Volume 37 Number 2 AUGUST 2006 ISSN 0046-5828
GEOTECHNICAL
ENGINEERING
Journal of the
SOUTHEAST ASIAN GEOT
SHNICAL SOCIETY
Sponsored by
ASIAN INSTITUTE OF TECHNOLOGY Vol. 37, No. 2, AUGUST 2006
CONTENTS
‘Main Papers:
DEM Analysis of SoiLPipeline Interaction in Sand Under Lateral and Upward Movements at Deep Embeent
by. Vimsin and K. Sopa
laflunce of Salt Additive in Lime Stabilization on Organic Clay
byS.Koslanan,K. Onitsuka and T. Negom
‘eterinaton of Partitioning and Dispersion CoefMcint Using Column Tests
by MA. Tanchuling, J Takemura, M.A. Khan and O.Kusakabe
Seale Effect on Fracture Mechanisms of Non-Persistent Rock Joints Under Uniaxial Loading
yO. Mughiede and I Khawaldeh
Consolidation Behavior of Reconstuted Bangkok Clay
by B. Ulrichon, TH. Seah, V.Budsayaplatorn and P. Luthenaprast
‘Technical Note:
Modelling Liquefaction of Sand ina Geoteenica Centrifuge
by PP. Bartake, 8 Roghanandon and DN. Singh
Abstracted and/or Indexed in Geotechnical Abstracts
3%
95
na
bt GEOTECHNICAL ENGINEERI
Published by the
SOUTHEAST ASIAN GEOTECHNICAL SOCIETY
EDITOR
NOPPADOL PHIENWEJ
ASSOCIATE EDITORS
W.H. TING, Malaysia ME. CHANG, Singapore
Y.S. FANG, Taiwan
EDITORIAL ADVISERS
AS. BALASUBRAMANIAM, Australia T.W. LAMBE, U.S.A.
EW. BRAND, U.K. S.L. LEE, Singapore
RP. BRENNER, Switzerland ZC. MOH, Taiwan
V.F.B. DE MELLO, Brazit H.G. POULOS, Australia
W.H. TING, Malaysia C.D. OU, Taiwan
SEAGS GENERAL COMMITTEE
KY. YONG (President) cD.0U
A'S. BALASUBRAMANIAM S.F.CHAN
Z.C. MOH ME. CHANG
SS GUE W. MAIRIANG
S.M.WOO T.A. OO1
WH, TING S.K. TANG
S, SAMBHANDARAKSA K. PETCHGATE
S.L. LEE D.T. BERGADO (Secretary-General)
N. PHIENWES (Editor) JC.LI (Immediate Past President)
Geotechnical Engineering is the oficial journal of the Southeast Asian Geotechnical Society. It is published
three times a year in April, August and December and is sent free to members of the Society. The annual
subscription rate for non-memiers is USS30 to individuals, and USSS0 to libraries and companies. Back issues
are available, Cheques or money orders should be made payable o the Asian Institute of Technology.
‘Membership application forms and other details can be obtained fom:
Tie Secretariat, SEAGS
Row 211, AIP Library
Asian Institue of Technology
P.O. Bos 4, Klong Luang
Pathumthani 12120, Thailand
Cover Photograph: Construction of foundation ofa five-star hotel located on a beach front hill-slope made of
residual soils of weathered granitic rock using a combination of shallow and piled foundation in Phuket,
Thailand, EDITORIAL PANEL
Dr. RP. Brenner
Weinfelden
Switzerland
Prof. Cheng-Hsing Chen
National Taiwan University
Taipei, Taiwan
Prof, In Mo Lee
Korea University
Seoul, Korea
Prof. San-Shyan Lin
‘Taiwan Ocean University
Keelung, Taiwan
Dr. Warakom Ms
Kasetsart University
Bangkok, Thailand
Prof. Harianto Rehardjo
Nanyang Technological University
Singapore
Dr. Satoru Shibuya
Kobe University
Kobe, Japan
Dr, Jiro Takemura
‘Tokyo Institute of Technology (TIT)
‘Tokyo, Japan
De. Tanaka, Hiroyuki
Port and Airport Research Insitute
Yokosuks, Japan
Prof. Wilson H. Tang.
‘The Hong Kong University of Science & Technology
Kowloon, Hong Kong
Dr. TH. Seah
MAA Geotechnies, Co, Lid
Bangkok, Thailand
Dr. Jin-Chun Chai
Saga University
Saga, Japan
Prof, B Indraratna
University of Wollongong
Wollongong, Australia
Prof. Chun-Fai Leung
[National Univesity of Singapore
Singapore
Prof. Madhira R. Madhav
Indian Institute of Technology
‘Kanpur, India
Prof. Hiroyasu Ohisu
Kyoto University
Kyoto, Japan
De. Mobamad R. Selamat
University Sains Malaysia
Pulau Penang, Malaysia
Prof. Mitsutaka Sugimoto
Nagaoka Univesity of Technology
Nagaoka, Japan
Prof. Siew Ann Tan
National University of Singapore
Singapore 117576
Prof. Jian Zhao
Nanyang Technological University
Singapore 639798
Dr. Charles W. W. Ng
‘The Hong Kong University of Science and Technology
Kowloon, Hong Keng
Dr. S.L Shen
‘Shanghai liao Tong University
‘Shanghai, China NOTES FOR GUIDANCE TO AUTHORS
Three copies of the orignal paper (and all figures and tables) together with a Microsoft Word file should be
submitted to the Editor, Geotechnical Engineering Journal, School of Civil Engineering, Asian Institute of
‘Technology, P.O. Box 4, Klong Luang, Pathumthani 12120, Thailand. Papers under review, accepted for
publication, or published elsewhere ae not accepted, ‘The review and publication procedures will be faster ifthe
basic requirements listed below are followed.
1, The manuscript including abstract and references must be typed double-spaced in Times New Roman 12
‘on one side of Ad paper witha margin of 25 mm on each sie.
2. The length of titles must not exceed 70 characters and spaces.
3. The maximum length of papers in the print format of the Joumal is 12 two-column pages in single-
spaced in Times New Roman 9 including figures and tables. A Journal page contains approximately
11040 words, Authors can approximate manuscript length by counting the number of words on atypical
‘manascrpt page and multiplying that by the numberof total pages (except for tables and figures), Add
‘word-equivalents for figures and tables by estimating the portion of the journal page each will occupy
‘when reduced to fit on a 160 mm x 240 mm joumal page. A figure reduced fo one-quarter of a page
‘would be 260 word-equivalents. When reduced, the igure must be legible and is type size no smaller
‘than 6 point font after reduction)
4. Figures: Line art should be submited in black ink or laser printed; halftones and color should be
original glossy art. Figures should be submitted at final width Le. 90 mm for one column and 185 mm
for two columns. ‘The font ofthe legends should be in Times New Roman and should use capital eters
forthe first letter of the first word only and use lower case for the rest of the words. Background
screening and grids are not acceptable.
5. Bach table must be typed on one side ofa single sheet of paper
6. All mathematics must be typewritten and special symbols identified. Letter symbols should be defined
‘when they first appear.
7. The paper must end witha set of conclusions.
8. Practical applications should be included, if appropriate
9. [experimental data andor relations fitted to measurements are presented, the uncertainty of the results
‘must be sited, The uncertainty must include both systematic (bias) errors and imprecision.
10. Authors need not be Society members, Each author's full name, Society membership grade (if
applicable), present title and affiliation, and complete mailing address must appear as e footnote atthe
‘bottom of the first page ofthe paper.
11, Each author must use SI (Intemational System) us
given in parentheses or in an appendix.
12, References should appear in the text as the author name(s) followed by the year of publication in
parentheses. A list of references must be given atthe end ofthe text in alphabetical order in capitals and
‘double spacing must be used between references. An example of the format to be used in reference list
are given below.
and units acceptable in SI. Other units may be
HOEK, E. and BROWN, E-T. (1982). Underground Excavations in Rack nd Editon). London: The Institution
of Mining and Metallurgy, 527p.
PUN, W-K. and AMBRASEYS, N.N, (1992). Earthquake data review and seismic hazard analysis for the Hong,
Kong region. Earthquake Enginoering and Structural Dynamics, Nol 21, No. 1, pp. 433-43,
RAMASAMY, N. (1992). Sof ground tunneling in Bangkok subsoils, M. Eng. Thesis, Asian Instute of
Technology, Bangkok, Thailand,
TATSUOKA, F. and KOHATA, Y. (1994). Stiffness of hard soils and soft rocks in engineering applications.
Proceedings of the I" International Conference on Prefailure Behaviour of Geomaterials, Balkema,
Rotterdam, The Netherlands, Vol. 2, pp. 947-1063. DEM ANALYSIS OF SOIL-PIPELINE INTERACTION IN SAND
UNDER LATERAL AND UPWARD MOVEMENTS AT DEEP
EMBEDMENT
pene interactions under lateral and ypsard pipe movements in sand ae investigated using DEM
fnalysia The simulstions are performed for both medium ad dense stad condition at different emsbedment ratios of up
60. The comparison of peak dimensionless forces ftom the DEM and caler FEM analyses shows tht, for medium sand,
both methods show simile peak dimensionless fores. For dese sand, the DEM analysis gives more gradual tanstion of
‘hallow to deep failure mechanisms than the FEM analysis andthe pak dimensionless forces avery deep depth are higher in
the DEM analysis thn inthe FEM analysis. Comparison of the deformation mechanism suggests that this due to the
Aifferences in soil movements around the pipe associated with its pariculete nature
The DEM analysis provides
‘supplementary data of the sl-ppelin interaction i sand at dep embedent condition,
‘Keywords: DEM Arayss, FEM Analysis, Sil-ppolin inerastion, Numerical simlations|
INTRODUCTION
‘An understanding of pipeine response to vertical and
Intra ground movements fs esential in pipeline design,
Rational design under these conditions requires knowledge
of soil forces reating from the relive soil-ppelie
Aisplacement. "The design formulations of the Toad:
tdsplacement characteristics for soil-piptine interactions
sre given by ASCE's "Guideline forthe Seismic Design of
Oil and Gas Pipeline System (1984)". This Guideline
recommends that the peak dimensionless fores and the
displacement at peak ores are functions of the embeeent
depth ofthe pipe and sol ition angle, and thatthe non
linear relationship between the force and displacement is
rmodeled by a rectangular hyperbola. These design values
ae derived ffom the database of pipe testing at relatively
Shallow embedment depths. However, thee are suations
‘where the pipe embedment ratios ean be deeper than the
‘ales given in the guideline. Typical scenarios are the
Construction of a large embankment over an ara
iting pipelines and installation of deep ipeines dc to
Siready congested underground space If the desig is
based on a linear extapolation fom the ASCE dat, the
Guideline would provide vory large vales fr te peak Toad
exerted on the pipe and the sitfass of the sai-ppeine
Ieraction This potential overestimation would in ura
results n uneconomical design
‘The deep embedded pipeline problem has been
invesigated. by ‘Yimsin et al (2008). They examined
various analytical solutions availble for the peak forces
‘onto a pipe or srp ico and showed ha there arc lege
differences in the computed peak dimensionless forces for
‘deep embedment conditions Tt was concluded that «Hest
fextupolition fiom the ASCE data derived fom shallow
depth condition should not be employed because of the
change in flue mechanism fom shallow (ce. ground
Surface dependen) to deep embedment conditions (ve
(ground surface independen). ‘They performed finte
lement analysis to examine the tansition in both median
nd dene sands. Unfortunately its not possible to validate
their findings because ful-sal pipe loading experimental
results or any ther spplementary dasa ot availabe
‘Assistant Profesor, Deparinet of Chil Eien.
Barapa Universi Senook, Mean, Chonbus 2011, Thad
(Conespondig aur Ema yiamisbouae)
"Spear, Deparment of Engoning, Unversity of
‘Cantige,Trampingion See, Cobre CBD TPZ, UK.
This stay hasbeen underaken ww provide
supplementary dats for deep embedded pipes using the
Distinet Flemeat Method (DEM). Fist, DEM analysis wat
conducted for shallow embedment “depth conditions #0
Calibrate the input parameters by comparing the results to
‘he experimental data epated by Trautmann and O'Rourke
(1983). The analysis was then extended to deeper
fembedment conditions. Duct its dinconinsows nate, i
{considered that DEM should tetersimulte the soll
‘movement closet the pipe at large pipe displacements and,
hence, may yield more realise fete to the problem
compared the contnaum-based Finite element aa
DISTINCT ELEMENT METHOD ANALYSIS
‘Model Assumptions
‘The pipe loading experiments were simulated using the
Aisint element method based onthe approach by Cundal
snd Stack (1979). The distinct clement code PFC®
(tasca, 1999) was employed. Arbitrary sized spherical
particles are weated as gid bodies and are allowed to
fverip one another atthe ‘contact points cesting
Intrpricle forces. The contacts between particles are
characterized through the sles and slip condition. No
bonding Deoeen panicles is employed ia this sudy 10
"Slate uncemented dry snd
‘The linear elastic contact model vas employed: the
contact free and eative contact diaplacement ate inealy
‘elated by a constant contact stiffness. Two siffess values
te required foreach conse: they are () normal contact
stiffs Ly and. (3) tangeatal contact siiess
(orceldsplacement), In this study, they are assumed to be
equal (sand ~ krsand). The parametic study hows that
‘he change inthe partle tangential coast fess has
minima eflects on the pipe losing resus the change in
ip sand ffom fy sand to SKN sand results in negligible
change in the force-displacement curve. Particle siding
focure when the tangential contact force reaches is
‘maximum allowable valus, wbich is computed fom the
coeficiem of ine-pariclefiton ange berween the (wo
contacting entities multiptied by the normal contact free,
‘The tank and pipe were modeled by senes of planar wal,
“The dimensions ofthe tank are the same asthe actu tank
used by Trautmann and O'Rourke (1983) forthe casts of
HID or Hib = 30 (see Fi, 1}, Whereas the with was dou-
JOURNAL OF THE SOUTHEAST ASIAN GEOTECHNICAL SOCIETY / AUGUST 2006 63, pens on 00
230 nm
scl pipe loading (UD < 30)
() Upward ppe leading (HED =30)
Fig. 1 Various replons of the DEM mode! (/D < 30
and HetD 30),
Bled forthe cases of HD oF HD 2 30 0 reduce any
‘ourdayefets Ge Fig 2). Heri, H's the depo the
tom ofthe pipe (ormaly wed in eter pipe losing
ses) i the depth te center Ft pie nomaly
{ed io upward pipe loading eases, and Ds the pipe
flamer "The cela erosseson of te pe wes
‘modeled bya Ise polygon The an wall wat assumed
{oe smooth: tht i nora cot shes ea
{0 that of the parle tat bar aero tngetl cont
Sess an eo sce fcton. ‘The pipe ha decal
enact slfness in bath nal ab tape ions
fd the vals ae equ 0 hat ofthe pele. The pipe
tsi ufc icon angle (hy) equal alo einer
parle fet angle of san (The pranetic say
Sons ta the change in he pipe sre ton se
between = atl fe = hs nein
the force Spacer
Sand was modeled as acolleton of spheres pails
with its size distribution folowing the normal Gaussian
isebution. Due to compton imitation, i was nok
possible to model using the actual size particles (Djp~ 0.2
fm end Diy = 0:82 rm). Therefore, the sand partes
‘were modeled with larger size partles. To incense the
computational sped, the size Of the patil was also
varied in various regions of the mode; smaller paricles
‘were used inthe region where arg soi defermation is
txpested For the eases of HD o& HL/D = 30, typical
‘umercal models wed for the simulations are shown in
Figs 1. tthe region near the pipe (Region A). the pales
Irae Faye 125 mim with standard deviation = 2. mm
(@5-timce larger than the acual sand. awed inthe
experiments). Further away (Region 8), the piles have
rye" 25° man with standard deviation = 8.0 mm (50
timer large), When 40D > 175 or H/D > 17, another
region (Region C) vas added and it bas panicle sizes of
Prwage "378 mm with standard deviation = 7.5 mm (75
‘dnct lage) Figure 2 shows typical so models for eases
with HD or #2/0 2 30. The width is twice a large a8 at
‘sed in he sallow embedment cass inorder wo redoee the
boundary effects At Region D, the patiles M6 rnnge =
25.0 mm with standard devon = 5.0 rm (50 tines lgger
than actual sand) Purter away (Region E, ares prices
ae pled: Faunge = 37.5 mim and stand deviation = 75
‘nm (75 times larger). This allowed the numberof particles
to e less than 18,000 even for the decpent ease, Results
fiom the parametric study on particle size effets shox that,
the case) without small paicles (12.5 mm averge
diameter) nea the pipe compates a slighly lager peak
force than the case with small paaicis, Mowever te
diference was les than 8. Hence ination to the pipe
ameter, the particle size used was eonsiered wo Be salt
enough in aed 1 eval pipe loads,
‘The DEM analysis consists of the flowing fou steps
1. Tank setup: Test tank was filled with predetermined
number of reduce-siz panicles, ‘The rads of the
spheres was expands 10 specified value oan 3
predetermined void aio.
2 Inka conditon: The generate bas were subjected 19
_graviationl force a rach the geosttie condition,
ws es |
(4) Late pipe loading (UD 2 30)
“t | ans
tom
(b) Upward pipe loading (402 30)
Fig. 2 Various regi
and D230)
f the DEM model (110 2 30
'84/ JOURNAL OF THE SOUTHEAST ASIAN GEOTECHNICAL SOCIETY / AUGUST 2006 During tis stage, there were some movements of balls
due to the deformations atthe panicle contacts which
resuled in Some changes in void ratio and embedment
Aepth This wa taken into account by til and eror of
‘he initial void ratio and embedment depth during the
tank setup stage. Typical void ratios obtained were 063
{0 069 for medium sand (0.67 was the arget value) and
1.56 to.0.62 for dense sand (0.58 waste target value)
5. Pine instalation: The pipe was placed into the model by
delcing the balls that were lated atthe desired pipe
Teeation, The model was then permited to. reach
aulibriam by allowing the pipe to slightly adjusts
Jocaton by maietaining the equilibrium conditions of the
Pipe. The examples. of the DEM models after pipe
{nsalation are shown in Fig. 3
4.Pipe pulling: The pipe was pulled in lateral or upwar
Aiecions by imposing displacements 10 the pipe The
ther direction was fee 19 move, The ration ofthe
Pipe was not allowed, which i consistent with the
‘experimental ee-up by Trautmann and O'Rourke (1983),
Material Input Parameters
‘The input parameters for DEM snalysis are listed in
Table 1. Most of the parameters were determined by
calibrating the numerical results with the experiments Sat
Of (i) triaxial test results ofthe sands used fr the tank
experiments (Turner and Kulhawy, 1987) and Gi thea
Dipe loading test results at shallow depts (Trautman and
O'Rourke, 1983),
Results fom the triaxial test simulations show tht the
DEM analysis (sing Asin = sand) can simulate the
Table 1 Input parameters for DEM analysis
Parameters Values
‘Normal conse sinew oF From Bes Vor
patch ans a
‘Tangent contac ifthess of
panicle, Rs
‘Nona contact sifines of pipe,
sss
“Tangent conact hess oF PS
rape
‘Nonna eoiact iff of wal,
Irena Int
a
ama
tne nang OTD
ate ond eon
Tawar mg,
bass
Density of panicle, p(kgim!) 2740
Varies in Regions
Radius of parle, ‘Yee inex
measured stress-strain relationship by wing a high value of
interparticle fiction angle (an gps = 30) a8 shown in
Fig, 4. This unusually high valve is mainly due to the we
of ‘stmoot spierealparicles, which allows excessive
Panicle rolling (eg. Thomas and Bry, 1999). At low
‘ales ofthe interptle Faction ange, was not posible
to achieve a good Fito the experimental data, The contact
stfltess was also varied in order 1 fit he dats, Figre $
shows the contetsffnessevahated a different confining
stresses andthe Tllowing empirical reitionship was
blaine.
04340 p, (pin kPa) o
() Upward pipe loading (medium sand, HdD= 25)
Fig. 3 Examples of DEM model
5
'
Fig. 4 Calibration of DEM res
test dat
JOURNAL OF THE SOUTHEAST ASIAN GEOTECHNICAL SOCIETY / AUGUST 2006 /85 Initially this relationship was used forthe pipe louding
simulations by assigning the initial effective overburden
stress at the pipe level as p,. Unforunately it wes found
thatthe caleulated pipe force was smaller and the foree-
displacement characteristic was les sf tan the measured
‘ones by Trautman and O'Rourke (1983). This is primary
because thre i significant increase inthe mcan pressure
in fom of the pipe asthe pipe displaces the surounding
sci The increase of mean pressure fom he inital py"
rch greater thay that observed in typical axial testing
‘This in tur resulted in large increase in contac stiffness of
the soil around the pipe. By considering thatthe mean
pressure and hence the contact siffbess vary at various
arts of the model during pipe loading, a strict appliation
Of Eq, (1) is posible by implementing i ino the stress
Aependent contact silfness model k= MY), where fy is the
lnterparicle normal comact force. Instead, however, 2
different approach was employed in this study. AS
described an the next section, the contctstifhess and
‘nterpaniole fhction values at different embodiment depts
were fist back-zaleulted by fing the DEM pipe loading
ure tothe experimental data at shallow depts reported
by Trautmann and O'Rourke (1983) and then the calibrated
values were used fo simulate the deep embedment ase.
i
Fig. § Relationship between contact stifness and
30, the DEM
gives somewhat lager esi dimensionless force thn the
FEM and a more gradual vanstion from shallow to deep
failure mechanisms is observed. Figur 12 (a) also nodes
{JOURNAL OF THE SOUTHEAST ASIAN GEOTECHNICAL SOCIETY / AUGUST 2006 /87 the relationship proposed by Oversen (1964), which is
recommended in the ASCE Guideline. Oversen’s model
ves relatively good match tthe computed transition of
the peak dimensionless force om shallow to deep
tembedment conditions but slightly overestimates the peak
dimensionless forces. At HID of 40, Oversen's model
overestimates the peak dimensionless forces frm the DEM
amalysis by approximately 20% for both medium ard dense
sd cases
Similar observation can be made for upward pipe
movement as show in Fig. 12(0). For medium sand, the
DEM analysis yields consistent resuls with FEM anal.
For dense Sand, the DEM analysis yields consistent resus
‘with FEM only when A/D is smaller than 40. For HD >
40, the peak forces computed from the DEM analysis ae
somewhat greater compared to those by FEM and the DEM.
Gata exhibit a mace gradual ransiin ftom shallow to deep
failure mechanism. Figure 12(0) also includes. the
‘lationship proposed by Meyetof and. Adams (1968),
‘Meyethof&e Adem model does not match the computes
teasiion ofthe peak dimensionless force from shallow to
‘eep embedent conditions and underestimates the citcal,
peak dimensionless forces. At H/D = 60, Meyethot &
‘Adams’ model underestimates the citcal pea
Gimensonless forces fom the DEM. resus by
sppronimately 30% and 40% for medium and dense sands,
spectively
(2) UD ~40, Medium sand
(6) WD ~=40, Dense sand
Fig.10 Force-dsplacement relationships at deep
‘embedment depth of lateral pipe loading
Stauyesnetm on
© “Soteieenn ee
(@) Herd =30, Medium sand
(0) He/D~30, Dense sand
Fig. 1 Rorce-isplacement relationships at deep
embedment depth of upward pipe loading
DISCUSSIONS,
Soll Deformation Pattern
‘The soll displacement pattems fom the experiments,
FEM, and DEM (ian §, = 30) analyses of pipeline
movement in mediim sand ata shallow embedment depth
are shown in Figs. 13 and 14 for lateral (H/D = 11.5) and
upward (HD = 13) pipe loading cases, respectively
Similay, Figs 15 and 16 show the soil displacement
pattems for the dense sand cases. The DEM simulates
Circular metion of soi behind the pipe, which was observed
inthe experiments. I pears that larger soil mss 5
influenced by the pipe losing in the DEM compared to the
FEM, especially for dense sand. This may be the reason fr
theft that the DEM computes larger peak foes at deper
depts.
“The computed soil displacements from the DEM
analysis of lateral pipe lading ata deopembedment depth
(WUD = 30) ae shown in Fig. 17. For medium sand, the
‘eformation patter shows deep failure mechanism,
whereby the upward sol deformation doesnot extend to the
‘ground surface. However, for dense sand, sallow flue
‘mechanism with upward sil deformation atthe ground
surface is sll observed, This is consistent with the peak
dimensionless force plot presented in Fig, 12(a), which
shows thats deep file mechanism is reached at ~ 30,
for medium sand but the mechanism at the same
‘embedment ati is til inthe wansition state in dense sand
{8 JOURNAL OF THE SOUTHEAST ASIAN GEOTECHNICAL SOCIETY / AUGUST 2006 oR
sm et
nn eee
(2 Lateral pipe loading
no
(8) Upward pipe loading
Fig. 12 Comparison of peak dimensionless force from
DEM, FEM, and analytical solutions at deep
cambedment depth
‘The computed soil displacements from the DEM
analysis of upward pipe loading at deep embeément depth
(4D = 30) are shown in Fig. 18, For medium sand, the
eformation patern shows deep fale mechanism;
however, for dense sand, shallow failure mechanism is
shown. This is also consistent withthe peak dimensiones
force pot presented in Fig. 1200), which inicates that a
deep failare mechanism reached a i/D = 30 fr medium
‘sand but the mechanam atthe same embedment rai is til
inthe transition sae for dense sand
The extent of sil mass involved in the pipe loading
increases ab the contact sifaess increases oF the inter
panicle friction angle decreases. It was found tat the soil
4 the model boundary moves significant amount for the
‘ave of an = 0.5, indicating thatthe model boundary i
infuencing the pipe loading force. This is consistent with
the large peak frces computed forthe eases of tan = OS
sas shown in Fig. 10(b) and 116),
Selection of Input Parameters
In this study, the values forthe contact stifness and
interparticle friction angle ‘were selected by fitting the
{canal compression data and pipe loading das. The fiting
exercise indicates that both values are equally important
land various combinations of the contac stiffness apd tan
‘an yield similar peak forces When & larger value of
contact stifess 16 used, a smaller interparticle tition
gle needs to be selected tot the experimental data,
Hence, the bickccaleulted values require proper
evaluation
As discussed before, the use of a high value of the
iner-paricle fiction angle is due to the use of smooth
sphere pariles, which allow excessive particle rolling
‘This is indicated from the wiaval compression text
simulation where tan, needs tobe greater than 3.0 6
‘he DEM data to the experimental results. For pipe losing
simulation, it was possible to use smaller values of inter-
pst fiction angle (ke. tan g = 0 and 1.0) if larger
values ofthe contact stifhess ae adopted. However, the
low inte-paricle fiction angle produced unrealistic so
deformation pater by pipe loading. Hence, te ease ofan
$,= 301s more realistic than the ease of tan g,=0'. This
1S confimed by the fact that the force-dsplacement
relationship obtained from the DEM analysis ofthe case of
tan g,~ 3.0 matched well tothe experimental data
The power coefficient forthe stress dependency on
contact stiffess wes 0.41 from the axial compression er
‘Srultion (se Eq. (1) This i within the normal values
for soils, which is approninwtely 0.5 (Hardin and Black,
1966). "It has been shown that the taxial stiffness
(Young's modulus) has relationship with confining sess
in the sume manner asthe contact stilfioess does (Chang and
Lio, 1998). The power coefficient derived from the pipe
loading simulations was. much greater Gan the normal
values for soils (9 Eqs. (2) and), Inthe simulations, 3
Constant valve of contact stiffness was assigned throughout
the model. As the conlact sifness varies with mean
pressure, a sit application of Eg. (I) i posible by
Implementing it nt the stress dependent comtctsuiffcss
‘model. This approach was not pursued inthis study due ©
‘computational complication. Henee, it should be noted that
Bs (2) and (3) are model iting parameters suitable for the
particular boundary value problems investigated in this
Ssudy and should not be considered asa ral sil bebavier.
JOURNAL OF THE SOUTHEAST ASIAN GEOTECHNICAL SOCIETY / AUGUST 2006 89 Se Rewes Bisel yhg wea
(2) Experimental tank test (Trautmann & O'Rourke, 1983)
aS AMEAPLEULL Et tie
(©) FEM results
90 JOURNAL OF THE SOUTHEAST ASIAN GEOTECHNICAL SOCIETY / AUGUST 2006 Tent No. 39 Bionatar (> = 102 me
Behesty = T2.6 wund —— Bagtn saa 5 TP
70 ae
saa
aa thee
10g
‘a8 ea aS 7
(4) Experimental tank test (Trautmann & O'Rourke, 1983)
(0) FEM resules (DEM results (Gan = 3.0)
Fig. 16 Comparison of di
(eID = 13, Dense s
ment pattern of upward pipe loading from experiment, FEM and DEM
)
JOURNAL OF THE SOUTHEAST ASIAN GEOTECHNICAL SOCIETY / AUGUST 2006/91 (6) Dense sand (tam $= 3.0)
Fig. 17 Displacement patter from DEM of laters! pi
192 /JOURNAL OF THE SOUTHEAST ASIAN GEOTECHNICAL SOCIETY / AUGUST 2006 (©) Dense sand tan = 3.0)
ig. 18 Displacement pattera from DEM of upward pipe loading (He/D =)
JOURNAL OF THE SOUTHEAST ASIAN GEOTECHNICAL SOCIETY / AUGUST 2006 /93 CONCLUSIONS,
‘Te sol pipeline interactions under lateral end upward
pipe movements in sand’ were investigated using DEM,
‘The simulations were performed for bath medium and
dense sand conditions a differen embedment ration of up
to 60. For medium sand, the comparison of the pee
imensonless forces from the DEM and FEM analysis
shows good mtich. For dense sand, the DEM dat show
‘more gradual tasition fom the shallow to deep fale
‘mechanism and have a tendency 10 give somewhat larger
peak forces at very dep depths The investigation of the
‘oil displacement paters computed by DEM show's larger
soll mass movement compared to FEM da, especially for
dense sand. The advantage of DEM over FEM is ability
to simulate large movement of sil around the pipe. Als,
‘he DEM analysis can continue with unlimited pipe
‘movement until eather ultimate peak force (or farther,
‘wheceas the continuum based FEM of stops a some pipe
displacements before the peak Tooe ean be reached ecae
large distortion of the mesh causes nurerial convergence
problem.
‘The DEM results from this study confirm the findings
by Yimsic etal (2004 thatthe calculation of the sol
pipeline interaction at deep embedment conditions should
fot basen a linear extrapolation from the ASCE data of
shallower embedment depts. However, the DEM resis
suggest” that”) there may be a possibilty of
“underestimation of the peak dimensionless force at very
ep embedineat depths ifthe design chart proposed by
‘Yimsiri et al. (2008) is employed and (i) the transition
from shallow to deep failure mechanism may proceed rore
gradually than suggested by Yimsir et sl. (2004), ‘The
‘esuls fom this DEM analysis together with earlier FEM
analysis will serve asa Class-A prediction of future fill
Scale tank ests ofthis problem,
ACKNOWLEDGEMENT
“The research work reported in this paper wa supported
bya grant from the Thailand Research Fund (TRE).
[REFERENCES
ASCE COMMITTEE ON GAS AND LIQUID FUEL.
LIFELINES (1984). Guidelines for the Seismic Design
‘of Oil and Gas Pipeline Systems, ASCE Techical
‘Council on Lifeline Earthquake Engineering,
CHANG, C. Sand LIAO, CL. (1994), Estimates of
clastic. modulus for_media of randomly picked
granules. Applied Mechanics Review, Vol. 47, No, 1,
Paet2, pp. 8197-5206
CUNDALL, P. A. and STRACK, 0, D. L. (1979) A.
iste numerical mogel for granular assemble,
Geotetigue, Vol. 29,No. 1, pp. 87-65.
HARDIN, B.O. and BLACK, W. L (1966), Sand stifness
unde’ various tiawal sues Jounal of the Soil
Mechanics ond Foundations Division, ASCE, Vol. 92,
No. SM2, pp. 27-42
ITASCA CONSULTING GROUP, INC. (1999). PFC.
Panicle Flow Cade in 3 Dimensions, Version 20, Val
1233.
MEYERHOF, G. G. and ADAMS, J. 1. (1968). The
Veins ae
Geotechaical Journal, Vo. 5, No.4, pp. 225-244
OVERSEN, NK. (1968). Anchor slabs, calculation
‘methods and. model tess. Bullen No. 16, Darish
Geotechnical Institute, Copenhagen, Denmark, pp. 1
2%.
THOMAS, P. A. and BRAY, J. D. (1999). Capturing
Nowsphercal Shape of Granular Maia with Disk
Clusters. ASCE Journal of Geotech and
Geoenvironmental Engincering, Vol. 125. No. 3. pp
178.
TRAUTMANN, CH. and O'ROURKE, T. D. (1983).
Behavior of Pipe in Dry Sand Under Luteal and Uphit
Loading. Geotechnical Engincering Report 8)
Comell Univesity,
TURNER, J.P and KULHAWY, FL HL (1987)
Experimental Analysis of Drilled ‘Shaft Foundations
Subjected to Repeatod Axial Loads Under ‘rained
Conditions." Report to. Eleevie Power Research
Instat, Come! University, USA,
YIMSIRI, S, SOGA, K., YOSHIZAKI, K., DASARI, 6.
R, and O'ROURKE, T.D. (2004, Lateral and Upward
Seil-Ppeine Interaction in Sand for Deep Embedment
Conditions. ASCE Jounal of Geotech ond
Geoenvironmenal Enginesing, Vol. 130, No. 8
p. 830-842,
94 JOURNAL OF THE SOUTHEAST ASIAN GEOTECHNICAL SOCIETY / AUGUST 2006 INFLUENCE OF SALT ADDITIVE IN LIME STABLIZATION ON
ORGANIC CLAY
', Koslnant K. Onitsuka’ and. Negami?
seuy investigate the eft of sls used aan adtve in ime-stabilzed organic ely. Genealy ime i
‘not the most effective stabilizer for organic clay due to low increase in stength Organic matters have the tendene to coat
{he sol particles, caving the obstruction or delay ofthe pozzlane reaction when the lime isinwoduce. However, the test
results ndiated tat when sodium sat (NaCI or calcium salt (CaCl, is added tothe ime-orgaicelay mixture, the strength
of mixture increases with increasing salt concentration. The sienghinreases by 10 times with an additional of 10% sal.
Furthermore, cementing products very visible in mirographs of lime salt mixcures these results wre concurrent with the
strengh development inthe subiized soil. Comparing between the two sals mixtures, eta 10% salt concentration the
Strength of clay stabilized with lime and sodium chide i higher than clay stabilized with lime and eakium chloride
‘Adversly, ats 2.5% salt concentration, the siength of calcium chloride mixtces is higher. The process of lime salt on
‘organic clay may be ae fllows; salts may coagulate with the cation of organic leaving the cay pariles exposed t the ime
for pozzolane reaction. In ado, sal also can flcculte soil particles and reduce pore space between sol particles,
reauting inthe required cementation o bridging between aggregates relatively lw. Hence, the higher strength of line-salt
‘ubilzaton could be achieved,
Keywords: Lime stabilization, Salt ative, Scanning letron micrographs, Organic cay
INTRODUCTION
Soil stabilization by lime and cement is & common
improvement method used for sabilizing the so clay aed
the dredged sludge. The soft clay found near the surfice
andthe dredged sludge usually contains high organic
‘content, which reduces the effectiveness of lime or cement
sailaton.
The influence of the organic matter has significant
effect onthe strength gain of lime and cement stabilization
‘ue tothe obstruction of pazzalanic reaction. Thompson
(1966) suggested. that the pozzolanie reaction we
‘bstncted in organic oil because sol patclswhic ae &
primary souee of silica andor alumina used in stabilization
Drocess were coated by organic ations. This “masking
effet" prevented the silica andor alumina dissoluson.
resulting in an absence of siiea andlor alumina in
pozzolanic reaction
lowever, not all of the organic compounds have a
negative effec (Tremblay ea, 2002). Many experiments.
were carried out to investigate the possibilities of wsing
various binders including gypsum (Kuno et al, 1989),
catborate-aluminae salt Kamen etal, 1989), plverized
fuel ash, and blast Furnace slag (Hebb abd Farrel, 2003) 0
stabilize organi soils. Not ony the types of binders, but
ko the salt_conconwation plays en imporant role in
Stablzation of organic sil (Onitsuka ct al, 2002). It was
reported that the adverse efles of orgie matter on
satengh of abiized Soil can be misgated in sol baving
high salt concemration (Onitsuka etal, 2003). To extend
this finding in an application, binders of lime with addition
‘of sodium chloride (NaCI) and caleium chloride (CaCI)
‘were examined for organic soil stabilization. Lime was
losted at main binder inthis research because lime was
‘more sensitive with organic than cement (Onitsuka et a,
2003) and recently cement stbiization was found to cause
—Doaart Gakate, Dept of Coil Fagreeiag, Suse
Unive Hone Mash, Sapueh Sap. 8408802 pan
(Carespendingsubor Era O3851@et espe 3)
“proenor, Dept of Civil Engnwenng, Saas Unies,
‘Hone Machi Sap, Sagn 40-850, apa.
rch Associ, Dep of Civ Engneeig, Sega
Univ Honjo-ichs, Sap, Sag, $4850, pan
water and soil pollution by leaching of hexavalent
‘homium (Tsunecka eta, 2003),
In order to identify the mechanism of stabiiation and
chemical reaction during the cementing development, the
{isoved elements in solution were obsorved by chemical
‘composition analysis. The main objective ofthis tidy is
valuate te effectiveness of alvlimo-organie clay
tmixtres in comparison with lime-organie cy etre,
‘The strength and cementing products of mixtures were
‘monitored by unconfined compression tess and seanning
clecton microscope (SEM)
MATERIALS AND METHODS
Materials
Soft clay ffom Rokkaku river bank in Ariake ara,
Japan at dept 0.5 m was used in this study. ‘The cay as
high plasticity with guid limit of 122% and aplastic iit
of 54%, Tho organic matter by ignition loss method and
humic acid are quite high, with value of 12.8% and 12%,
respectively. The clay faction (< 0.008 mm) is 67%
Himic acid is known as one offctors retarding the ability
of soil to react with binders and decreas. the soil
improvement effect (Kuno et al, 1989; Tremblay et al,
2002; Onicuka et al, 2008). Soil cannot be improved
effectively by quieklime i tha sil containing bumic acid
‘more than 1% And water conten higher than 100% (Kune
tal, 1985),
In order to investigate the influence of organic mater,
the organic maser was exacted by the alkaline extraction
method by method suggested by Clare and. Sherwood
(1986), The extraction method iss follows:
1. Natural lay (500g of dry clay) was sired wit 1 Her
(0 0.5 mol Sodium hydroxide (NaOH) for 24 hous by
‘mechanicel shaker.
2. The undissolved cly was removed by cenifiging
snd decamation,
JOURNAL OF THE SOUTHEAST ASIAN GEOTECHNICAL SOCIETY / AUGUST 2006 /95 43. The process was repented and the two portions of
liguid were combined to find the humie acid later.
4. The eated clay was then washed with distiled water,
and the suspension was brought to api of sbout 7.6
by adding hydrochloric acd (HCD #0 a normal level
for natural clay.
5 The ey was seu asin by cei ent
‘The material obtained by the allaline extraction can be
Aivided into two parts, which are humic acid and fulvic
acid. The humic acid was precipitated by aeidiieavon of
the extracted Fiquid to 2 pH of 2. by using HC! and
Fluaton. After extraction, the weate clay (with organic
‘matter removed in the alkaline enriction process), was
sed forthe unconfined compression test by mixing with,
10% of ime at curing peiods of 7, 14 and 28 days.
Some properties of clay were changed because of
organic removal process. Liquid and plastic limits were
lower to about 122% and 49%, respectively and specific
gravity was inereased from 252 0 236. The eay Faction
of treated clay was decreased from 67% to 46%, Properties
ofraural and woated clas are summarized in Table |
Uneonfined Compression Tests
Various binders, including quicktime, normal Portand
cement, sl, and gypsum were chosen forthe stabilization,
All samples were compacied by hand-vibaton 19 reduce
the entrapped air. The specimens, 50 mm in diameter and
100 mm in height, were cured at temperatures of 20°C and
«a umidity of 90% In general, the unconfined compression
fests were performed on the tested samples at curing
petods of 7 14 and 28 days, but fr some stmples tated
‘with lime and Time salts, the tests were also conducted at
curing periods of I, 3, and 5 days
‘Table 1 Properties ofthe so
“Treated Cay
Rokkaku (Extracting
Properties River organle
Clay matter by
NaOH)
Depth fm 05-10
‘Water content, wn (%) 13.2 5
Ligui imi, LL (4) 1416 ms
Plastic imi, PL (6) st 485
Plastic index, PL 375 wa
Specific gravity, Gs 232 156
Panicle sie dition, 6)
Sand (2000-0075 mm) 2 2
Silt(0.075.0.005 mm) a 2
Clay (< 0.005 em) o “6
Color Gay Gay
NaCl gt) 44 1s
anion loss (4) Rs ns
Humic acid 6) 12 :
Unit weight (m3) 142
ptt
(Chemis Composition An
‘The chemical compotion analysis was conducted on
tested samples affer performing, unconfined compression
tests. The preparation procedue is follows:
|The samples were crushed by @ pestle and mora nto
particles unéer2 mm diameter.
2. The samples were mixed with distilled water ata ratio
of 1: 20 (aol ligu),
3. The slury was stived for 24 hours by @ mechanical
shaker
44. A centrifugation for a period of 30 minutes at 3000,
‘pm was performed before Maton to separate the
‘lid from the gu
5. The concentrations of specific chemicals in the liquid
slong with tsp valve were determined
‘The concentrations of cations, such as Ca, Na, Mig and
Fe, were determined by the atomic absorption spectrometer
(Shimaden AA-6580) the concentation of Si was salyand
With’ & spectophotometer (Hach DR2400), The” Al
Cconcenuation was measured by the inductively coupled
argon plasma spectometer (Shimadzu 1CPS-8100), by
Which the pH value wat determined fom the HyO' a
OF ay ions.
‘Scanning Electron Microscope (SEM)
the microsructure of stabilized
samples, samples Stabilized with lime and lime-alt of the
“unconfined compression tests with curing period of 28 days
‘were prepared for SEM. A small pice of sample was cut
with an iron wire before dying by a freeze-dry method to
minimize the volume change. The fractured surfaces of
samples were coated with gold before seanning by the JSM-
3800 JEOL microscope
[RESULTS AND DISCUSSIONS,
Strength Development
Figure shows the unconfined compressive strength of
clay stabilization with various binders. In general, the
Surength inreases with the time, indicating a continuous
pozzolmic reaction from the additives. The stengths of
lay with 10% and 20% quicklime at curing period of 28
ays were 100 kNim® and 199 kN? which ae relatively
low compared to other binders. The strength valuss of clay
with 10% coment; 5% quickime and 59% gypsum; 10%6
(uicklime with 10% NaCl; and 10% quicktime with 10%
CaCl, at curing period of 28 days were G44 KNim!,
579 KNin* 1,295 KNim and 1,039 KN, respectively
‘To cxamine the influence of organic, the organic
content of cay was removed through addition of sodium
hydeonide (susie soda, NaOH). The emul indicated that
the suength of ueated clay-ime increased to 1060 KNIm
ata 28 day curing period.
Among these binders, lime with NaCl and lime with
CaCl, were the two most effective binders for this clay
Figures 2 and 3 sow stess-sin curves of clay, one
stabilized with ime and sodium chloride (NACI, and the
{96 JOURNAL OF THE SOUTHEAST ASIAN GEOTECHNICAL SOCIETY / AUGUST 2006 coer sabilized with lime and calcium eblorde (CaCI) in
‘aris proportions. The figuesindieate that stfness of
organic ‘lay stabilized with lime as improved. with
sddtion of sais. The salt may contibte by two folds, one
4 to cagglate with the organic cation, leaving the’ clay
particles exposed to lime for porzolanic reaction and the
‘other i to ease Mocculaton ofthe so
CALLE L LE EL
Notes
LI'L Na, Cu Tos Ce, denote iting, NaCI, CAC,
‘Treated cy, oral Porand cement and p=,
repeal
2. TOLHona dences 10 of fie an 10% oF NCI
So tedontes safe so wep
Fig. 1 Unconfined compressive strength of
et ormteens
sn
¥ ow a
°
es
ee earner
oat
organs
‘The extnction of organic mater by NaOH is an
fective mothod tht can remove the organic content up 0
80% (Stevenson, 1998), AMer extraction, the Liquid limit
deceased from 1423 to 123%, specie gravity Inresed
fiom 2520238 and stengih inreased from 100 kN to
1060, kN st 28 day curing period with 10% lime
stabilization, despite of no changes in inition loss. These
clearly showed the effect of organic mater in the ely
Decreasing in clay faction fiom 63% 10 46% may be
resulted fom Mocculaion by increasing" of salt
oncenuation afer organic extraction.
= tat 10%
cece ‘sox
Ses 00%
iy i 1%
e567
‘strain %
eo 0
Fig. 3 Resuts of unconfined compression test on clay
‘abiized with 10% lime and various contents of
‘alelum ebloride (CaCh) ata 28 day curing
period.
‘The ignition loss value of clay before and after organic
extraction was quit high about 12.8% The determination
Of the change in mass of a soil sample
ssimction of organic matter by ig
lemperatre is subject to eror. Lambe and Marin (1986)
reported thatthe organic mater determined by the ignition
Joss and the chemical analysis agreed wel if non-organic
Portion (quar, feispar, ete) of the soil are not
Adcomposed by heat and donot contain significant amounts
of adsorbed water. The igution less sa very poor measure
fof organic matter in soils, which have minerals, such as
‘montmorlloites, that have large quantities of adsorbed.
‘moisture, and which ace driven off by heat. The dominant
Clay mineral of Arake clay the Low swelling ype of
Smectite(Ohtsubo eta, 1982). In Aiake cay, te ignion
Joss method overestinais the organic ontent because both
Inorganic and organic constituents ose mass during
heating. The ignition loss value ofthe treated lay was sill
high eventhough the orgnic mater has been removed
Chem
1 Composition Analysis
Prior to analysis, he lime-cly reatton mechanisms are
briefly described as follows, When lime (C20) is added 0
Soil. hydration reaction occurs and. forms Ca(OH).
feleasing about 17x10" J of heat per kg of CaO (6) added
(Boardman, 2001), Some water evaporates due 10 the
Increase inthe temperature by this reaction. The cretion of
Caleum hyttonide neeases inthe pH ofthe pore wate
‘The presene of addtional Ca” may result nam inereas in
‘occulation of partite, modifying ts plasty
Solidification, refering 10 the Iong-erm strength gain
of sil, occurs as result ofa pozzolanc reaction involving
the formation of cementitious compounds within the clay
structure. The compounds are formed by the availabilty of
JOURNAL OF THE SOUTHEAST ASIAN GEOTECHNICAL SOCIETY / AUGUST 2006 97 caloiu, while the alumina and silica oxides dissolve from
the elay structure at such high alkali environment. The
component reacts to form calcium silicate hydrate (CSH)
tnd calcium atuminate hydrate (CAH), cementitious
products tend to bond the clay particles. The degsee of
‘cementation depends greatly on the available Time, clay
minerals and overall environment. Stength gun can
‘continue for several months of even years undet suitable
‘condition (Lit, 1996),
‘The results illustrate the changes in the pl value
(ig. 4) and the elemental analysis of the soltion from
chemical analysis at varying curing periods forthe different
sixtures. The pi value of the Rokkaky River clay was
about 7.6, which is else to neutrality. Aer the adition of
Time, the pH value increases significantly after 1 day 35
expected, There ia clear drop inthe pH vale after 3 days,
except for solution containing. a 10% lime with slight
decrease asa proportion of the strength incease, However,
the pf ofeach mixture was mainaiped 1a sullen level
to allow mineral dsolution, I should be noted thatthe
solubility of AO, and SIO, is pH dependent as israted
inFig S
TNL, Na Ca Te Ce, G doi tine, NC), CHC
“raed ey cement and ypu respectively.
2. LOLONe donner 185 fine ao TDN cP NCI
3. Todentes Sear ary so weight
Fig. 4 pl of clay stabilization with various binders
Exar range of sl pl valuos,
cain,
Sobbty, mmoit
oT2T 4S ST Bo OMI
nga of pH vaues
stl wer andrain
Fig. Solubty of soil minerals at various pH (after
Loughnan, 1969)
ley + Lime 30%
:
i
|
:
2
:
sos
Cann Tie. ays
6 Chemical composition analysis of lay stabilized
with 10% time 24h, 30 revi, 1:20,
ta
* sw
Cy Line 14
b4 * f
i » “|
i
a” rt &
z F so 2
Pa i
a - 8
ee
° “a ©
eae
ovat oe
Fig. 7 Chemical composition analysis of natural lay
stabilized with 10% lime + 10% NaCl 24h, 30
revimin, 1:20 slidiquid)
Figures 6 through 9 indicat the development of ions
concentration including Ca, Na, Si Al and Me forthe 10%
lime, the 103% ime with 10% NEC, the 10% lime with
(CaCl, and 10% lime with treated clay, respectively. It
should be nowd that Fe fons were not detected in any
Solution and Mg concentration was relatively low compared
swith other ons.
‘Tae Ca ions decreased wit ime in all mixtures due to
the progress of reaction. As expected, mixtures of fme with
CaCh gave high coneenintion of Ca ions due to the
sition of CaCl, Inthe same manne, with NaCl, Ns ion
‘wat highest inthe lime with odium-salt mixture
Based on lime-lay reaction, the strength of mature
depends onthe quamity of cementing agent which includes
CCSH and CAH, Calum (Ca) and hydroxyl (OH) were in
excess for the reaction ay indicated in Figs. 5 though 9
“Therefore, the Si and AT concentrations gover. the
‘production rate of the reaction
However, the dissolved silica concentration is much
rate than alumina fo ime and Time-salts stabilization on
98 JOURNAL OF THE SOUTHEAST ASIAN GEOTECHNICAL SOCIETY / AUGUST 2006 ‘uted clay as asad in Figs. 6 through 8. In Figures
and 10, comparing beeen the untreated ely and rated
clay stabilized with 10% lime, it can be seen thatthe
‘isioled silica and alumina concenvation of wexed clay
‘mixture was higher than that of untreated clay Ths may be
resulted fom the removal of organic mates through
‘masking eet, producing higher stent
In case of saltlime mixtures, the silica concentration
‘depends on type of sl rather than the content. nthe lime
sodium chloede matures, the silica concentration was
tach higher than tat dn the unteatd clay whieh was
observed in previous research by Her (1955). I was
‘epted that the sodium exoride (NaCI) greatly aceleats,
the formation of soluble silica. Soil sles disolution
proces in ime-clay reaction may be eataly2d by presence
‘of sodium ions. In Hme-calcium chloride mixtures, the
Silica concentration was slightly lower than that nthe lime-
Sodium chloride mixtures, but higher thn tat in untreated.
slay metre
(Cay Lime 10% CCL 10% SL
KC
Ms —*
oe
Chg oe
Fig. § Chemical composition analysis of watural lay
‘stallzed with 10% time + 10% CaCl, Qh, 30
revimin, 1:20 soliiqui)
“rete Clay Lime 10%
py Sale
edo ;
fe |
Z¢ 8
ia a ai
24 co
M9?
8 oS ws
cure irs, cay
Fig. 9 Chemical compostion analysis of treated clay
stabilized with 10% lime (24h, 30 rev/in, 1:20
soliiiqui)
Te tonteet ites coy
‘Uncontnes Corgressve Seng
may rg.
g
° yo
pes 8078 woo
§
i. -
8
!
a a |
bo 2507s oo
acter cach
Fig. 10 Average values of dissolved silica at diferent
‘mistares and unconfined compresive strength
at 28 day curing period
‘By comparing between the two salts mixtures at 10% of
salt content the stength of clay sabilzed with lime and
cium chloride is higher than clay stabilized with ime and
calkiom chloride. However at 2% of salt coment, the
Strength of caleium chloride mixture is higher, Stevenson
(1958) reported that organic collid canbe coagulated by
clectrolytes such as salt solution, Salt may congulate with
the cation of organic, leaving the cay particles expose to
the lime for pozzlanie reaction. As investigated by SEM
and sol propets, Oita etal. (2003) also stated that salt
Solon can loceulate the sol particles then reducing the
pore spice, therefore, the requited cementation to bridging
between aggregates to grain the strength is relatively low
Consequently by these two ations, the Timeslk mixurs
can yield higher eng.
lemay conclude that not only dissolved silica can affect
fn strength of lime stabilization, but also, salt may
‘ontribate to coagulate with organic cation and reduce poe
space between soi particles
Figure 1 illustrate the proposed schematic of organic
lay partiles before and fer adding Sls i the xe ia
‘according 10 assumption that sas may coagulate with te
‘ation of organic leaving the clay patiles exposed tothe
lime for pozzlanie rection and pore spaces are reduced
‘ue to flection, However, is noted that in economical
tnd envionment aspects additive salt may be cosy and
itutlicing withthe large quantity, it wil be haf 0
JOURNAL OF THE SOUTHEAST ASIAN GEOTECHNICAL SOCIETY / AUGUST 2006/9 = ‘ay paricle
Note: Ciay aggregate contsning humle cid eanne form
Aeovgute
Shaves pore spc by Ros
1) Clay patie after adding alt
Fig. 11 Purposed schematic of organi clay particles
before and after adding salt
cnvironment (Ariizumi, 1977. Therefore, using of additive
salt in ime stabilization shal be performed wit cautions
Microstructure
Scanning elcton microscope observations were done
on the same stabilized samples afer performing unconfined
ompression texts at 28 day curing period. For the clay
Stabilized with ime 10% shown in Fig. 12, « Moceuaing
ssmucure is visible. On the other Rand, for the clay
ilzed with 10% lime and NaCl, with lime 10% and
CaCI; 10%, and weatad clay with 10% ime (Figs 13
through 15), cementation canbe observed, The cementing
products with thin plate-form ereate an archi store
fn and between the ageregtes, which are mainly elem
Slicmte ydates (CSH ~ the main cementing product
responsible for cement seting, Choquette et al, 1987)
CCompacing with Fig. 12, the sol particles are no as dstint
since they are covered with reaction products, and the vids
betveen ol particles are alo less distinc. In summary, the
results were in agreement with strength development he
‘Sabilized sil
Fig. 12 Micrograph of clay with 10% lime at a 28d
curing period
Fig. 13 Merograph of clay with 1096 lime + 10% NaC
‘ata 24 day caring period
100 JOURNAL OF THE SOUTHEAST ASIAN GEOTECHNICAL SOCIETY / AUGUST 2006 CONCLUSIONS.
‘This research work concentates on the effectiveness
and mechanism of organic soil stabilized. by lime and
Seiad alte, Based on the rests, the. following
‘conclusions are made:
1. Lime satilizaton with additional sodium of ealium
salt‘on organic oil yields elatvely high steagth
Compared tha of lime stabilization without anys
The proces of lime-alt on ergani clay may be as
follows salts nity coayulate withthe ation of organic
Jeaving the clay) patcles exposed to the lime for
pozzolanic reaction In addon, salt an Nooculate the
Teil particles reducing pore space, then the ried
emonttion (© briduing between aggregates. is
felaively low, Hence, higher stengih could be
achieved
2. Comparing between the two salts at 10% of salt
ontet the tength of clay stabilized with lime and
fodium ehloide i= higher than clay stabilized with
Time and ealeivm chloride, However at 25% of salt
‘conten, the strength of ealeum chloride mixture is
higher
ACKNOWLEDGEMENTS
“The experimental work by Mx. K._Obtani of Saga
University Is ratefully acknowledged, The aulors also
cxpress their appevition 10 Prof. K. Inoue, Prof J. C
Ghai De T. Hino and Dr. H, Kawakita of Sage University
for providing esuipment fr chemical composition anal
'A special thank yu to Dr TI. Seah of MAA Geotecnies
Co" Lid, Thailand fr his advice and assistance.
Fig. 14 Micrograph of clay with 10% lime + 10% CaCI?
ata 28 day curing period
REFERENCES,
ARIZUMI, A. (1977). Mechanism of lime stabilization.
Journal of Japanese Sactety of Soll Mechanics and
Foundation Engineering, Tsvch-To-Kieo, Vol. 28,
No.l. pp. 916 Gin apaness).
BOARDMAN, Di, GLENDINNING, S. and ROGER,
DF. (001), Development of sbilization and
Sélidfeaion in limeclay mixes. Geotechniqu, Val,
5, No.3, pp. $83-583,
CLARE, KE. and SHERWOOD, PT. (1956). Pure
studies onthe effet of organic mater on te sting of
foil-cement mixtures, Journal of Applied Chemis,
No. 6,97. 317-324
JOQUETTE, M, BERUBE, MA. and LOCAT, 4
(1987). Minerslogieal and micrtextral changes
sssocited wih lime stabilization of marine clays from
Eastem Canada. Applied Clay Setnce, Vol. 2, pp. 215:
ct
232,
HERIB, S, and FARRELL, E.R, (2003). Some experiences
sum fon the sabiizaton” of Irish peats, Canadian
— Geotechnical Jounal, Vol 40, pp 107-120,
ILER, RK, (1955). The Colloid Chemistry of Sioa and
‘Slates. New York: Cornell University Press, 324p.
Fig, 1S Micrograph of treated clay with 10% line at 2
28 day curing period
JOURNAL. OF THE SOUTHEAST ASIAN GEOTECHNICAL SOCIETY / AUGUST 2006 /101 DETERMINATION OF PARTITIONING AND DISPERSION
(COEFFICIENT USING COLUMN TESTS
M.A. Tanchulig"”, J. Takemura’, M. R.A. Khan? and O. Kustkabe!
ABSTRACT: To predict the transport of a contaminant in soi, the paritioning coeficient, dispersion eoeficient and
Porenater velocity ate required. The method of determining pore water velocity has been wellstablised, but farther work
heeds to be done to determine the partioning coefficient andthe dispersion coeficlen. The objective ofthis study i 0
‘evelop a lboratory procedure to determine such parameters. This paper describes laboratory procedure which uss the
concentration depth profile in a laboratory colura. ‘The concenttion depth profile tom the experiment bad a. good
‘agreement with the analytical solution ofthe advecton- Da because of very low pore Water velocity. D, is
centimated using the non-linear least squares estimation
‘method contained in the commorcialy avaiable software
MATHEMATICA’ V4.1 (Wolfam, 1999). The top
boundary is assumed to beat the surface of heey ayer
with the assumpzion that the upper and bottom porous
plate have negligible effets on the omputation of the
parameters
‘The oter parameters needed to solve Ea. 7 are water
content and pore water veloiy. The water content was
measured afer the test. Pore water velocity can be
tbtined from the Aow rato measured st the drinage
‘outlet. I was modified to get a good mass balaooe ofthe
‘contaminant, Which s explained ater.
RESULTS AND DISCUSSIONS
‘Testing and Flow Conditions
‘There are nine test case included in this study,
Iabeled as He, M2, 6, M7, eB Te, M10, Te
tnd Ti-12 (Table 2. The ec presse as the hydealic
Toad was vaced from 29:4 KPa o.98. kPa to be able to
investigate any effect of the hydraulic gradient on the
sorption behavior. Likewise, he sepage time was also
varied ffom 1262 days to $423 days check
esulibriam conditions.
The measured parameters of the testcase ae given
in Table 3. Tho porosity was computed ffom the void
ratio based on the average moisture content measured
after each column tet. The permeability coeficint, K
was messed fom the Darcan relationship, where the
‘velocity i based onthe flow rate ofthe effluent that was
‘measured fom the actual et,
Void
and Permeability
Figue3 shows the void ratio ~ vertical consolidation
pressure telationship of the remolded ite clay withthe
Same intial water content 2 that used forthe column
feats Tn the oodomcter text in which the clog. P
‘elatinship was abtaine, a Tlling bead permeability tet
‘was als coniucted following the procedure deseribed in
‘Amatya (2002)
‘The eog? relation of the normally consolidated
condition is given ase = 2.878 0.76Slog P. Except for
1-6, al the et cases Show a higher vod aio than what
JOURNAL OF THE SOUTHEAST ASIAN GEOTECHNICAL SOCIETY / AUGUST 2006, 105 ‘Table’ Measured parameters
"Test Case Tet Tice me Teel nena
Flow rte, O(a) 321522 378 497437386 42
Nyéraulicconductviy, K(e 10 mis)" 73827284 48° 68 55 Sd
‘Average porosity, 065 065 066 066 0st 066 0.70 0.62
Porewater velocity, (x 104mye 12 668 531 900 22 109 996 106
‘Adj porewater veloiy, adj v(x 10% mis) 7.08 3.08270 580 82 650 656 3.40
Pore volumes (vere) S68 LIL 93168 224246 24818
‘Average void ati, 183182 191194 174198 230 168
Total mas retained in colums, me, 446 654986 155651
“ota input mass (exp) mg 5381007 747A S98 TOK 7K 2363443
‘Tova input mass (ai), ma 532488380377 M7719 t0s2 1712165,
‘abianed Based on flow rate a fant
+ based on experimental veloctty
‘sshased on adjusted velocity
is expected from te tress history. This may be atrbuted
to swelling ofthe clay which occurs rom the end ofthe
tes, when the loads ar taken ofT upto the time when the
Sample is collested. from the column. The sare
bservation was made in Orava etal. (2002) The vid
rato that was used forthe determination ofthe pore water
velowity was based on the acuel values, as measured
fom the water content of the experiment.
5001000
30" 100
log (kPa)
Fig. 3 cog P relation of ite
.
Figure 4 shows th relationship between hydraulic
condvctvity (K) obtained tom the” falling heed
Permeability test. The relationship ean be approximated
Py ult rion i gag wale aK (L860
"22° From Fig. 4 iis also seen that over-consalidation
oes nt ae the hydrulic conductivity ofthe lay, and
that the relationship between the void ratio and the
hydraulic conductivity isthe same for both normally and
‘over consolidated clays. Since hydraulic conductivity
ales dispersion, the findings of this study ean be
compared With Ozawa etal (2000) which stidied the
effect of overconsaidation on dispersion. It was
‘concluded in their stay thatthe vest history has 90
elf onthe determination ofthe dispersion cotficen.
Determination of the Partiioning and Diffusion
Coetficents
The partioning coefficient, K, was determined using
a, (6) with the measured concentration of contaminant
the soil (Cus) at the top layer using the constant
influent concenraion of 480 mg/L as the pore water
concentration, -C. With the partioning coeicint
knowa, the relatdation factor, R ean be solved using Eq
(2). Using the derived, the concenttion-depth profile
(ich isthe concentration a ferent pons ta ine
porosity, and. porewater velocity as Input data the
Aifusion cootcint was estimated in Eq, (7) by ising
non-linear last squresfiting method contained in the
software MATHEMATICA W.1 (Wolfam, 199).
log Hyco condi, K (mi)
‘Nonally consolidate]
Over enolate
log Void ratio, e
Fig. 4 Variation of & with
Figure 5 shows repression curves of 2 cases (l-2
and Ilt-I1.1eshows that thre is incosisteney in mass
alone berween the computed and measured ons. In
“Table 3, the tual mass retained in the column afer the
test are shown to be less with the toa input mass based
‘on the measured velocity. The velocity wis adjusted toa
‘quanti that results ina total mass which ¢ 110% of the
‘ass retained in the column. The pore water velocities
that were used in the compuiations of the diffsion
cveticiem were reduced by 32-99% to be able to have a
oo mass balan. I-12 and Ht] have the grates
(0.32) and least (099) reduction factors, respective)
‘The toa input mass based on the adjusted velocity ate
also shown in Table 3.
‘The velocities used are shown in Table 3, lubed as
the adjusted pore water velocities, The ajstinent in the
velocity canbe justified bythe fice hat the Now rat may
106 / JOURNAL OF THE SOUTHEAST ASIAN GEOTECHNICAL SOCIETY / AUGUST 2006 t $
‘Contaminant concentration in sols su
ome
Contaminant concentration a si, ©
nent
(woe)
Fig. 5 Regresion carves wing experimental
velocities of porewater velocity
have been overestimated by sidewall leakage. The
values corresponding co the adjusted velocity are graphed
‘with the void ratio taken fFom the elog P relation and
shown in Fig 4. Tho measured X values ft wel in he log
‘log e graph, but it should be noted thatthe permeant
‘cd inthe consoldation tet is de-ionized water while
the K values fom the column test used zine nitrate
soaton as permet
‘The background zinc concentration of illite clay was
measured oe 0.0066 mg which is asumed nelgible
‘Thus, the concentration at any point inthe column atthe
start ofthe testis zero. The top boundary i defied to be
a the surface of the clay layer, wih he assumption that
the porous stone as negligible elec. Aer one column
tes, the porous stone was soaked in a 10% nite acid
‘olution (pH = 1, andthe eoncentation inthe acd bath
‘was measured to be 20 mg/l. This is the ine
Cconcentation that the sone is assumes to hold and for
purposes of data analyses, nas considered negligible
‘The results ate shown in Teble 4, Thre test cases
(at.2, lr, te9) have 3 core samples each, I, Iie,
Ie and M11 and Uh-12 have to, and Hl10 has one
Results of the test cass fit fairly well with the cure
defined by the advection-dispersion equation, as shown
‘sully bythe graphs of two fest cases (Figs. 6a ~b) and
xcept for It, quantified by an acceptable conelation
factor? (om 79 10.96). Tae results ofthe eave fing
show thatthe profil is defined oaly onthe upper 2 em
‘art for most of the test cases with concentrations being
eligible at locations lower han that
‘Table 4 Coefficients ofall test eases
R&D
(able) (10min)
T1036 iO
m2 8105.00 829096
Te 789463 379097
m7 849838 330079
mes 730483 308 oh
Ths 126388 393090
M10 16301181 393088
Melt 1580 998 BEG OM
Merz $46 380 393 ons
‘The poritioning coefficient forthe seven test eases
(ith pl equal t 6) ranges from 4.53 - 1181 mbig. The
También podría gustarte
- 48 ZapataDocumento14 páginas48 ZapataDAVID CCALAAún no hay calificaciones
- 20 Iparraguirre Rodríguez 2020 OkDocumento192 páginas20 Iparraguirre Rodríguez 2020 OkDAVID CCALAAún no hay calificaciones
- 24 Luna Yzaguirre 2019 OkDocumento139 páginas24 Luna Yzaguirre 2019 OkDAVID CCALAAún no hay calificaciones
- 22 Larrea Rivas 2019 OkDocumento138 páginas22 Larrea Rivas 2019 OkDAVID CCALAAún no hay calificaciones
- 23 Linares - 2019Documento37 páginas23 Linares - 2019DAVID CCALAAún no hay calificaciones
- 7 Cerna Montalvo 2020 OkDocumento106 páginas7 Cerna Montalvo 2020 OkDAVID CCALAAún no hay calificaciones
- 10 de Paz - OkDocumento200 páginas10 de Paz - OkDAVID CCALAAún no hay calificaciones
- 16 Gamarra Marino - 2021 - OkDocumento34 páginas16 Gamarra Marino - 2021 - OkDAVID CCALAAún no hay calificaciones
- 17 Gambini Zelada 2021 OkDocumento90 páginas17 Gambini Zelada 2021 OkDAVID CCALAAún no hay calificaciones
- 14 Estrada - Pintado - 2019 - OkDocumento95 páginas14 Estrada - Pintado - 2019 - OkDAVID CCALAAún no hay calificaciones
- 13 Eche - Pelaez - 2019 - OkDocumento191 páginas13 Eche - Pelaez - 2019 - OkDAVID CCALAAún no hay calificaciones
- 5 ArbaizaDocumento2 páginas5 ArbaizaDAVID CCALAAún no hay calificaciones
- Espirales en CarreterasDocumento28 páginasEspirales en CarreterasDAVID CCALAAún no hay calificaciones
