. In addition, BIM champions ± BIM
managers who have the technical skills and lead the implementation, were vital as
1234 N. Bui

institutional entrepreneurs. The two BIM champions exerted mimetic pressure through
experience sharing, training, and workshops. They supported project members in imple-
menting BIM without official guides. Finally, the project members used the BIM models
to increase the design quality and support the approval process.
In summary, this paper presents a cross-case analysis of BIM implementation in two
infrastructure projects in Norway and Vietnam. The contextual differences led to dif-
ferent ways of BIM implementation. Through the institutional lens, this paper elabo-
rates how the project teams overcame obstacles for successful implementation. How-
ever, the data focused on managerial positions. Further studies shall cover specialists
for a broader view. Also, a cross-case study in the construction phase will add more
insights to the contextual influence on BIM implementation.

References

1. Barbosa, F., J. Woetzel, J. Mischke, M.J. Ribeirinho, M. Sridhar, M. Parsons, N. Bertram,

and S. Brown: Reinventing construction: A route to higher productivity, (2017).
2. Jin, R., P.X. Zou, P. Piroozfar, H. Wood, Y. Yang, L. Yan, and Y. Han: A science mapping
approach based review of construction safety research. Safety science 113, 285-297 (2019).
3. Hjelseth, E.: BIM understanding and activities. WIT Transactions on the Built Environment
169, 3-14 (2017).
4. Cheng, J.C., Q. Lu, and Y. Deng: Analytical review and evaluation of civil information
modeling. Automation in Construction 67, 31-47 (2016).
5. SmartMarket Report: The business value of BIM for infrastructure: Addressing America's
infrastructure challenges with collaboration and technology, (2012).
6. buildingSMART, Technical Vision, https://www.buildingsmart.org/standards/ technical-
vision/, last accessed: 30 January 2019.
7. Smith, P.: BIM implementation±global strategies. Procedia Engineering 85, 482-492 (2014).
8. Tran, H.M., V.H. Nguyen, N.B. Ta, and T.H.A. Le: BIM implementation worldwide and the
status of BIM adoption in Vietnam construction industry. Vietnam Construction Economics
Megazine, (2014).
9. BIM Steering Committee, News, http://bim.gov.vn/tin-tuc/tin-tuc, last accessed: 1
November 2018.
10. Scott, W.R.: Institutions and organizations: Ideas, interests, and identities. Sage Publications, (2014).
11. Wahid, F. and M.K. Sein: Institutional entrepreneurs: The driving force in institutionalization
of public systems in developing countries. Transforming Government: People, Process and
Policy 7(1), 76-92 (2013).
12. Colyvas, J.A. and S. Jonsson: Ubiquity and legitimacy: Disentangling diffusion and
institutionalization. Sociological theory 29(1), 27-53 (2011).
13. Bui, N.: BIM technology implementation in Vietnam: an institutional perspective on a bridge
project. In 22nd Pacific Asia Conference on Information Systems (PACIS 2018). Yokohama,
Japan (2018).
14. Bui, N., C. Merschbrock, B.E. Munkvold, and A.K. Lassen: An institutional perspective on BIM
implementation ± a case study of an intercity railway project in Norway. In 27th International
Conference on Information Systems Development (ISD2018). Lund, Sweden (2018).
 Application of BIM tools in technician training, a case
of Ho Chi Minh City Construction College, Vietnam

NGUYEN, Tung Lam and NGUYEN, Thi Xuan

Ho Chi Minh City Construction College,

265 No Trang Long Str., W. 13, Binh Thanh Dist., Ho Chi Minh City, Viet Nam.
thixuannguyen.tcxd@gmail.com

Abstract: In vocational training, keeping up with new technology is a challenge. Build-

ing Information Modeling (BIM) tools are now increasingly replacing traditional skills,
which is irreversible advance in architecture and construction sector. This leads to
changes in the curriculums of training courses, teaching and learning methods, especial-
ly in Technical Vocational education and Training (TVET). This article explores an ap-
plication process of BIM tools in BIM Modeler or architecture technician training
courses at Ho Chi Minh City Construction College as a case study, reflecting experi-
ence of change management toward new technology integration, its implication in and
challenges for 79(7¶VPHWKRGRORJ\7KHauthor also recommends that cooperation be-
tween industry and TVET in training is one of the most effective solution to solve hu-
man resource problem for construction sector.

Keywords: TVET, BIM Modeler, competency.

Ho Chi Minh City Construction College (the College) has been specialized in voca-
tional training for construction and architecture technicians since 1977. The &ROOHJH¶V
vocational courses are continuously updated with new applications, enabling students
to apply the knowledge effectively in industrial situation.
This article shares practical experiences through BIM tools training, one of the latest
technology applications, with the desire to obtain extra ideas, knowledges and meth-
ods from BIM community.

1 The process of training architecture technicians

In the 1990s, the College began developing courses for architecture technician who
were able to carry out the construction layouts by hand. Technicians training pro-
grams have been upgraded with CADD design software (CADD: Computer aided
design and drafting operator). Students learnt to work in AutoCAD environment.
Then, in addition to developing layouts, students were also able to illustrate archi-
tectV¶ ideas with SketchUp and 3DSMax.
Following the rapid advance in construction technology, with support from the indus-
try, the College had early approached to the concept of BIM (Building Information

© Springer Nature Singapore Pte Ltd. 2020 1235

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_198
1236 T. L. Nguyen and T. X. Nguyen

Modeling) in 2012 in form of BIM seminars ZLWK WKH SDUWLFLSDWLRQ RI $XWRGHVN¶V
experts. Numbers of seminars, training, technology transferring etc. concerning BIM
tools training were implemented afterward.

2 BIM application (BIM Tools) integration process

Since 2015, BIM tools has been incrementally updated in the architectural technician
curriculum, and gradually succeeding the existing training curriculum of the College.
Initially, BIM tools have been step by step integrated into the syllabus through small-
sized project, including Revit Architecture, Revit Structure, Revit MEP respectively.
In that way, learners use BIM tools to implement the process of developing models in
cyberspace. The structural components were developed corresponded to the LOD 200
[3].

After the Ministry of Construction issued Guidelines to apply the Building Infor-
mation Modeling (BIM) [3] , the training of BIM tools[2] [5] has been orientated to
proceeding actual building, including construction projects of the College in the
South Saigon and other projects such as social housing, offices, etc. are piloted for
students to apply BIM experimentally. Creating models in local-central network sys-
tem have assisted students to have experience when deploying BIM in architecture,
structure, water and electricity systems; quantity calculation, detection of conflicts
between disciplines, etc. Structural components were developed in line with LOD 300
[3]
.
The beginning results from the BIM tools training for the College are very positive,
expressed through the rate of post-graduate employment, through changes in teaching
and learning process.
Firstly, the high percentage of graduates with specialized jobs reflected the proportion
of students getting jobs with positive assessments from businesses. The College has
cooperated with 12 BIM consultants for placement. Through enterprise surveyed data,
after finishing the internship, 88% of apprentices have the capacity to work immedi-
ately after the internship process. Furthermore, BIM modelers are straightforwardly
able get jobs with income 1.5 times as high as the average income of existing techni-
cians with CADD skill only.
Secondly, BIM tools had also positive implication on student. 88% of surveyed stu-
dents using BIM tool (Revit) think that 3D models are visualized to insight under-
standing of projects structures. In addition, learners can develop themselves by chang-
ing their perception of learning, cooperation, personal responsibility in group activi-
ties; 85% of surveyed students said that learning BIM tool helps them to improve
communicating, sharing and teamworking.
Thirdly, the application of BIM tools not only equips students with modeling skills,
but also facilitates other subjects. Applying the architecture, structure and MEP disci-
plines of the BIM model vividly simulates and illustrates buildings, helps learners
visualize the components of a house, the structure, pipeline layout etc., thereby sup-
porting the statistics of volume of supplies and equipment.
Application of BIM tools in technician training... 1237

3 Determining precise target groups

According to experts [7], the positions of employees formed during the implementation
of BIM such as BIM Modeler / technician (Model Builder); BIM Specialist / technol-
ogist; Coordinator (BIM Coordinator) and Manager (BIM Manager). Through needs
survey in some partner companies, experts believe that in the four positions men-
tioned above, the need for the number of modeling specialists (BIM Modeler) at pre-
sent is 60%, the highest rate in the process of model establishment and analysis. This
position is also consistent with training at intermediate TVET level [7].
A BIM Modeler is a technician who carries out responsibility of creating and updating
BIM information models for different purposes such as produce design layouts, shops
drawing, visualization, build 3D library etc. Modelers are specialized in one subject or
multi-disciplinary modeling [7]. This target group requires quite similar skills, can be
upgraded quickly from the content of training traditional technicians. Therefore, BIM
Modeler is chosen as the main target groups of the College training courses.
The idea to develop competitive BIM modelers are now endorsed by a large number
of
Experts DQGFRPSDQLHV¶PDQDJHUV who are day by day facing with the lack of techni-
cians working in BIM models despite of a huge number of engineers graduated from
university every year. However, the current national-level BIM pilot training pro-
grams in Vietnam are now focusing on group of universities without any significant
engagement of TVET [2].

4 Clearly identify new competencies in the curriculum

The description of BIM technician¶VSRVLWLRQ mainly requires core competencies and

technical competencies [7].
The core competencies consist of communication, cooperation among project mem-
bers, ensuring collaboration between stakeholders, which are key factors for success.
In order to work in BIM process successfully, learners must be initially trained to
develop teamwork and collaboration skills such as communicating, transferring in-
formation and ideas, as well as improving the ability to utilize English. In addition,
students are to be coached to actively learn, apply and share new knowledge and skills
acquired at work.
The technical competencies demonstrate their professional ability to apply BIM suc-
cessfully. The course can reduce some of the existing skills concerning CAD etc.,
only focus on the ability to use tools that create information models for each specific
purpose such as developing component libraries; developing and testing information
model; making animation, extracting build quantity from the model [7].
The above competency groups are defined as the basic competency groups, suitable
for intermediate level training course within 1,5 to 2 years duration. Furthermore,
graduated learners are also enabled to obtain additional capabilities in the process of
working and lifelong self-learning process.
1238 T. L. Nguyen and T. X. Nguyen

5 "Classroom at enterprise" could be an effective solution.

According to experts the capacity to work with BIM can be developed in several con-
ducts such as formal education, vocational training or job training (on-the-job train-
ing) and do professional work [6]. In fact, the BIM modeler training process at the
College shows that the most important task is to create a learning environment that
simulates the real working environment. In order to offer a skilled BIM modeler after
a short period of 1,5 to 2 years, the course must provide students opportunity to prac-
tice in teamwork, working with authentic projects, simulating the process of project
and information exchange [4]. This process effectively helps students to get direct
access to the work environment after graduation
The combination of TVET and industry [4] is a key factor to strengthen the pedagogi-
cal capability of training institutions. Cooperation plan must be concretized by de-
tailed implementing plans, that allows students to participate in practical projects at
companies, helping students to strengthen their knowledge and skills learned at
school. Before being guided to the enterprise, students¶ knowledge and skills are syn-
thesized through a general project, helping students confidently work at the enterprise.
During the internship, the business appoints experts to guide them the culture, pro-
cesses and methods of doing business tasks in the enterprise. The College also engag-
es teachers to support students in a two-month internship. At the end of internship,
enterprises would comment, assess and mark points, participate in surveys and pro-
vide further suggestions.
In the next phase, the College are to conduct training courses with higher rate of in-
dustrial engagements. Learners should involve in placement at the commence to learn
business culture and code of conduct, etc. OTJ (on the job training) process should
increase toward 70% of training time under the form of placement where learners are
guided by experienced technician to conduct real projects. More authentic document
and practical building codes and standards related to BIM should be included in learn-
ing sources etc.
These activities aim at enhancing workforce of the business in process of creating
BIM products also effectively support the comprehensive skills of the students to
meet the demand of the industry.

6 Some challenges in implementing BIM tools training

The integration of BIM tools in training requires long-term, genuine investment with
time and rational cost. In fact, the process brings out some barriers in strengthening
teaching capacity; improve the IT system; and develop training curriculum that ap-
propriates to the entry level of TVET students.
Regarding innovating WHDFKHUV¶ capability, training courses were organized at the
college and enterprises. However, only about 30% trained teachers are qualified,
some of them have moved to businesses or universities. Visiting tutors from the in-
dustry have been invited, but training methods are also their barriers. This is resulted
from the sluggish change in the perception of teachers to a new and unfamiliar issue.
Application of BIM tools in technician training... 1239

Actually, teachers are working as consultants in their own businesses. However, the
majority of their businesses are small and medium-sized enterprises which still apply
the traditional methods; thus, BIM tools are found unnecessary for their job. Conse-
quently, 65% of the opinions of the surveyed teachers believe that there are still very
few businesses applying BIM. In reality, the application of BIM currently focuses on
a number of large corporations and some companies specializing in model processing.
Thus, it is still controversial about the popularity of BIM in reality, leading to a di-
lemma of changes.
Besides, in order to fully deploy BIM tools which are increasingly advanced, the ex-
isting IT system needs to be regularly upgraded and synchronized. The investment
can only be taken by small steps, thus, supports from industries are highly appreciat-
ed. In addition, due to *RYHUQPHQW¶V education and training policy, the entry level
qualification with post-secondary is an immense challenge for the College. 44% of
surveyed students said that learning BIM tools application is more difficult than other
applications; teachers also claim BIM tools for more of effort than other applications.
It is due to the complexity of the interface, three-dimensional space cognition, team-
work skills etc.
In order to slowly clear these barriers, beside endless efforts of TVET, more supports
from national BIM development programs are expected to stimulate the application of
BIM in TVET of which changing mindset of stakeholders on the role of TVET and
adequate resources for long-term cooperation between TVET and the construction
industry are essential.

7 Conclusion

Building Information Modeling (BIM) [1] is increasingly widely used in construction

operation and management. The development of BIM, therefore requires a large num-
ber of human resources with new job positions and new competencies compared to
the previous technology.
The BIM applications incrementally applied in training have strong positive implica-
tion on both learners and teachers in TVET. It has changed the training methods, re-
quiring early determination of positions learners would take charge of the enterprise
with BIM application as target groups; requiring clear identification of new concreted
competencies in the training content that straightforwardly enable learners to obtain
successful career objectives in the fields of BIM.
Especially, the process also emphasizes the cooperation between businesses and
schools as the most effective solution to provide competitive employees that could
effectively utilizing BIM for construction industries. This is also to highly appreciate
the commitments and willingness from BIM consultants to joint training program for
BIM modelers.
This article also calls for further recognition and effective support TVET from the
JRYHUQPHQW¶V%,0GHYHORSPHQWSURJUDPVto clear the barriers in BIM modeler train-
ing. That is to say, the training of BIM modeler when being rationally analyzed in
TVET, showing the necessity and effectiveness, affirming the role and position of the
1240 T. L. Nguyen and T. X. Nguyen

TVET system in human resource development working within the field of building
information modeling./.

References:
1. Government, Approving the Scheme on application of construction information model
(BIM) in construction and management of project operation, 2500 / QD-TTg, (2016).
2. Ministry of Construction, Announcing the program of training and retraining framework
applying Building Information Model (BIM) in the pilot phase, 1056 / QD-BXD, (2017).
3. Ministry of Construction, Announcing the guidelines for temporary application of Build-
ing Information Model (BIM) in the pilot phase, 1057 / QD-BXD, (2017).
4. References to train source lecturers on BIM at Hanoi University of Construction in
1/2018.
5. Nguyen Bao Ngoc, Nguyen The Quan, "Frame Bim training program for building prac-
tices: patterns and recommendations", Journal of Construction Economics (ISSN 1859-
4921) No.02/2018. (2018)
6. B, Succar and W, Sher, "A competency knowledge-base for BIM learning", Australasian
journal of construction economics and building - Conference series Vol. 2 No. 2 (2014)
7. Nguyen Nhu Trang, "Building capacity to work with BIM in vietnam", Journal of Con-
struction Science and Technology No. 01-2018 (2018).
 Bridge assessment for PSC Girder Bridge using Digital
Twins Model

NgocSon Dang1[0000-0002-3341-3034], ChangSu Shim2

1 Ph.D. student, ChungAng University, Seoul, South Korea
2
Professor, ChungAng University, Seoul, South Korea
(*)dangngocson@cau.ac.kr

Abstract. Bridge maintenance nowadays is no longer a "reactive activity" to the

severe degradation or unexpected disaster. Aging and deterioration are inevitable,
sometimes arises exponentially if maintenance strategy is neglected. Leading to
structural failure, or worse case is the functional failure and seriously threaten the
safety, interrupt the public transportation system. This paper proposes a new con-
cept for preventive maintenance strategy for existing aged PSC Girder Bridge
using a digital twin model. Digital Twins concept is mainly based on the use of
parallel models: digital twin model (DTM), reality twin model (RTM) and me-
chanical twin model (MTM). DTM authoring is introduced using an integrated
BIM mod-el, while RTM is verified by a reversed engineering surface model
based on 3D scanning data. The RTM is generated continuously during bridge
lifecycle in or-der to make a field-verified replica of the structure, aims to records
all the bridge damage time by time. The mechanical twin model is derived di-
rectly from DTM through the interoperability of BIM model and adjusted by
damage information from RTM. Discussion on bridge assessment model is in-
duced based on the supposed analysis model and deterioration history, significant
support for the decision making team in order to make a long-term strategy for
bridge maintenance task.

Keywords: DTM, BIM, Preventive Maintenance, Reality Model, 3D Scanning.

1 Introduction

Nowadays many developed countries are paying significant effort for a well maintain-
ing the service of their transportation system. Thousands of bridge almost reach or be
expired the designed life and arise the demand in term of maintenance or rehabilitation
job much urgent than ever. It is necessary to develop a new revolution in term of pro-
active strategy, so-called "preventive maintenance". Several bridge maintenance sys-
tem (BMS) for the existing bridge is proposed and discussed (Shim et al. 2017, Dang
et al. 2018).
The deterioration assessment of bridge has been extensively studied with a variety
of approaches and discussion. During bridge service life, it is needed to focus on the
lifetime reliability index, especially the material deterioration model according to time
(Saydam & Frangopol, 2011). Moreover, the prediction model for the evolution of the

© Springer Nature Singapore Pte Ltd. 2020 1241

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_199
1242 N. S. Dang and C. S. Shim

reliability of deteriorating structures under maintenance actions is introduced. The per-

formance of a deteriorating structure can be predicted using a probabilistic framework
based on lifetime performance indicators, then decide the adequate maintenance strat-
egy (Yang, Frangopol, Neves, 2006).
The traditional 2D-based methodology is no longer suitable for collaborative work-
ing with different stakeholders. The main limitations are the implicit errors from dif-
ferent sources accompany data incompatibility during the transfer process. In this re-
gards, BIM technology can dramatically enhance the performance of this collaboration.
Object-oriented rule-based models are useful for communication and maintain all the
related information entirety. The significant benefits of BIM lie in its ability to probably
use both analysis and evaluation programs during the design process. Based on the re-
quirement in terms of the target-purpose type as well as the level of detail, the infor-
mation exchange between the original model and the analysis/assessment model can be
implemented through geometric and semantic mappings. On another hand, the interop-
erability between existing commercial tools needs to be taken into account and en-
hanced through digital software libraries for model development and transformation.

2 Digital Twin for maintenance

Digital twin concept has been around for two last decades, and always named as one of
the top strategic technology trends by time. In this concept, a Digital Twin Model
(DTM) is paired with a physical entity and then represented for its existence. In other
word, it is a digital replica or virtual counterparts for a real entity which can be assets,
processes, systems or even services. After pairing with physical worlds, it allows the
features of the monitoring system and data analyzing, aims to catch up and head off
problems before they even occur. Uncertainty risk, therefore, can be prevented, and
even new challenges can be assumed to plan for the future by simulation. In this paper,
it is combined between "a physical 3D model with damage records" and "a supposed
analysis model based on that deterioration history". It is probably the parallel aspects
of one overall system, which work derivatively, but simultaneously and interactively
each other.

Figure 1. DTM concept for bridge maintenance

Bridge assessment for PSC Girder Bridge using Digital... 1243

3 Model authoring

3.1 DTM

Digital model for bridge maintenance purpose is proposed to use an integrated model
including 3D model from design, calculation result from analysis task, and all the rec-
ords during bridge service lifecycle. Start with the as-built document of the existing
bridge, a data schema for 3D information model which oriented to maintenance purpose
is strictly required following the employer’s information requirement (or tender docu-
mentation). As following object-oriented approach, the structures of bridge is specified
into inventory system and defined by individual ID. Naming convention should be
taken into account in order to well manage the member information in the entire system.
The common data environment of information system has two main components: “at-
tribute” which includes the physical information of the structure, and “archive” which
consist all the related information along with history of the bridge. Bridge alignment
also needs to be taken into account, not only for the modeling purpose but also for
analysis task on later. Since the data schema and bridge alignment is presented, the 3D
geometry model can be easily generated using parametric modeling concept with the
aid of an open-source application programming interface.

Figure 2. Digital alignment-based parametric model

1244 N. S. Dang and C. S. Shim

3.2 RTM

Basically the reality twin model is a reversed 3D-surface model, which created through
the 3D scanning procedure, to capture the current status of the bridge. Lateral and top
surface model is created by photo scanning using the drone, while bottom surface model
based on laser scanning cloud data. The GPA data on another hand can be attached onto
the top surface model.

Figure 3. Reversed-surface model

Inspection data from the scanning procedure is automatically converted into tech-
nical damage reports based on the photo processing and photo tracing technology, and
directly updated onto the DTM.

Figure 4. Fragmented surface damage and mapping to the DTM as archive data
Bridge assessment for PSC Girder Bridge using Digital... 1245

4 Federated model for assessment

DTM and RTM are overlapped based on predefined marks, which is included in both
the visual model and attached to the real bridge before 3D scanning procedure. The
overlapped model also called “initial model”, which can be considered as the physical
3D model including damage records, and represent the real current status model at the
beginning of the maintenance project. The mechanical model is derived directly from
the initial model through the interoperability of BIM solution which created the DTM,
which can maintain the integrity of the 3D physical model. Different from the analysis
model for design check in design procedure, this is a supposed model for assessing the
future behavior of the bridge.

Figure 5. Interoperability of DTM and supposed analysis model

The behavior of the damage effect model leaves the basic assumption for the analysis
task. Current deterioration or damage of bridge changes the input structural parameters
of the analysis model. For example, the modulus of elasticity and moment of inertia,
are reduced. As long as the damage is detected, for instance, the cracks, material deg-
radation, corrosion of steel elements are considered as the reduction of the structural
parameters. A chain of analysis cases is conducted according to different combined
load cases. Thereafter, the results are compared and discussed, the future behavior of
the bridge also suggested. Furthermore, in comparison with the real behavior of many
other similar bridges, the bridge management team can timely figure out certain repair
or strengthening measures for the bridge. Data management system on the network
level is also an important issue for the DTM in order to maintain an automated and
timely update of information in the common data environment.
1246 N. S. Dang and C. S. Shim

5 Conclusion

This paper proposes a new concept for the bridge maintenance work which can provide
much improvement in term of a long-term strategy. DTM is suggested to use an inte-
grated BIM model, through a digital rule-based model and the open-source BIM plat-
form. On the other hand, RTM is based on the advancement of 3D scanning technology
and reverse engineering tools. Hence the federated model using DTM/RTM provides
an information-rich resource for maintenance purpose. During bridge lifecycle, the in-
itial model and deteriorated model are continuously curated. The accumulated damage
and repair history is always kept-tracking, significantly supports the project team to
timely react again unexpected situation, as well as creates a premise to orient a long-
term strategy for bridge maintenance. A pilot system including a full application of
BMS (for a real bridge in Korea) has been developing and testing for a year based on
the suggested concept and surveyed information requirements. Hence, maintenance
performance was significantly improved due to the federated models.

Acknowledgements

This research was supported by a grant “Development of life-cycle engineering tech-

nique and construction method for global competitiveness upgrade of cable bridges
(16SCIP-B119960-01)” from Smart Civil Infrastructure Research Program funded by
MOLIT and KAIA.

References
1. Shim, C.S., Kang, H.R., Dang, N.S., Lee, D.K: Development of BIM-based bridge
maintenance system for cable-stayed bridges. Smart Structures and Systems, 20(6):
697-708 (2017).
2. Dang, N.S., Shim, C.S.: BIM authoring for an image-based bridge maintenance
system of existing cable-supported bridges. IOP Conference Series: Earth and En-
vironmental Science, 143(1) (2018).
3. Dang, N.S., Kang, H.R., Lon, S., Shim, C.S.: 3D digital twin models for bridge
maintenance, Proceedings of 10th International Conference on Short and Medium
Span Bridges, Quebec city, Quebec, Canada (2018).
4. Saydam, D., Frangopol, D.M.: Time-dependent performance indicators of dam-
aged bridge superstructures. Engineering Structures, 33(9): 2458-2471 (2011).
5. Yang, S. I., Frangopol, D.M., Neves, L.C.: Optimum maintenance strategy for de-
teriorating bridge structures based on lifetime functions. Engineering Structures,
28(2): 196-206 (2006).
6. Shim, C.S., Dang, N.S., Lon, S., Jeon, C.H.: Development of a bridge maintenance
system for PSC bridges using 3D digital twin model, Structure and Infrastructure
Engineering. (Accept on 26-Feb-2019).
$SSOLFDWLRQRI%XLOGLQJ,QIRUPDWLRQ0RGHOOLQJ([WHQGHG
WUDFNLQJWHFKQLTXHDQG$XJPHQWHG5HDOLW\LQ%XLOGLQJ
2SHUDWLQJ0DQDJHPHQW

7KDQK%DQ1*8<(1$QK%LQK75$1 0DQK7XDQ1*8<(19DQ+RDQ
3+$0.KXRQJ/(1*8<(1

1DWLRQDO8LYHUVLW\RI&LYLO(QJLQHHULQJ*LDL3KRQJ+DQRL9LHWQDP
8QLYHUVLW\RI7UDQVSRUW7HFKQRORJ\7ULHX.KXF+DQRL9LHWQDP

ELQKWD#QXFHHGXYQ

$EVWUDFW7KLVSDSHUSUHVHQWVDFRPELQDWLRQRI$XJPHQWHG5HDOLW\$5([
WHQGHG7UDFNLQJ7HFKQLTXH(77DQG%XLOGLQJ,QIRUPDWLRQ0RGHOOLQJ%,0WR
LQWHUDFWZLWKEXLOGLQJLQIRUPDWLRQLQUHDOWLPHRQWKHFRQVWUXFWLRQVLWH8VLQJWKLV
PHWKRGZHKDYHSURJUDPPHGDQDSSOLFDWLRQUXQQLQJLQDPRELOHGHYLFHWRKHOS
PDQDJHDQGRSHUDWHWKHEXLOGLQJIDFLOLW\V\VWHPV,QWKLVDSSOLFDWLRQWKHEXLOGLQJ
PRGHODQGLWVLQIRUPDWLRQDUHVWDFNHGDQGDXJPHQWHGRQWKHPRELOHGHYLFHFDP
HUDZKHQSURMHFWLQJRQWKHWDUJHWLPDJHPRXQWHGRQWKHEXLOGLQJ(77DOORZV
H[WHQGLQJWKHREMHFWGLVSOD\HGRQWKHVFUHHQHYHQZKHQWKHPRELOHGHYLFHFDP
HUDQRORQJHUFDSWXUHVWKHWDUJHWLPDJH%DVHGRQWKLVSURFHVVHQJLQHHUVFDQLQ
WHUDFWZLWKWKHVRIWZDUHWRDGGFKDQJHWKHLQIRUPDWLRQQHHGHGRIIDFLOLWLHVRQ
FRQVWUXFWLRQ VLWH VXFK DV VL]H PDQXIDFWXUHU GDWH RI PDLQWHQDQFH GDWH RI UH
SODFHPHQWGHYLFHOLIHF\FOHHWF

.H\ZRUGV$XJPHQWHG5HDOLW\%XLOGLQJ,QIRUPDWLRQ0RGHOOLQJ%XLOGLQJ2S
HUDWLQJ0DQDJHPHQW([WHQGHG7UDFNLQJ7HFKQLTXH

 ,QWURGXFWLRQ

%,0LVDSURFHVVUHODWHGWRLQIRUPDWLRQHQULFKPHQWRIEXLOGLQJFRPSRQHQWV,WLVD
FRQWLQXRXVSURFHVVWKURXJKRXWWKHSURMHFWOLIHF\FOHZKHUHE\EXLOGLQJLQIRUPDWLRQLV
XWLOL]HGDQGXVHGWKURXJKVWDNHKROGHUV7KLVWHFKQRORJ\KDVVLJQLILFDQWO\FKDQJHGWKH
FRQVWUXFWLRQSURFHVVLQFUHDVLQJWKHSURMHFW
VYLVLELOLW\%,0LVLQWHJUDWHGLQWRWKHZRUN
IORZWRSHUIRUPWDVNVVXFKDVGHVLJQDQDO\VLVFRVWPDQDJHPHQWLQFUHDVHGFROODERUD
WLRQDPRQJSURMHFWPHPEHUVIURPYDULRXVGLVFLSOLQHVSURYLGLQJWKHQHFHVVDU\LQIRU
PDWLRQIRUEXLOGLQJRSHUDWLRQSKDVHDQGPDLQWHQDQFHSKDVHRIWKHIDFLOLW\V\VWHP%,0
LVFRQVLGHUHGRQHRIWKHPDLQVWUHDPWUHQGVRIWKHFRQVWUXFWLRQLQGXVWU\,QSDUWLFXODU
WKHFRVWRIPDQDJLQJDQGRSHUDWLQJEXLOGLQJVDFFRXQWVIRURIWKHWRWDOFRVWRIWKH
ZKROHSURMHFWLQFOXGLQJGHVLJQFRQVWUXFWLRQPDQDJHPHQWDQGRSHUDWLRQFRVWV>@
7KHUHIRUHLWLVQHFHVVDU\WRUHVHDUFKDQGDSSO\%,0IRUFRQVWUXFWLRQPDQDJHPHQWDQG
RSHUDWLRQSKDVHV

© Springer Nature Singapore Pte Ltd. 2020 1247

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_200
1248 T. B. Nguyen et al.

$5WHFKQRORJ\>@DOORZVDXJPHQWLQJLPDJHVRQWKHFDPHUDRIDFRPSXWHURUPR
ELOHGHYLFH:KHQXVLQJ$5LWLVRIWHQQHFHVVDU\WRDWWDFKWRDWDUJHWLPDJHRUDWWDFK
WR*36FRRUGLQDWHVDQGPRELOHGHYLFHVHQVRUV$5LVZLGHO\XVHGLQHGXFDWLRQLQWHULRU
DQGH[WHULRUDUFKLWHFWXUDOVKRZVJDPHVQRWDEO\WKH3RNpPRQ*RUHOHDVHGE\1LDQWLF
7KH FRPELQDWLRQRI $5 DQG %,0 WHFKQRORJ\ZLOO FUHDWHDQ DSSOLFDWLRQWKDW HQDEOHV
XVHUVWRYLHZDQGLQWHUDFWZLWK'PRGHOVHPEHGGHGLQUHDOZRUOGDQGUHDOWLPHHQYL
URQPHQWVLQYDULRXVSHUVSHFWLYHV(QKDQFLQJWKH%,0PRGHOIRUUHDOFRQVWUXFWLRQZLOO
KHOSVLPSOLI\WKHPDQDJHPHQWDQGRSHUDWLRQSKDVHIRUVWDNHKROGHUV)RUFRQVWUXFWLRQ
UHODWHGHGXFDWLRQLWDOORZVVWXGHQWVWRVHSDUDWHHDFKFRPSRQHQWRIWKHPRGHOGLVSOD\
' PRGHOV UHODWHG YLGHRV ZKHQ SXWWLQJ FDPHUDV RQ VSHFLILF LPDJHV LQ ERRNV 7KDW
KHOSVVWXGHQWVDEVRUEWKHOHVVRQPRUHHDVLO\DQGHIIHFWLYHO\
7KHUHDUHVHYHUDOVWXGLHVFRPELQLQJ$5DQG%,0LQWKHFRQVWUXFWLRQLQGXVWU\$5
DSSOLFDWLRQLVXVHGIRUFRQVWUXFWLRQVLWHPRQLWRULQJDQGGRFXPHQWDWLRQZLWKDFRPEL
QDWLRQRIDHULDOYLVLRQDQGDQ$5PRELOHLQWHUIDFHWKDWDOORZVUHOHYDQWLQIRUPDWLRQWR
EHDFFHVVHGGLUHFWO\RQVLWH>@$XJPHQWHG5HDOLW\FDQEHXVHGLQLQWHULRUGHVLJQ>@
DQGDUFKLWHFWXUDOSODQQLQJ>@RULQWKHPDQDJHPHQWRIFLYLOLQIUDVWUXFWXUHV\VWHPV>@
,WDOORZVXVHUVWRSODFHWKHLURZQWDJVGLUHFWO\RQWKHVLWHDQGEULQJWKLVLQIRUPDWLRQ
EDFNWRWKHGDWDVWRUHWKDWUHPRWHJURXSPHPEHUVFDQDFFHVV
,Q WKH UHFHQW SDSHU >@ ZH SURSRVHG WR FRPELQH $5 DQG %,0 LQ PDQDJLQJ DQG
RSHUDWLQJWKHPHFKDQLFDODQGHOHFWULFDO0 (V\VWHPLQEXLOGLQJV,QWKLVSDSHUZH
SURSRVHDQH[WHQVLYHWUDFNLQJWHFKQRORJ\WKDWDOORZVWKH%,0PRGHOWREHDXJPHQWHG
RQWKHPRELOHGHYLFH
VFDPHUDHYHQZKHQOHDYLQJWKHWDUJHWLPDJH7KLVLVDQLPSRUWDQW
VWHSWRH[SDQGWKHVFRSHRIDSSOLFDWLRQRI%,0DQG$5FRPELQHGVRIWZDUH7KLVDSSOL
FDWLRQFDQEHDSSOLHGWRWKHVWDJHRIEXLOGLQJRSHUDWLQJPDQDJHPHQWLQJHQHUDODQGIRU
WKH0 (V\VWHP

 ([WHQWHG7UDFNLQJ7HFKQLTXH

(77 DOORZVNHHSLQJ WKH LGHQWLILFDWLRQSRVLWLRQ DQG PRYHPHQW RI WKHPRGHO DF

FRUGLQJWRWKHURWDWLRQDQJOHDQGWKHPRELOHGHYLFHZLWKRXWUHORDGLQJWKHPRGHOZKHQ
WKHFDPHUDOHDYHVWKHWDUJHWLPDJH>@7KLVWHFKQLTXHXVHV'HYLFH7UDFNHUWRLPSURYH
WUDFNLQJSHUIRUPDQFHDQGPDLQWDLQWKLVWUDFNLQJSURFHVVHYHQZKHQWKHWDUJHWLPDJHLV
QRORQJHULQWKHILHOGRIYLHZ)29RIWKHFDPHUD7KHDXJPHQWZLOOVWLOOPDLQWDLQWKH
SRVLWLRQRIWKHPRGHOLQWKHUHDOZRUOGDQGPDWFKWKHRULJLQDOUHIHUHQFHIUDPHGHILQHG
E\WKHWDUJHW(77XVHVWKLVWHFKQLTXHWRGHWHUPLQHWKHORFDWLRQRIWKHWDUJHWE\YLVXDOO\
PRQLWRULQJWKHHQYLURQPHQWDQGXVLQJRWKHUHQYLURQPHQWDOLQIRUPDWLRQ)RUWKLVSXU
SRVH(77ZLOOFUHDWHDPDSDURXQGWKHWDUJHWDQGWKHHQYLURQPHQWDOFKDUDFWHULVWLFV
ZLOOEHDGGHGWRWKHWDUJHWLPDJHGDWDEDVHDQGWKHREWDLQHGFDPHUDIUDPHVZLOOFRP
SDUHZLWKWKHQHZGDWDEDVH>@
Application of Building Information Modelling... 1249


D,PDJHWDUJHWLQ)29  F,PDJHWDUJHWLVRXWVLGH)29
E$XJPHQWHG5HDOLW\0RGHO          G([WHQGHG7UDFNLQJ
)LJ,PDJHWDUJHWLQ)29DQGRXWVLGH)29RIGHYLFH
)LJDVKRZVWKDWZKHQWKHFDPHUDLVSURMHFWHGGLUHFWO\RQWKHWDUJHWLPDJHLQWKH
ILHOGRIYLHZWKHUHVXOWGLVSOD\VDSHUIHFWPRGHORIEXLOGLQJV)LJE:KHQWKHXVHU
PRYHVXSWRWKHWRSRIWKHEXLOGLQJWRVHHPRUHGHWDLOVWKHWDUJHWLPDJHLVDFFLGHQWDOO\
RXWVLGHWKHFDPHUD
VYLHZDUHDDQGWKHPRGHOLVORVW)LJF,Q)LJGZKHQPRYLQJ
WKHFDPHUDXSZDUGVWKHWDUJHWLPDJHLVRXWRI)29EXWWKH(77WHFKQRORJ\VWLOOUHWDLQV
WKH HQWLUH PRGHO DFFRUGLQJ WR WKH RULJLQDO UHIHUHQFH IUDPH WR KHOS XVHUV FRQWLQXH WR
H[SHULHQFHWKHPRGHOHIIHFWLYHO\

 &RPELQLQJSURFHVVIRU%,0$5DQG([WHQWHG
7UDFNLQJ

,QWKHSUHYLRXVSDSHU>@ZHSURSRVHGWKHSURFHVVRIFRPELQLQJ%,0DQG$5WHFK
QRORJ\LQPDQDJLQJDQGRSHUDWLQJ0 (V\VWHPVLQEXLOGLQJV7KLVSDSHUH[WHQGVWKH
SURFHVVE\FRPELQLQJ(77WRIXUWKHUFODULI\WKHSURFHVVRIFUHDWLQJDWDUJHWLPDJHGD
WDEDVHDQGDGGLQJWKHRSWLRQWRVWRUHWKH%,0PRGHORQWKHVHUYHU$QHZSURFHVVLV
SURSRVHGLQ)LJ


1250 T. B. Nguyen et al.

)LJ7KHSURFHVVLQFRUSRUDWHV%,0DQG$5
:HSURJUDPPHGDPRELOHDSSOLFDWLRQWRLQWHJUDWHWKHSURFHVV7KHDSSOLFDWLRQDO
ORZVYLVXDOL]LQJ'PRGHOVDQGREVHUYLQJWKHLQIRUPDWLRQRIWKHEXLOGLQJULJKWRQWKH
PRELOHGHYLFHEDVHGRQWKHLGHQWLILFDWLRQRIDWDUJHWLPDJHPRXQWHGRQWKHUHDOEXLOG
LQJ7KHSURFHVVLVH[SODLQHGLQWKHIROORZLQJVWHSV6WHS'HVLJQLQJIDPLOLHVDQG
GHFODULQJVKDUHGSDUDPHWHUVLQ%,06WHS&UHDWH%,0PRGHOVDQGUHSRUWLQWHJUDWHG
LQIRUPDWLRQRI0 (V\VWHP6WHS,)&,QGXVWU\)RXQGDWLRQ&ODVVHVILOHRXWSXW
6WHS9XIRULD6'.6RIWZDUH'HYHORSPHQW.LWORDGV,)&ILOHV6WHS,QLWLDOL]H
GDWDVHWIURPWDUJHWLPDJH6WHS0DWFKWKHLGHQWLILFDWLRQLPDJHIURPWKHGHYLFHFDP
HUDZLWKWKHSUHYLRXVO\FUHDWHGLPDJHGDWDZHGHVFULEHGLQGHWDLOWKHFUHDWLRQSURFHVV
RQ WKH PRGHO DQG WKH FRUUHVSRQGLQJ ,)& ORDG 6WHS   'LVSOD\WKH EXLOGLQJ LQIRU
PDWLRQPRGHORQWKHGHYLFH6WHSXVH([WHQGHG7UDFNLQJWHFKQLTXHZKHQWKHWDUJHW
LPDJHLVQRORQJHULQWKHFDPHUD
VYLHZDUHD6WHS8SGDWH%,0PRGHOFKDQJHVWR
WKHVHUYHU
7KLV DSSOLFDWLRQ KHOSV XVHUV WR KDYH D FOHDU SUDFWLFDO H[SHULHQFH RI WKH EXLOGLQJ
PRGHO7KHSURMHFWVWDNHKROGHUVFDQHYDOXDWHDQGRSWLPL]HWKHGHVLJQDWDQHDUO\VWDJH
DQG WKLV %,0 PRGHO FRQWLQXHV WR EH GHYHORSHG LQ WHUPV RI LQIRUPDWLRQ WR VHUYH WKH
PDQDJHPHQW RI WKH HQJLQHHULQJ V\VWHP 5HJDUGLQJ WKH %,0 PRGHO LW LV FRPSOHWHO\
FUHDWHGIURPGLIIHUHQWVRIWZDUHDQGHDVLO\H[SRUWHGWR,)&ILOHV8VHUVFDQPRYHWKH
FDPHUDLQDURRPWRJHWUHDOZRUOGH[SHULHQFHVDWHYHQPRUHDQJOHVLQVWHDGRIKDYLQJ
WRNHHSWKHFDPHUDLQDFHUWDLQGLUHFWLRQ

 $SSOLFDWLRQGHYHORSPHQWDQGWHVWLQJ

7REXLOGWKHDSSOLFDWLRQZHXVH9XIRULD6'.8QLW\'(QJLQHIRUWKHSXUSRVHRI
FUHDWLQJPXOWLSODWIRUPSURGXFWVVXFKDVL26$QGURLGDQG:LQGRZV,QDGGLWLRQZH
DOVRXVHWKH/HDQ7RXFK3OXJLQWRKDQGOH'WRXFKVZLSHDQGURWDWHHYHQWV1RWHWKH
LQLWLDOVWHSFRQILJXUDWLRQLQ8QLW\DVIROORZV(QDEOH$5YLD)LOH!%XLOG6HWWLQJV
!3OD\HU6HWWLQJV,WPLJKWEHQHFHVVDU\WRGRZQORDGDQGLQVWDOO9XIRULD6XSSRUW
5HPRYH WKH PDLQ FDPHUD IURP WKH VFHQH  $GG $5 FDPHUD YLD *DPH2EMHFW !
9XIRULD!$5&DPHUD$GGLPDJHWDUJHWYLD*DPH2EMHFW!9XIRULD!,PDJH:H
FRQGXFWHGDWHVWWRLGHQWLI\DQGHQKDQFHDKLJKULVHEXLOGLQJXVLQJD'GUDZLQJDVD
WDUJHWLPDJH$IWHUXVLQJ9XIRULD7DUJHW0DQDJHURQWKHFORXGSODWIRUPZHREWDLQHG
GDWDZLWKVWDUUDWLQJZLWKWKLVWDUJHWLPDJH$VDQRWHGXULQJWKH'PRGHOGHYHORS
PHQWLWLVQHFHVVDU\WRWDNHLQWRDFFRXQWWKHUDWLREHWZHHQWKHYLUWXDOPDUNHUVL]HWKH
VmarkerDQGWKHDFWXDOPDUNHUVL]HRmarkerDQGWKHUDWLRRIYLUWXDOZRUOGVL]HVworldDQG
WKHUHDOZRUOGVL]HRworldPXVWEHWKHVDPH>@LQSURSRUWLRQ


7KHWHVWVRQRXUDSSOLFDWLRQDUHDVIROORZV)LJVKRZWKHDFWXDOVWUHQJWKHQLQJRI
WKH0 (V\VWHPLQWKH%,0PRGHOZKHQWKHFDPHUDLVSURMHFWHGRQWKHZDOOLQ5RRP
$DWWKH8QLYHUVLW\RI&LYLO(QJLQHHULQJ+DQRL9LHWQDPDQGLWKHOSVWRGLVSOD\
Application of Building Information Modelling... 1251

WKHQHFHVVDU\LQIRUPDWLRQRIWKHHOHFWULFZLUHWXEHDQGDLUFRQGLWLRQHUWXEHDQGZDWHU
SLSHVZKHQXVHUVWRXFKWKH'PRGHO([WHQGHGWUDFNLQJWHFKQLTXHKHOSPRYHWKHFDP
HUDDURXQGWKHPRGHOZKHQWKHFDPHUDOHDYHVWKHWDUJHWLPDJH

)LJ'LVSOD\SRZHUOLQHVDQGDLUFRQGLWLRQHUVDQGZDWHUSLSHVKLGGHQLQZDOOVWRXFKWR
GLVSOD\LQIRUPDWLRQ
7KHDSSOLFDWLRQKDVEHHQWHVWHGDQGZRUNVZHOOZLWKPRELOHGHYLFHVZLWKPLQLPDO
FRQILJXUDWLRQDVIROORZVLQFKVFUHHQVL]H*%5DPPHPRU\RIPRELOHGH
YLFHV4XDOFRPP6QDSGUDJRQ&38SURFHVVRU4XDGFRUH*+]RSHUDW
LQJV\VWHPYHUVLRQ$QGURLGL2603FDPHUDDQGVWDUUDWLQJWDUJHW
LPDJHIURP9XIRULD7DUJHW0DQDJHUIRUWKHEHVWLGHQWLW\TXDOLW\+RZHYHUWKHUHLVD
OLPLWDWLRQZKHQVWRULQJWKH%,0PRGHORQWKHVHUYHUFRPSDUHGWRWKHGHYLFHWKDWDIWHU
GRZQORDGLQJ WKH PRGHO WR WKH GHYLFH YLD WKH LQWHUQHW LW LV GLIILFXOW WR UHIHUHQFH WKH
FRUUHFW SRVLWLRQ RI WKH PRGHO RYHUODSSLQJ WKH WDUJHW LPDJH OHDGLQJ WR DQ XQUHDOLVWLF
H[SHULHQFH7KHUHIRUHWKLVVROXWLRQVKRXOGEHDSSOLHGIRUWKHSXUSRVHRISURYLGLQJPRUH
LQIRUPDWLRQDQGLQVWUXFWLRQVIRUXVHUDWKHUWKDQDXJPHQWLQJWKHPRGHOWKURXJKWKHWDU
JHWLPDJH,QDGGLWLRQWRLPDJHUHFRJQLWLRQWHFKQRORJ\WKH2EMHFW5HFRJQLWLRQWHFK
QRORJ\XVHV9XIRULD2EMHFW6FDQQHU926FRPSRQHQWWRVFDQSK\VLFDOREMHFWVIURP
ZKLFKWRGHWHFWDQGFROOHFWVDPSOHLPDJHVFDSWXUHSDWWHUQVDURXQGDQREMHFWDQGFRY
HULQJDQREMHFWZLWKDSHUIHFWO\PDWFKHG'PRGHO,WLVDOVRDQLGHQWLILFDWLRQWHFKQRO
RJ\WKDWLVZLGHO\XVHGWRSURYLGHHQKDQFHGLQIRUPDWLRQRQUHDOREMHFWVVXFKDVFRP
SRQHQWVLQWKHHOHFWULFDODQGPHFKDQLFDOV\VWHPVLQWKHEXLOGLQJ

 &RQFOXVLRQV

7KHSDSHUSURSRVHGDSURFHVVWKDWLQFRUSRUDWHV%,0DQG$5WHFKQRORJ\DQG(77
6LQFHWKHQWKHDXWKRUVKDYHEXLOWDQDSSOLFDWLRQWKDWUXQVRQPRELOHGHYLFHVWRPDQDJH
EXLOGLQJVGLUHFWO\LQWKHILHOGVXFKDVPDQDJLQJPDLQWHQDQFHDQGUHSODFLQJHTXLSPHQW
PRUHHDVLO\%DVHGRQWKH(77WKHFRQVWUXFWLRQLQIRUPDWLRQFRQWLQXHVWREHGLVSOD\HG
RQWKHPRELOHGHYLFHVFUHHQHYHQZKHQWKHFDPHUDLVQRWSURMHFWHGRQWKHWDUJHWLPDJH
PRXQWHGRQWKHEXLOGLQJ7KHH[SHULHQFHRIWKHDFWXDOSURMHFWKDVSURYHGWKHVWDELOLW\
DQGXWLOLW\RIWKHPRELOHDSSOLFDWLRQWKDWZHKDYHSURJUDPPHG3URJUDPPLQJLQWHJUDWHV
DSSOLFDWLRQVRQVSHFLDOL]HGJODVVHVOLNH+RKROHQV0LFURVRIWZLOOEHRXUQH[WGHYHORS
PHQWGLUHFWLRQV
1252 T. B. Nguyen et al.

$FNQRZOHGJPHQW

7KLVUHVHDUFKLVVXSSRUWHGE\0LQLVWU\RI&RQVWUXFWLRQRI9LHWQDPXQGHUJUDQWQXP
EHU5'

5HIHUHQFHV
&KXFN(DVWPDQ3DXO7HLFKRO]5DIDHO6DFNV.DWKOHHQ/LVWRQ%,0+DQGERRN 
,QF1HZ-HUVH\
$XJPHQWHG5HDOLW\IRU0RELOH'HYLFHV,QVWDOOHG%DVHRI0RELOH$XJPHQWHG5HDOLW\$SSVWR
5HDFK  %LOOLRQ E\  $YDLODEOH DW KWWSVZZZWUDFWLFDFRPQHZVURRPSUHVV
UHOHDVHVLQVWDOOHGEDVHRIPRELOHDXJPHQWHGUHDOLW\DSSVWRUHDFKELOOLRQE\
>$FFHVVHG0D\WK@
 6 =ROOPDQQ & +RSSH 6 .OXFNQHU & 3RJOLWVFK + %LVFKRI DQG *
5HLWPD\U $XJPHQWHG UHDOLW\ IRU FRQVWUXFWLRQ VLWH PRQLWRULQJ DQG
GRFXPHQWDWLRQ  YRO  QR  SS

 6LOWDQHQ 6 :RRGZDUG &$XJPHQWHG LQWHULRUV ZLWK GLJLWDO FDPHUD LPDJHV
+REDUW$XVWUDOLD&53,73LHNDUVNL:SS

2OVVRQ76DYLVDOR$+DNNDUDLQHQ0:RRGZDUG&8VHUHYDOXDWLRQRIPRELOHDXJPHQWHG
UHDOLW\LQ DUFKLWHFWXUDO SODQQLQJH:RUN DQG H%XVLQHVV LQ $UFKLWHFWXUH (QJLQHHULQJ DQG &RQ
VWUXFWLRQ*XGQDVRQ* 6FKHUHU5HGV 5H\NMDYLN,VODQG-XO
SS
 $ + %HK]DGDQ 6 'RQJ DQG 9 5 .DPDW $XJPHQWHG UHDOLW\
YLVXDOL]DWLRQ $ UHYLHZ RI FLYLO LQIUDVWUXFWXUH V\VWHP DSSOLFDWLRQV
YROSS
1JX\ӉQ7KDQK%ҧQ7UҫQ$QK%uQK1JX\ӉQ'X\9NJ1JX\ӉQ1KұW1LQK4XҧQOêKӋWKӕQJ
FѫÿLӋQF{QJWUuQKVӱGөQJF{QJQJKӋWăQJFѭӡQJWKӵF$5YjP{KuQKWK{QJWLQF{QJWUuQK%,0
̩ ͹ 7UѭӡQJĈ+;'WұSVӕ
 7D\ORU 6 5RVWHQ ( 'UXPPRQG 7 5REXVW )HDWXUH 0DWFKLQJ LQ XV LQ

.DPHURQ:.LQFDGH$Q$XJPHQWHG5HDOLW\$SSOLFDWLRQIRU'LVSOD\LQJ'0DSVRQD0R
ELOH'HYLFH&RORUDGR6FKRRORI0LQHV
3HGUR1LFRODX0DFKDGR&DUYDOKR0RELOH$5IRUODUJH'6FHQHVXVLQJ0DUNHUV8QLYHU
VLGDGHGR0LQKR

 Protocol for Standard Contract Forms for projects
involving Building Information Modelling (BIM) in
Vietnam

Nguyen Nam Trung1 and Dinh Tuan Hai2

1
Power Engineering & Consulting JS Company 3, Hochiminh city, Vietnam
2
Hanoi Architectural University, Hanoi, Vietnam

trungnn@pecc3.com.vn

Abstract. This study will review international practices for BIM protocol and
current contractual environment in Vietnam, interview participants in current pi-
lot BIM projects on contractual practices, analyze possible conflicts between Vi-
etnam Laws and the BIM Protocol for pilot projects (1057/QD-BXD), and rec-
ommend steps for future BIM protocols in Vietnam.

Keywords: BIM, Contract, Vietnam Construction Laws.

1 Introduction

In 2016, Vietnam government declared decision 2500/QD-TTg established steps for

application of BIM in Vietnam construction industry. Expected timeline for the plan is
to prepare necessary conditions and skill training sessions for application of BIM from
2017 to 2019, initiate pilot BIM projects during 2018 to 2020; and introduce regulations
and guidelines on universal utilization of BIM in construction and operating manage-
ment activities from 2021. One important mission is to establish the legal framework
over application of BIM by create standard contract templates that include the contents
of BIM application [1].

This study will review international practices for BIM protocol and current contractual
environment in Vietnam, interview participants in current pilot BIM projects on con-
tractual practices, analyze possible conflicts between Vietnam Laws and the BIM Pro-
tocol for pilot projects (1057/QD-BXD), and recommend steps for future BIM proto-
cols in Vietnam.

2 International Practices for BIM protocols

International practices for contracts in projects involving BIM can be classified in 3

categories:

© Springer Nature Singapore Pte Ltd. 2020 1253

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8_201
1254 N. N. Trung and D. T. Hai

- One-size-fits-DOODSSURDFK8.¶V&,&3URWRFRODQG6LQJDSRUH¶V%&$%,03DU
ticular Conditions
- Ready-made approach: 8.¶V1(&1(&
- Custom-tailored approach: FIDIC 2017
2.1 One-size-fits-all approach
One-size-fits-all approach provided standard Protocol/Particular Conditions to be in-
cluded in the main contract/agreement for any party who carrying out BIM in the pro-
ject. 2 prominent examples are 6LQJDSRUH¶V%&$%,03DUWLFXODU&RQGLWLRQVDQG8.¶V
CIC Protocol.
Contents of CIC BIM Protocol: 1. Definitions, 2. Coordination and Resolution of
Conflicts, 3. Obligations of the Employer, 4. Obligations of the Project Team Member,
5. Electronic Data Exchange, 6. Use of Information, 7. Liability in Respect of Proprie-
tary Material, 8. Remedies ± Security, 9. Termination, 10. Defined Terms
Key clauses in CIC BIM Protocol³1.4 This Protocol forms part of the Agreement.
In the event of any conflict or inconsistency between the Protocol and any other docu-
ments contained in and/or forming part of the Agreement, such conflict or inconsistency
VKDOOEHUHVROYHGLQDFFRUGDQFHZLWKWKH$JUHHPHQW«´ [2]
Contents of BCA BIM Particular Conditions: 1. Definitions, 2. General Principles,
3. BIM Management, 4. BIM Execution Plan, 5. Risk Allocation, 6. Intellectual Prop-
erty Rights, 7. Electronic Data Exchange, 8. Termination, Rescission or Expiry of Prin-
cipal Agreement.
Key clauses in BCA BIM Particular Conditions:
³7KH3ULQFLSDO$JUHHPHQWIRUDQ\SDUW\ZKRLVFDUU\LQJRXW BIM shall include
WKH%,03DUWLFXODU&RQGLWLRQV´
³$Q\SDUW\WRWKH3URMHFWZKRKDVWKH%,03DUWLFXODU&RQGLWLRQVLQLWV3ULQFLSDO
Agreement shall, by a term in the relevant agreement or contract, include the BIM Par-
ticular Conditions in the respective agreement or contract with all of its subconsultants,
VXSSOLHUVDQGVXEFRQWUDFWRUVZKRZLOOEHLQYROYHGLQFDUU\LQJRXW%,0IRUWKH3URMHFW´
³,QWKHHYHQWRIDQ\LQFRQVLVWHQF\EHWZHHQWKH%,03DUWLFXODU&RQGLWLRQVDQG
the relevant Principal AgreemHQWWKH%,03DUWLFXODU&RQGLWLRQVVKDOOSUHYDLO´ [3].

2.2 Ready-made approach

NEC4 suite contains 20 standard contracts with many ready-made forms and options
for the Contract users to integrate to make most suitable contract for their project.
Regarding BIM, NEC4 provided Option X10 ± Information Modelling, with following
contents: X10.1 Defined Terms; X10.2 Collaboration; X10.3 Early warning; X10.4 In-
formation Execution Plan; X10.5 Compensation Events; X10.6 Use of Information
Model; X10.7 Liability. [4]

2.3 Custom-tailored approach

In response to comments from academics and SUDFWLWLRQHUVRQ),',&¶V³VLOHQFH´ RQ
BIM [5], FIDIC published a new set of Rainbow Suite of FIDIC Conditions of Contract
(Red/Yellow/Silver Books) in 2017 with Advisory Notes provided at the end of each
Protocol for Standard Contract Forms for projects... 1255

Book. The Advisory Notes does not provide any specific clauses for users to incorpo-
rate into the contract, but instead it provided advices to users of FIDIC Contracts where
the project uses BIM and it listed specific clauses to be thoroughly reviewed when
drafting the Particular Conditions. FIDIC also advise that professionals with engineer-
ing, construction, insurance and legal expertise should be consulted during contract
drafting process, and legal counsel should review the contract to ensure it does not cre-
ate risk in some jurisdictions [6].
List of Sub-Clauses in FIDIC 2017 Contracts should be thoroughly reviewed for
projects involved BIM:
1. General Provisions: 1.1 Definitions; 1.3 Notices and Other Communications; 1.5 Pri-
ority of DocXPHQWV(PSOR\HU¶V8VHRI&RQWUDFWRU¶V'RFXPHQWV&RQWUDF
WRU¶V8VHRI(PSOR\HU¶V'RFXPHQWV/LPLWDWLRQRI/LDELOLW\
2.The Employer: (PSOR\HU¶V3HUVRQQHODQG2WKHU&RQWUDFWRUV6LWH'DWDDQG
Items of Reference
3. The Engineer: 3.2 EQJLQHHU¶V'XWLHVDQG$XWKRULW\
4. The Contractor: &RQWUDFWRU¶V*HQHUDO2EOLJDWLRQV&RQWUDFWRU¶V'RFXPHQWV
4.6 Co-operation; 4.7 Setting Out; 4.9 Quality Management and Compliance Verifica-
tion Systems; 4.20 Progress Reports
6. Staff and Labour: 6.&RQWUDFWRU¶V6XSHULQWHQGHQFH&RQWUDFWRU¶V5HFRUGV
8. Commencement, Delays and Suspension: 8.3 Programme
9. Tests on Completion: &RQWUDFWRU¶V2EOLJDWLRQV
13. Variations and Adjustments: 13.3 Variation Procedure
17. Care of the Works and Indemnities: 17.3 Intellectual and Industrial Property
Rights; 17.4 Indemnities by Contractor; 17.5 Indemnities by Employer; 17.6 Shared
Indemnities
19. Insurance:19.2.6 Other insurances required by Laws and by local practice

3 Current Contractual Environment for BIM projects in

Vietnam

3.1 Contractual environment

A set of standard contract forms has been set-up in Vietnam by Ministry of Construction
and Ministry of Planning and Investment:

Table 1. Standard contract forms in construction issued by government agencies in Vietnam

Procurement Contract form Contract form Descriptions
Method by Mistry of by Mistry of
Planning and Construction
Investment (MoC)
(MPI)
Tradi- Works 03/2015/TT- 09/2016/TT- Standard Form of Contract for
tional BKHDT BXD Construction Works
(Design- Service 01/2015 /TT- 08/2016/TT- Standard Form of Contract for
BKHDT BXD Consulting Services
1256 N. N. Trung and D. T. Hai

Bid- Supply 05/2015/TT- Standard Form of Contract for

Build) BKHDT Supply of Goods
Design- 11/2016/TT- 30/2016/TT- Standard Forms of Contract
Build/ BKHDT BXD Agreement for Engineering, Pro-
EPC curement and Construction
(EPC)
While the forms in the traditional procurement are typical contract of a civil law
country: simple, with number of provisions implied by law into contractual relation-
ship, the standard contract forms for EPC projects (11/2016/TT-BKHDT and
30/2016/TT-BXD) are based on FIDIC Silver Book (1999 edition) and much more
complicated with its root from English common law [7] ~ [13]).
3.2 Legal environment for BIM projects
The first legal document mentions BIM is the Construction Law 50/QH13/2014
since 2014, in which BIM is mentioned in Article 4 & 66. Since then, many other reg-
ulations has gradually provide frameworks for BIM applications.
One key milestone for legal framework for BIM project is December 2017, when the
Ministry of Construction of Vietnam (MoC) issued Interim Guidelines for pilot BIM
projects by Decision 1057/QD-BXD. The BIM Protocol was provided in Part II of
Interim Guidelines, with 6 Articles to be included as Appendix of Contract between
parties in pilot BIM Projects [14].

4 Survey & Interview

In 2018 the Ministry of Construction of Vietnam issued list of 20 pilot BIM projects
by Decision 362/QD-BXD. The 1st author conducted interview by sending questions
(in Vietnamese) to selected persons who works in the pilot BIM projects. There are 7
interview answers among 34 recipients, count for 20%.
Key results of the survey and interviews:
- Occupation of interviewees: 2/7 interviewees from the Employer, 3/7 from Pro-
ject Management Board, 1/7 from Design Consultant and 1/7 from BIM Con-
sultant
- Source of fund/ or investment type: 3/6 interviewees working in Public invest-
ment projects, 2/6 in state-owned enterprise investment projects, 1/6 in private
investment project.
- Contract form: 2/6 use Contract forms issued by MPI, 2/6 use Contract forms
issued by MoC, 1/6 use FIDIC Contract, 1/6 use bespoke Contract
- Incorporation/reference of BIM: There is specific Protocol/ Particular Condi-
tions on BIM in 6/7 projects, Requirement for BIM models was mentioned in
WKH³RXWSXW´³KDQGRYHULQIRUPDWLRQ´³DV-EXLOWGRFXPHQW´RIWKH&RQWUDFWLQ
project
- The role of BIM Manager: BIM Consultant fulfilled this role in 3/6 projects,
and Design Consultant fulfilled this role in 3/6 projects.
- Interoperability in bidding /contracts documents: There were specific choice of
software platform in 4/7 projects, there were requirements on export/import in
Protocol for Standard Contract Forms for projects... 1257

IFC platform in 1/7 project, and interoperability was not mentioned in 2/7 pro-
jects
- Documents described the roles and responsibilities of each party in relation to
BIM: 3/7 described in Contract Agreement, 2/7 described in Contract Protocol/
Appendix/Addendum, 1/7 described in BIM Execution Plan, and 1/7 does not
mentioned.
- Providing comments/feedback on BIM models: the appraisal/supervision con-
sultants review 3D Model on CDE and provide comments/feedback by emails
in 3/6 projects, while in 3/6 projects the appraisal/supervision consultants re-
view on drawings (pdf/hardcopies) and comment by report/ letter.
- Application of BIM Protocol (issued by in part II of Guideline attached to De-
cision 1057/QD-BXD by Ministry of Construction): 2/6 projects has applied,
while 4/6 projects have not applied.
- Recommendation for BIM Protocol: 1/5 interviewee think as per current form
(a common BIM protocol which will be annexed to other contracts), while 4/5
interviewees think that BIM Protocol should be revised to separated BIM pro-
tocols for construction contract, design consulting contract, BIM consulting
contract, EPC contract

5 Conclusions and Recommendations

5.1 Conclusions

Conclusions from reviewing of international practice and current contractual envi-

ronment, as well as interviews of participants in pilot BIM Projects:
- Local contract forms (MoC/MPI) are mainly applied in pilot BIM projects in
Vietnam.
- Contract protocol in BIM Guidelines 1057/QD-BXD are not fully aware and
applied in pilot BIM projects
- Conflicts or disputes may be occurred to project when a standard Contract Pro-
tocol/Annex on BIM can prevail and govern over all other contract/agreements
in that project. Conflicts of laws are more likely to occur when that BIM Proto-
col is from common law system while all other main contracts/agreements are
based on civil law system.
- Majority of survey participants preferred to change BIM Protocol from one-
size-fits-all approach to ready-made approach.

5.2 Recommendations

Based on above analysis and review, the author would like to propose that a set of
standard Particular Conditions/Protocols shall be prepared for each type of contract
form mentioned in Table 1.
1258 N. N. Trung and D. T. Hai

References
1. Prime Minister of Vietnam: Decision 2500/QD-TTg approving the Plan to utilize Building
Information Modelling (BIM) for construction activities, management and operation of
buildings/facilities, Hanoi (2016)
2. Building and Construction Authority: BIM Particular Conditions. Version 2. BCA, Singa-
pore (2015).
3. Construction Industry Council: Building Information Modelling (BIM) Protocol, Second
Edition, London (2018).
4. NEC: NEC4 -box one- the contracts, 2nd Edition, NEC, London (2017)
5. Centre of Construction Law and Dispute Resolution - .LQJ¶V &ROOHJH /RQGRQ: Enabling
BIM through Procurement and Contracts, London (2016)
6. International Federation of Consulting Engineers: Conditions of Contract for EPC/Turnkey
Project, Geneva (2017).
7. Ministry of Construction of Vietnam: Circular 08/2016/TT-BXD Guidance on a number of
terms of Construction Consulting Contracts, Hanoi (2016)
8. Ministry of Construction of Vietnam: Circular 09/2016/TT-BXD Guidance on Construction
Contract, Hanoi (2016)
9. Ministry of Planning and Investment of Vietnam: Circular 01/2015/TT-BKHDT on Making
of Request for Expression of Interest, Invitation to Bid, Request for Proposals for Consulting
Services, Hanoi (2015).
10. Ministry of Planning and Investment of Vietnam: Circular 03/2015/TT-BKHDT Specifying
Preparation of Invitation to Bid in Construction Works, Hanoi (2015).
11. Ministry of Planning and Investment of Vietnam: Circular 05/2015/TT-BKHDT providing
specific provisions on preparing the invitation for bid on the procurement of goods, Hanoi
(2015).
12. Ministry of Construction of Vietnam: Circular 30/2016/TT-BXD on Guidelines for Engi-
neering, Procurement and Construction Contract, Hanoi (2016)
13. Ministry of Planning and Investment of Vietnam: Circular 11/2016/TT-BKHDT on Guid-
ance on Bidding Documents for Engineering-Procurement of Goods- Construction (EPC),
Hanoi (2016).
14. Ministry of Construction of Vietnam: Decision 1057/QD-BXD Interim Guidelines for Pro-
jects involved Building Information Modelling (BIM) during pilot stage, Hanoi (2017).
Author Index

A Bui, Quoc-Bao, 143

Abdelouas, Abdessalam, 639 Bui, Tan-Trung, 143
Aboseif, Eyad, 1019 Bui, Xuan Cay, 513
Adeli, Hojjat, 3
Alberio, Alessandro, 673 C
Aljer, Ammar, 1123 Cajka, Radim, 653
Amandine, Rojo, 793 Camey, M., 1111
Anh, Nguyen Thuong, 81 Camille, Christophe, 471
Anh, Tuan Phan, 355, 361 Canh, Khuu Minh, 1105
Ashraf, Mahmud, 179, 185 Canh, Nguyen Duc, 1105
Asnani, Karan, 519 Castelletti, Matteo, 673
Azmee, Norzaireen Mohd, 331 Charef, Nouar, 823, 829
Chau, Truong Linh, 709
B Cheng, Jack C.P., 1149
Badhrudeen, Mohamed, 1185 Chen, Po-Han, 1211
Bae, W., 679, 703 Chinh, Luong Minh, 215
Bai, Xian-Xu, 397 Chinh, Vu Dan, 123
Baker, J., 691 Christophe, Lanos, 793
Bart, M., 495 Christophe, Poinclou, 793
Basu, Biswajit, 877 Chu, Hong Nhung, 489
Beakou, Alexis, 111 Chu, Long Hai, 403
Berrah, Yacine, 823, 829 Chu, Tuan Long, 1205
Bhatt, Yashraj, 519 Chu, Youngkyu, 555
Bhuiyan, Muhammed A, 1025, 1075 Clarke, Todd, 471
Boltze, Manfred, 35, 941 Collet, F., 495
Boumezbeur, Abderrahmane, 823, 829 Combe, Gaël, 715
Bourbon, Xavier, 639 Cui, L., 781
Brahmi, Serhane, 823 Cuong, Dinh Quang, 123
Buchanan, Matthew, 1105 Czerepak, Adam, 907
Bui, Gia Linh, 1167
Bui, H.H., 311, 889, 895 D
Bui, Nam, 1229 da Cunha Oliveira Santos, Mariana, 1105
Bui, Phuong Trinh, 507 Dang, Cong Chi, 647

© Springer Nature Singapore Pte Ltd. 2020 1259

C. Ha-Minh et al. (eds.), CIGOS 2019, Innovation for Sustainable Infrastructure,
Lecture Notes in Civil Engineering 54,
https://doi.org/10.1007/978-981-15-0802-8
1260
Dang, Cong-Thuat, 299
Dang, Hong-Lam, 197
Dangla, Patrick, 871
Dang, Liet Chi, 647
Dang, NgocSon, 1217, 1241
Dang, Thuc Trang, 1117
Dang, Trang Thuc, 977
Daniel, C,, 397
Dao, Thuy-Ninh, 1211
Dao, Van Dong, 513
Darde, Benjamin, 871
David Jeong, H., 1131
Derrible, Sybil, 1185
Desrues, Jacques, 715
Dimitrakopoulos, Elias G., 245
Dinh, Hai Dung, 983
Dinh, Van Nguyen, 221, 877
Djamai, Zakaria Ilyes, 453
Djelal, C., 603
Djellali, Adel, 817
Doan, Nhat-Phi, 841
Doan, Noi Thi, 227, 1093
Do, Duc-Phi, 197
Doh, J.H., 203
Dong, Nguyen Duy, 525
Dong, Phan Huy, 805
Do Nguyen, Van, 1205
Dong, Tran, 131
Do Quang, Khanh, 769, 853
Do, Tan, 697, 811
Do, Van-Dong, 111
Do, Van-Toi, 349
Duc, Dang Viet, 209
Duc-Nghiem, Nguyen, 933
Dung, Nguyen Anh, 373, 609
Dung, Nguyen Thi Tuyet, 989
Dung, Tran Anh, 93
Duong, Minh-Thuan, 343
Duy, Nguyen Phuong, 787
E Huang, Kuangyu, 1173
Einav, I., 691
Er, Michael, 1223
F Huong, Nguyen Thi Thanh, 921, 959, 971
Fall, M., 781
Feng, Wenxian, 1173
Ferrier, Emmanuel, 239, 531
Fountain, Micah, 167
Franck, Cassagnabère, 441
Fukada, Saiji, 161
Fukubayashi, Yoshinori, 585
Author Index
G
Georgin, Jean-François, 543
Giap, Van Loi, 385
Graupner, Bastian, 775
Gross, Dennis, 131
Gugliemi, Yves, 775
H
Hai, Dinh Tuan, 221, 1253
Ha, Le Hai, 93
Ha, Manh-Hung, 1137
Hanafi, Eliya, 597
Hancock, Gregory J., 275, 281
Hang, Do Thi, 117
Hanh, Pham Huu, 367
Hanifa, Rahma, 597
Han, Seung-Ryong, 901
Ha, Sangwook, 555
Ha, Tuan Minh, 161
Hemalatha, G., 251, 397
Hewage, Dayani Kahagala, 471
Hien, Ta Duy, 117
Hieu, Nguyen Duc, 81
Hieu, Nguyen Duy, 1105
Hieu, Tran Trung, 1099
Hilloulin, Benoît, 639
Hoang, Manh-Tuan, 799
Hoang, Nguyen Ba, 391
Hoang, Thi Thanh Nhan, 465
Hoang, Tran Tuan, 1105
Hoang, Tung, 433
Hoang, Vu, 1167
Ho-Huu, Chinh, 591
Ho, Khoa N., 1161
Ho, Lanh Si, 859
Ho, N.M., 203
Hosseini, Maryam, 167
Ho, Truong Son, 1013
Hou, Yao-Long, 841
Hoxha, Dashnor, 197
Hung, Khuat Viet, 1037
Hung, Nguyen Phi, 697, 811
Hung, Vo Duy, 191, 965
Huyen, Le Thu, 1063
Huy, Nguyen Viet, 391
Huynh, Le Anh Thi, 287, 293
Huynh, Ngoc Thi, 137
Huynh, Nhan Vo, 769
Huynh, Quoc Hoang Anh, 1007
Hwang, Chao-Lung, 447
Author Index
I Lepretre, Alain, 1123
Ito, Daisuke, 865
J Le Thu Huyen, -Ing, 1081
Jeong, Jaehong, 555
Joel Shelton, J., 251
K Leu, Tho Bach, 483, 489
Kada, H., 603
Kafle, Bidur, 179, 185
Kajita, Saeko, 501
Kawai, Kenji, 507
Kawai, Marina, 763
Kawakane, Hajime, 561
Kawamura, Kei, 1155
Khallaf, Rana, 1019
Khanh, Tran Viet, 391
Khoa, Nguyen Tan, 409
Kien, Pham Hoang, 865
Kieu, Phuc, 769, 853
Kim, Cuong Nguyen, 1155
Kim, Dookie, 901
Kim, Ki-Du, 221
Kim, Seong-Heon, 847
Kim, Seung-Eock, 105
Kimura, Makoto, 585
Kim, Yong, 555
Kirkland, Brendan, 471
Kitagaki, Ryoma, 477
Kremer, Paul, 179, 185
Kubori, Yasutaka, 561
Kubota, Hisashi, 933
Kumar, Amit, 627
Kumar, Virendra, 627, 633
L Michel, Mouret, 441
Lacasse, Suzanne, 45
Lam, My Ngoc-Tra, 69
Lam, Nguyen Ngoc, 537, 543, 549
Lam, Nguyen Tung, 1235
Lam, Trinh Xuan, 1099
Lanos, C., 495
Laouar, Mohamed Salah, 817
Larbi, Amir Si, 453
Laurent, Molez, 793
Le, Anh Thang, 379, 579
Le, Chau, 1131
Lee, Seungtae, 555
Le, Hung, 1057, 1069
Le, Linh A., 311
Le, Lu Minh, 1191
Le, Minh Tu, 155
Le, Ngoc Quyet, 1223
Le-Nguyen, Khuong, 305, 1247
1261
Le, Thien-Phu, 111, 299
Le Thi, Huong, 1031
Le Thuy, Phan Thi, 915
Le, Tien-Thinh, 1167, 1191
Le, Trung-Tri, 847
Le, Van Nam, 137
Le, Van Phuc, 621
Li, Lijuan, 1173
Li, Xin, 179, 185
Long, Hoang, 733
Long, Nguyen Ngoc, 117
Long, Tran Duc, 441
Loukili, Ahmed, 639
Luan, Pham Dang Manh Hong, 1105
Luong, H.T., 691
Luong, Minh-Chinh, 1143
Luong, Nguyen Hoang Phuong, 745
Ly, Hai-Bang, 1049, 1167, 1179, 1191
M
Machelski, Czesław, 907
Mai, Huong Tran Thi, 853
Mai, Huy Quang, 227, 233
Mai, Lam Tuan, 1205
Mai, Luu, 349
Mashiri, Fidelis, 471
Matos, Rute Sousa, 915, 977
Matsuda, Hiroshi, 427
Matsuyama, Toshikazu, 561
Matthews, Adam, 57
May, Nguyen Huu, 561
Menges, Karin, 941
Minh, Phan Quang, 367
Minh, Tan Nguyen, 361
Minh, Tran, 373, 609
Mirza, Olivia, 167, 471
Mizobe, Asuka, 573
Mohavedi, Ali, 1185
Moissette, S., 495
Mudgal, Ankur, 883
Muftah, Fadhluhartini, 257
Mulchandani, Rahul, 519
N
Nakarai, Kenichiro, 501, 561, 573
Nam, Tran Tuan, 149
Naranjo, Nalin, 1185
Ngoc, An Minh, 1037, 1043
Ngo, Anh Quan, 859
Ngo, Chau Phuong, 621
1262
Ngo, Quoc Trinh, 385
Ngo, Thi Thanh Huong, 659, 859
Ngo, Tu Anh, 977
Nguyen, Anh Rin, 137
Nguyen, Chau V., 421
Nguyen, Cong Phong, 1087
Nguyen, Cuong H., 1161
Nguyen, Dinh Tai, 1205
Nguyen, Duc Hanh, 715
Nguyen, Duy Liem, 337, 343, 349, 579
Nguyen, Duy-My, 299
Nguyen, Duy-Thao, 87, 99
Nguyen, Giang D., 311, 889, 895
Nguyen, Hoang Long, 385
Nguyen, Hung Quang, 1167
Nguyen, Huu Hue, 1205
Nguyen, Huu Truong, 679, 703
Nguyen, Huyen Thi, 915, 977
Nguyen, Huynh-Tan-Tai, 337
Nguyen, Huy Vuong, 659
Nguyen, Long Khanh, 385
Nguyen, Mai Lan, 155, 513
Nguyen, Manh Duc, 1167
Nguyen, Manh Tuan, 1247
Nguyen, May Huu, 501
Nguyen-Minh, Long, 591
Nguyen, Nhu H.T., 889
Nguyen, Phi Long, 239
Nguyen, Phong Dang, 233
Nguyen, Phong Thanh, 69
Nguyen, Phu-Cuong, 69, 75, 653, 901
Nguyen, Quang Binh, 1001, 1087
Nguyen, Quang-Huy, 155, 173
Nguyen, Quang Tuan, 465
Nguyen, Quoc Dung, 659, 859
Nguyen, Thai Huyen, 1117
Nguyen, Thanh Ban, 1247
Nguyen, Thanh Phong, 621
Nguyen, Thanh Quang, 709
Nguyen, Thanh Sang, 403
Nguyen, Thanh-Hung, 379
Nguyen, The Hung, 709
Nguyen, The-Quan, 1211
Nguyen, Thi Huong Giang, 513
Nguyen, Tien Dung, 513
Nguyen, Tran-Hieu, 1199
Nguyen, Tran-Trung, 653
Nguyen, Trong-Nghia, 69
Nguyen, Trung Kien, 715, 721
Nguyen, T.S., 775, 781
Nguyen, Tuan-Anh, 173
Nguyen, Van Bac, 621
Author Index
Nguyen, Van-Ben, 337
Nguyen, Van Hung, 621, 679, 703
Nguyen, Viet Phuong, 433
Nguyen, Xuan Nam, 507
Nguyen, Xuan-Toan, 99
Nhan, Tran Thanh, 427, 615
Nicolini, Emilio, 665, 673
Nong, Zhenzhen, 847
O
Ogawa, Yuko, 507
Okui, Yoshiaki, 209
Orbovic, N., 781
Ounaies, Sana, 1123
P
Page, J., 603
Papazafeiropoulos, George, 1143
Paraskeva, Themelina S., 245
Parekh, Viraj, 519, 567
Park, Sung-Sik, 841, 847
Pascal, Monnot, 793
Pereira, Jean-Michel, 871
Pernin, Celine, 1123
Peters, T., 203
Pham, Binh Thai, 1167, 1179, 1191
Pham, Cao Hung, 263, 269, 275, 281, 287, 293
Pham, Duc-Duy, 69, 75
Pham, Duy Khanh, 275, 281
Pham, H.Q., 1111
Pham, K.D., 1111
Pham, K.V., 1111
Pham, Lien Thuc, 137
Pham, My, 299
Pham, Ngoc Bay, 621
Pham, Ngoc Hieu, 263, 269
Pham, Song Hong, 281
Pham, Thi Hai Yen, 1123
Pham, Tuan Anh, 385, 1167
Pham, Tuan Hung, 483, 489
Pham, Van Hoan, 1247
Pham, Van Minh, 659
Pham, Van-Trung, 1143
Phan, Cong-Vu-Duc, 379
Philippe, Blanc, 793
Phong, Le Trung, 373, 609
Phung, Huynh Luu Trung, 1105
Phuong, Nguyen Viet, 367
Potin, Gildas, 639
Pradhan, Nischal P.N., 245
Pradipta, Giovanni, 597
Prajapati, Nishit, 567
Author Index 1263

Prasad, B.K., 627 Tang, Anh Minh, 871

Prasad, Brajkishor, 633 Tang, My Ngoc, 507
Pretot, S., 495 Ta, Ngoc Binh, 1223
Pribad, Krishna Suryanto, 597 Tang, Yunchao, 1173
Prud’homme, Elodie, 543 Tani, Shintaro, 561
Tarighat, Amir, 1155
Q Tensing, D., 397
Quang Trinh, K.H., 615 Terribile, Ivan, 665
Quan, Le Vu, 733 Thach, Tran Xuan, 427
Quan, Nguyen Hong, 1105 Thai, Hong Nam, 433
Quoc, Doan Bao, 81, 93 Thai, Huu-Tai, 323
Quyen, Vu Thi Bich, 317 Thai, Son, 323
Quy, Nguyen Minh, 733 Thang, Do, 835
Thanh, Bui Tien, 117
R Thanh, Ho Trung, 427
Rasmussen, Kim J.R., 263, 269, 287, 293 Thanh, Huong Pham Thi, 355
Reuge, N., 495 Thanh, Nguyen Ngoc, 787
Rilett, L.R., 23, 1111 Thanh, Truong Thi My, 1043, 1049, 1179
Robira, Maxime, 639 Thao, Nguyen Danh, 763
Rohmer, Jeremy, 721 Thao, Nguyen Duy, 191
Rojas Lopez, M.C., 953 Thi, Khanh-Dung Tran, 447
Roux, Jean-Noël, 871 Thuc, Pham Tri, 805
Rutqvist, Jonny, 775 Thu-Ha, Nguyen, 757
Toan, Trinh Dinh, 927, 947, 953
S Tomala, Piotr, 907
Saghafi, Behrooz, 817 Tong, Anh Tuan, 745
Saifullah, Halwan Alfisa, 573 To, Thuy Nga, 1057, 1069
Salvatore, Ferdinando, 453 Tran, Anh Binh, 1247
Sang, Nguyen Quynh, 989 Tran, Hieu T., 889
Sang, Nguyen Thanh, 409 Tran, Linh H., 1161
Sani, Mohd Syahrul Hisyam Mohd, 257 Tran, Manh Tien, 531
Sankaran, Shankar Binh, 1223 Tran, Minh-Phung, 349
Sarala, L., 397 Tran, Ngoc Linh, 305
Sarkar, Raju, 883 Tran, Thanh, 299
Satomi, Tomoaki, 685 Tran, Thanh-Tuan, 653, 901
Shafiq, Nasir, 331 Tran, Thi Hien Hoa, 489
Shahrour, Isam, 1123 Tran, Thu-Hang, 799
Shim, ChangSu, 1217, 1241 Tran, Trung Hieu, 513
Shiozaki, Masando, 1155 Tran, Van Hien, 727
Shirzadi, Ataollah, 1191 Tran, Van Quan, 385
Shrivastava, Amit Kumar, 883 Tran, Van Toan, 727, 1205
Singh, Prince, 633 Tran, Vu Tu, 579
Smith, Mitchel, 1105 Trinh, Son Hoang, 415, 459
Solanki, Kushal, 567 Trong, Quang Hoang, 769, 853
Somja, Hugues, 173 Trung, Le Quoc, 733
Subhani, Mahbube, 179, 185 Trung, Nguyen Nam, 1253
Sundar Manoharan, S., 397 Truong-Linh, Chau, 757
Truong, Viet-Hung, 105, 1137, 1143
T Truyen, Tran The, 81, 93
Ta, Binh, 57 Tsujimoto, Kazuhito, 763
Tai Nguyen, H.T., 579, 615 Tuan, Dat Truong Trong, 853
Takahashi, Hiroshi, 685 Tuan, Vu Anh, 921, 959
Talandier, Jean, 871 Tung, Bui Thanh, 81
Tam, Thanh Nguyen Thi, 769 Tung, Nguyen Xuan, 477
1264
Tu, Nguyen Thanh, 995, 1063, 1081
Turmo, Jose, 13
U 1075, 1087
Ueno, Toshiyuki, 161
Ung, Thi Thuy Ha, 483, 489
V Vu, Anh-Tuan, 1199
Van Dang, Tran, 131
Van Dao, Dong, 415, 459, 1049
Van Dong, Dao, 927, 947
Van Dong, Nguyen, 367
Van Dung, Nguyen, 697, 811
Vanhove, Y., 603
Van Hung, Nguyen, 733, 739, 751
Van Hung, Pham, 93, 391
Van, Le Anh, 561
Van Le, Hiep, 1167
Van Le, Thang, 763
Van, Nguyen Quoc, 525
Vanny, La, 573
Van Phan, Tho, 915, 977
Van Quan, Ho, 421
Van Thuan, Nguyen, 117
Van Thu, Tran Thi, 1105
Van, Tran Thi Thuy, 317
Van Tuan, Nguyen, 367
Van Tuoi, Nguyen, 421
Venkatesh, V., 251
Vien, Nguyen Dai, 427
Viet, Nguyen Vu, 525
Viet-Phuong, Nguyen, 933
Vo, Duc Phuoc, 1025
Vo, Duy-Hai, 447
Author Index
Vo, Duy-Hung, 87
Vo-Le, Dien Ngoc, 591
Vo, Ngoc Duong, 1001, 1007, 1013, 1025,
Vo, Nguyen Duc Phuoc, 1075
Vo, Trung Dung, 1007
Vu, Anh Quang, 763
Vu, Ba Thao, 659, 721, 859
Vu, Do Anh, 965
Vu, Huy Cong, 1057, 1069
Vu, Minh Chien, 685
Vu, Minh Ngoc, 871
Vuong, Vinh-Dat, 763
Vu, Quang-Viet, 1137, 1143
Vu, The Minh Ngoc, 1205
Vu, Thi-Bich-Nga, 337
Vu, Xuan Hong, 239, 531
W
Wong, Peter Kok-Yiu, 1149
Wong, Y.D., 953
Woo, Seung-Wook, 841
X
Xuan, Nguyen Thi, 1235
Y
Yamakita, Yoshiro, 477
Yehualaw, Mitiku Damtie, 447
Z
Zou, Xiangjun, 1173