University of Nottingham

Department of Manufacturing Engineering & Operations Management

Using Feature-based Product Modelling to Integrate Design and Rapid Prototyping

By
Robert Ian Campbell

Thesis submitted to the University of Nottingham for the degree of Doctor of Philosophy as a staff candidate

April 1998

Abstract

Rapid prototyping (RP) provides a means of producing physical models directly from computer aided design (CAD) data. The aim of this research was to determine the most effective method of integrating RP into the design process.

A review of the links between design and RP was undertaken. This revealed that RP is a technology which can benefit several key areas of engineering design. Many computer tools were identified which supported the designer's but CAD Using RP these this of of geometry alone. most use relied on using incomplete set of design information hindered the integration of RP into the design process.

A hypothesis was formulated which stated that a feature-based product become integrated RP to to an part modelling methodology was needed enable it design demonstrate To the the the methodology, was of process. validity of for (DSS) in design rapid prototyping. support system embodied a The DSS

designers RP, determined through using and a a survey of requirements were full specification for the system was defined. A demonstration version was implemented using a relational database coupled with a CAD system. The demonstration DSS enabled feature-based geometry and non-geometric information to be integrated within a single product model. An application data developed to optimise the the product model which used program was differing finish in RP the to meet surface order model orientation of an

requirements for each feature in a component. This example use of the system illustrated the benefit of using a feature-based product model to optimise the designer's use of RP.

Future work needed to improve the DSS to a state where it would be ready for development into a commercial package was identified. Finally, conclusions how drawn to as all the objectives were met and summarising the original were knowledge by to contribution made the research.

CONTENTS
Contents Declaration Acknowledgements Abbreviations List of Figures
1 vi Vil viii ix ix

List of Tables

CHAPTER ONE
1.1 1.2 1.3 1.4

INTRODUCTION
1 2 4 4

Background to Project ........................................................................... Research Methodology .......................................................................... Research Objectives ............................................................................... Structure of Thesis .................................................................................

CHAPTER TWO:
2.1 Definition

ENGINEERING DESIGN
5

2.2

2.3

Engineering Design Process 7 ................................................................... 8 2.3.1 Specification Stage ....................................................................

Classe 6 Engineering Design s of ............................................................... 6 2.2.1 Class 1( Creative) Design ........................................................ 6 2.2.2 Class2( Innovative ) Design ..................................................... 7 2.2.3 Class3 (Routine) Design ......................................................... 2.3.2 2.3.3 Concept Design Stage 9 ............................................................... Detailed Design Stage 9 ...............................................................

...............................................................................................

2.4

2.3.4 Optimisation Stage 10 .................................................................... Use of Computers for Engineering Design 10 ............................................

2.4.1 2.4.2 2.4.3 2.4.4 2.4.5

2.5 2.6 2.7

Feature-based Design 14 ............................................................................. Integration with Downstream Activities 16 ................................................ Conclusions 17 ............................................................................................ RAPID PROTOTYPING TECHNOLOGY

Electronic Draughting 11 ............................................................... GeometricModelling 12 ................................................................. Wireframe Modelling 12 ................................................................. SurfaceModelling 13 ..................................................................... Solid Modelling 13 .........................................................................

CHAPTER THREE:

3.1 3.2

3.3

18 Introduction to Rapid Prototyping ......................................................... 19 Leading CommercialRP Processes ........................................................ 20 3.2.1 Stereolithography ...................................................................... 21 3.2.2 Solid Ground Curing ................................................................. 22 3.2.3 SelectiveLaser Sintering (SLS) ................................................. 24 3.2.4 Laminated Object Manufacturing (LOM) ................................. 25 3.2.5 FusedDeposition Modelling (FDM) ......................................... 26 Other Rapid Prototyping Processes ....................................................... 26 3.3.1 3D Printing ................................................................................

3.4

3.3.2 3D Plotting - SandersPrototype 26 ................................................ 3.3.3 Ballistic Particle Manufacture 27 .................................................... 3.3.4 3D Welding 27 ............................................................................... 3.3.5 ShapeDeposition Manufacturing (SDM) 27 .................................. 3.3.6 Laser Generation 28 ....................................................................... Applications of RP Technology 28 ............................................................. 3.4.1 CAD Model Verification 29 ........................................................... 3.4.2 Design
Visualisation

3.4.3 3.4.4 3.4.5 3.4.6 3.4.7

3.4.8

Proof of Concept 30 ............................................... ........................ Marketing Models 30 ............................................. ........................ Form and Fit Analysis 30 ........................................ ........................ Flow Analysis 31 .................................................... ........................ StressAnalysis 31 ................................................... ........................
Mock-up Parts

..................................................................

29

3.4.9

Prototype Parts

................................................... ........................ .................................. ........................

32

3.4.10 Plastic Moulding Patterns

32 .................................................. ........................
33

3.5

Role of Rapid Prototyping within the Design Process 34 .... ....................... 3.5.1 Use of RP within Concept Design 34 ..................... ........................

3.4.11 Metal CastingPatterns 33 ....................................... ....................... 3.4.12 Electo-dischargeMachining (EDM) Electrodes 34 ........................ 3.5.2 3.5.3 3.6.1
3.6.2

3.6

Impact of Rapid Prototyping upon the Designer 36 ............ ........................

Use of RP within Detailed Design 35 ..................... ........................ Use of RP within Design Optimisation 35 ...................................... 36 ........................ ...............................
............................ ........................ 37

Increasein Part Complexity

Reduction in Material/Weight

3.7 3.8

7 3.6.3 Reducedneedfor Design for Manufacture ................................. 3.6.4 Increasein Customer Acceptance 37 ...................... ........................ Future Potential of RP 37 .................................................... ....................... Conclusions 39 .............................................................. ...................... LINKING DESIGN AND RAPID PROTOTYPING

CHAPTER FOUR: 4.1

4.2

Ensuring the Effective Use of Rapid Prototyping 41 .................................. 4.1.1 Determining if RP is Appropriate 41 .............................................. 4.1.2 RP SystemSelection 44 ................................................................. 4.1.3 Data Exchangefor RP 47 ............................................................... 4.1.4 Selectionof RP ProcessParameters 49 .......................................... 50 4.1.5 Using RP at Different Stagesof the Design process .................. Software Tools for Rapid Prototyping 51 ............................................ 51 4.2.1 SystemSelection .......................................................................
4.2.2 STL File Manipulation

4.3

53 4.2.3 Direct Slicing ............................................................................. 54 4.2.4 ProcessPlanning ........................................................................ 55 4.2.5 ProcessSimulation .................................................................... 56 4.2.6 Integration of Software Tools .................................................... 58 Requirement Statement ..........................................................................

..............................................................

51

ii

CHAPTER FIVE:

DEVELOPING THE SPECIFICATION FOR A DESIGN SUPPORT SYSTEM FOR RAPID PROTOTYPING

5.1 5.2

5.3 5.4

The Need for a Design Methodology which Supports the Use of RP 61 .... Problems Causedby Not ConsideringRP during the Design Process 62 ... 5.2.1 Unsuitable RP SystemSelected 63 ................................................. 5.2.2 Design is Difficult to Produce Using RP 63 ................................... 5.2.3 RP Model does not Meet Requirements 64 .................................... 5.2.4 Designer is Unaware of New Possibilities 64 .................................
Determining Designers' Requirements for a DSS for RP 65 ...................... Results and Analysis of the Survey of Designers 67 .................................... 5.4.1 General Information 67 .................................................................. 5.4.2 Use of RP 68 .................................................................................

5.4.3 5.4.4 5.4.5 5.4.6 5.4.7

5.4.8

SecondaryProcessingof RP Models 70 ......................................... 70 Effect of RP upon Design Process ............................................ 71 Choice of RP System ............................ .................................... CAD System 72 .......................................... ................................... 72 TransferringData to RP System ................................................
73 Relationship with RP Service ................. ...................................

5.5 5.6

73 5.4.9 Future use of RP .................................... .................................... 74 5.4.10 Conclusionsfrom Survey ................................... ....................... 76 Aim of the DSS for RP ................................... ....................................... 76 Characteristicsof the DSS for RP .................................... ...................... 77 5.6.1 Make RP an Integral Part of the Design Process ....................... 5.6.2 Enable the Designerto Considerthe use of RP 78 Stage in Design Process the at any .............................................
5.6.3

5.7

79 5.6.4 Ensure Right First Time RP Models .......................................... 80 5.6.5 Ensure the Correct Choice of RP Process ................................. 5.6.6 Improve Communicationbetweenthe Designer 80 Operator RP and ........................................................................ 80 5.6.7 Optimise the RP Build Parameters ............................................ 81 Conclusions ............................................................................................ DEFINING THE DESIGN SUPPORT SYSTEM FOR RAPID PROTOTYPING

79 Avoid the use of RP in Unsuitable Circumstances .....................

CHAPTER SIX: 6.1 6.2 6.3

6.4 6.5

82 Objectives .............................................................................................. 83 Design Information Required to Support RP ......................................... 85 Format of Information Required to Support RP .................................... 85 Design for Rapid Prototyping 6.3.1 Feature-based ............................. 86 6.3.2 Definition of Featuresfor RP Applications ................................ 89 6.3.3 RepresentingFeaturesfor Supporting RP ................................. 92 Using Design Information to Support the use of RP .............................. 93 Componentsof the Design Support System ........................................... 95 6.5.1 CAD Data Input ........................................................................ 91 6.5.2 FeatureEditor ........................................................................... 97 6.5.3 RP Usage Advisor ..................................................................... 98 6.5.4 RP Build Time and Cost Estimators ..........................................

iii

6.6

6.5.5 RP SystemSelector 98 ................................................................... 6.5.6 Adaptive STL Generator 98 ........................................................... 6.5.7 SurfaceFinish Optimisation Module 99 ......................................... 6.5.8 Adaptive Slicer 99 ................................ .......................................... 6.5.9 RP Data Output Module 99 ........................................................... Conclusions 100 .................................................. .......................................... IMPLEMENTATION AND EVALUATION OF THE DESIGN SUPPORT SYSTEM FOR RAPID PROTOTYPING

CHAPTER SEVEN:

7.1 7.2 7.3

Choice of Hardware and Software 101 ......................................................... SharedDatabase 102 .................................................................................... Other SystemComponents 104 .................................................................... CAD
7.3.1 Data Input

7.3.2
7.3.3

FeatureEditor

Surface Finish Optimisation

105 ...........................................................................
....................................................... 109

........................................................................

104

7.4

Using the System 115 ................................................................................... 7.4.1 Starting Procedures 116 ................................................................... 7.4.2 Initiating Software Packages 116 ..................................................... 117 7.4.3 Entering Part Data ..................................................................... 7.4.4 Entering FeatureData 117 ............................................................... 119 7.4.5 Readingin STL File(s) .............................................................. 119 7.4.6 AssigningFacetsto SurfaceFeatures ........................................ 121 7.4.7 Running the SurfaceFinish Calculation Procedure ....................
7.4.8 Sensitivity Analysis

7.5

128 7.4.9 Finishing Procedure ................................................................... 128 Evaluation of the System .......................................................................
7.5.1 7.5.2 7.5.3 129 Make RP an Integral Part of the Design Process ....................... 129 Ensure Right First Time RP Models .......................................... Improve Communication between the Designer

....................................................................

124

7.6

Operator 130 RP and ........................................................................ 130 7.5.4 Optimise the RP Build Parameters ............................................ 131 Conclusions ............................................................................................ FUTURE WORK
132 134 135 137 138

CHAPTER EIGHT: 8.1

8.2 8.3 8.4 8.5

Enabling the Direct Transfer of Non-geometric Design

Information from CAD Systems ............................................................ Increasing the Range of Data in the System ........................................... Increasing the Number of Application Modules ..................................... Enhancing the User Interface of the System .......................................... Creating a Commercially Available Version of the System ....................

CHAPTER NINE: 9.1

CONCLUSIONS

To Determine What Links are Required Between the Engineering 139 Design Processand Rapid Prototyping ..................................................

iv

9.2 9.3
9.4

To Design a ComputerisedSystemto Support the Designer's Use of Rapid Prototyping 140 ...................................................................... To Implement the Design Support Systemand Demonstratethe
Benefits its Use will Yield 141 ...................................................................... To Identify the Future Research and Development which is

9.5 9.6

Required to Transform the Systeminto a Commercial Package 142 ............ Original Contribution to Knowledge 143 ...................................................... Review of Progressand Validation of Work 143 .......................................... ........................................................................................... QUESTIONNAIRE USED FOR SURVEY OF DESIGNERS USING RP RESULTS OF SURVEY OF DESIGNERS USING RP DATABASE INPUT FORMS PROGRAM LISTING FOR "LOAD STL" MODULE 145

REFERENCES APPENDIX A.

APPENDIX B: APPENDIX C: APPENDIX D: APPENDIX E: APPENDIX R. APPENDIX G:

PROGRAM LISTING FOR "FACETSIN" MODULE PROGRAM LISTING FOR "SELFACET" MODULE PROGRAM LISTING FOR "ASSIGN_FACETS" MODULE PROGRAM LISTING FOR "CALL MODULE SURF FIN"

APPENDIX H:

APPENDIX I:

EXAMPLE OF NEUTRAL FILE CONTAINING FACET DATA EXAMPLE OF OUTPUT FILE WITH ORIENTATION VALUES RP USAGE ADVISOR QUESTIONS FROM LATEST VERSION OF DSS FOR RP PUBLICATIONS

APPENDIX J:

APPENDIX K:

APPENDIX L:

DECLARATION
I declare that this thesis is the result of my own work. It has not, with the form, been Chapter in different 2, the exception of presented whether same or a to this or any other university in support of an application for any degree other than that for which I am now a candidate. As previously declared within my dated Chapter 2 April 1997, 22nd of this thesis of notification submission, Warwick University to the of contains material which was previously submitted I for in Engineering, degree Science Master thesis the which as part of a of of in obtained July 1994.

R. Ian Campbell 24th April 1998

vi

ACKNOWLEDGEMENTS
I would like to acknowledge the receipt of a New Lecturers Grant from the University of Nottingham Research Committee. I would like to thank my advisor, Dr Phillip Dickens, for his help and advice, during the early stages of this research. particularly Finally, thank you to my wife Yvonne for her help with proof-reading and preparation of this thesis.

vii

ABBREVIATIONS
2D 3D ABS ACIS AP B-rep BS CAD CADCAM
CAE CAM

two dimensional three dimensional acrylonitrile-butadiene-styrene American committee for inter-operable systems application protocol boundary representation British standard computer aided design computer aided design / computer aided manufacturing
computer aided engineering computer aided manufacturing

CAPP CFD CNC CO2 CSG DSS

EDM

computer aided process planning computational fluid dynamics computer numerical control dioxide carbon constructive solid geometry design support system.
electro-discharge machining

FBD FDM FEA

IGES

feature-based design fused deposition modeling (US spelling) finite element analysis
initial graphics exchange specification

LOM MIT MJM MS NC PC RDBMS RP

SDM

laminated object manufacturing MassachussetsInstitute of Technology (US jet spelling) multi modeling Microsoft numerical control personal computer database management system relational rapid prototyping
deposition manufacturing shape

SLA SLS STEP STL UV

StereoLithography Apparatus laser sintering selective STandard for the Exchange of Product data STereoLithography (exchange file format) ultra-violet

viii

LIST OF FIGURES
Figure 1.1 Figure 2.1 Figure 2.2 Figure 3.1 Figure 3.2 Figure 3.3
Figure 3.4

Two-phase researchmethodology followed during project. Stagesin the engineeringdesignprocess. Parametertransformation in surfacemodelling. Methodology behind commercially availableRP systems. Principle of operation for Stereolithography. Principle of operation for Solid Ground Curing.
Principle of operation for Selective Laser Sintering.

2 8 13 18 20 22
23

Figure 3.5 Figure 3.6 Figure 3.7 Figure 3.8 Figure 4.1 Figure 4.2 Figure 5.1 Figure 5.2 Figure 6.1 Figure 6.2.
Figure 6.3

Principle of operation for Laminated Object Manufacturing. Principle of operation of FusedDeposition Modelling. Example of a non-realisableCAD model. Growth of RP sales. Stair-steppingeffect causedby use of layers. Relationshipbetweenvarious RP software tools. CAD model of a castlewith internal staircase. Correlation matrix betweendesigners'requirementsand
system characteristics

24 25 29 38 42 57 65
77

Example of B-rep geometry and associated topology. Example of a feature with its associatedattribute list. Correlation matrix between systemcharacteristics and systemcomponents Using a shareddatabase to integrate designand RP. Mapping an EXPRESS-G entity into an MS-Access table. Structure of the shareddatabase. Example of the facets display seenin AutoCAD. Relationshipbetween different feature editing modules. Schematicdiagramused to define surfacefinish. Relationshipbetweennormal vector angle and surfacefinish. Effect of rotation around X axis upon Vx, Vy and Vz. Effect of rotation around Y axis upon Vx', Vy' and Vz'. Flow diagram of the "calc_surf fin" software module. Component selectedto demonstrateusing the DSS. Optimum orientation of demonstrationpart. Result of sensitivity analysisfor averagesurfacefinish ratio. Result of sensitivity analysisfor worst surfacefinish ratio. Example of "feature attribute list" file createdin AutoCAD.

87 89

EXPRESS-G representation of the DSS for RP data structure. 92

Figure 6.4 Figure 6.5 Figure 7.1 Figure 7.2 Figure 7.3 Figure 7.4 Figure 7.5 Figure 7.6 Figure 7.7 Figure 7.7 Figure 7.9 Figure 7.10 Figure 7.11 Figure 7.12 Figure 7.13 Figure 8.1

94 95 103 103 107 108 109 110 111 112 114 115 122 125 126 133

LIST OF TABLES
Table 3.1 Table 3.2 Table 6.1. Table 7.1 RP leading Strengthsand weaknesses processes. of Classificationof the applicationsof RP models. Information required to support the use of RP. Results of sensitivity analysison surfacefinish requirements. 19 28 84 127

ix

CHAPTER ONE

INTRODUCTION

1.1

Background to Project

RP offers tremendous new opportunities to designers in the way of faster

realisation of designs and the creation of component shapesthat were previously impossibleor too expensiveto make. It can also revolutionise the design process in that design iterations can be performed more quickly and more often. Several alternative concepts can be created as physical models and evaluated in parallel it design. despite However, to the these significant making easier select optimum benefits to design which can be obtained from using RP, the link between RP and the design process is rather tenuous. This is becausemost users of RP transfer
information to the RP system using the STL file format. This necessitates a precise

CAD model, possibly with embeddeddesign intent, to be simplified into a seriesof triangles which are then transferred to the RP system. The triangles approximate the geometry of the design but contain no other relevant information. This can designer's fully do in RP the requirements,therefore meet models which not result in disappointment RP leading the to time, process. money and effort, and wasting

Therefore, it was decided to investigate the requirement for better integration between design and RP with the aim of optimising the effectivenessof the RP process.

1.2

Research Methodology

The two-phase researchmethodology followed during this is illustrated in project Figure 1.1. The initial phaseof the project involved a general review literature of relating to RP followed by a specific investigation into previous work in the area of linking design and RP. This was achieved through library and databasesearches including use of the Internet. Conferenceswhich related to this area were also attended. This gave the author the opportunity for discussionwith others working in the samefield to solicit their views on the proposed researchproject.

Ul O

U
Q)

Q) 1-Y

Review

o-P Engineering -1

Design

I Review
a) O
Q) y

of

Rapid

Prototyping

Review r

of

Linking

Design

and

RP

Requirement

Statement

Analysis

of Designers' (Questionnaire) for Design Design of Design of

Needs

o ,,

SpeciFication Definition Implementation

Support Support Support System

System System System

of

Evaluation

Conclusions

Figure 1.1

Two-phase researchmethodology followed during project.

The outcome of the first phasewas a thorough understandingof the current statein of-the-art terms of the relationshipbetween design and rapid prototyping. It was discoveredthat although many links existedbetweendesignand RP, someof which links, few of these went beyond using simple geometric were computerised information to support RP. Therefore a requirement statement was developed design data link that to stated which a product model which contained all relevant RP. and RP was neededto optimise the effectiveness of

The secondphaseof the project was aimed at satisfyingthis requirementstatement. This involved analysingthe needsand aspirationsof designersusing RP through a for develop then this to a a specification questionnaire and using questionnaire design support system which would use the product modelling approach. This definition implementation-independent then to of specificationwas used produce an the system. Suitable hardware and software were then selected and a

demonstration system was implemented and evaluated. Conclusions were then drawn from this evaluation showing that the use of a product model based for indeed did the the optimising potential provide system computerised future RP. Finally, work required to extend the system was effectiveness of identified.

Throughout the project, a seriesof Gantt charts were used to predict and monitor Also, a strategy of continuouswriting-up was used as much as possible. progress.

1.3

Research Objectives

The following specific objectives were formulated during the initial phase of the researchproject:1.3.1 To determinewhat links are required betweenthe engineeringdesign processand rapid prototyping 1.3.2 To designa computerisedsystemto support the designer's use of rapid prototyping 1.3.3 To implement the design support systemand demonstratethe benefitsits use will yield 1.3.4 To identify the future researchand developmentwhich is required to transform the systeminto a commercialpackage

1.4

Structure of Thesis

As a result of the decision to research into the link between design and rapid
it prototyping, was necessary to gain a thorough understanding of current RP

technology, its role within the design processand what work had been done in the integrating it design. Therefore, the next three chaptersof area of with engineering the thesis addressthe subjectsof engineeringdesign, rapid prototyping technology design linking Chapter 4 RP. the and endswith a requirement statementas of and to how RP must be better integrated into the design processusing a feature-based development describes Chapter 5 the of a specification product modelling system. for such a system and the functionality of the systemis defined in Chapter 6. The implementation of the system and its evaluation are the subjects of Chapter 7. Chapter 8 describesfurther work to be done and conclusions to the project are in Chapter 9. made

CHAPTER TWO

ENGINEERING

DESIGN

2.1

Definition

The word "design" has many different meanings. Even when used in the context of
design there is still no universally accepted definition. engineering Dixon and

Simmons state that "design is the human activity of creating the concepts and the detailed instructions that specify the manufacture of material parts, products, and [1]. systems" Brown and Chandrasekarandefine design as "a highly creative

involving diverse kinds techniques activity problem-solving and many of knowledge" [2]. Sriram et al argue that "design can be viewed as the process of description from that a specifying a of an artefact satisfies constraints arising knowledge" [3]. by diverse sources of number of sources using For Liu and

design is it is Trappey "the essence that of a plan to achievea purpose or to satisfy [4]. a need"

This author will not attempt to give yet another definition of engineeringdesignbut key in The draw the the to quotations. above elements attention some of rather detailed, be from these that are: concepts, statements extracted should words knowledge, process,constraints and need. manufacture,creative, problem-solving, This section is aimed at giving the reader an understandingof what all thesewords design in the context. engineering mean when used

2.2

Classes of Engineering Design

One of the reasonsit is so difficult to define engineeringdesign is that it be can divided into severaldifferent classes. Each class of design is suited to a particular type of design problem. Sriram et al observethat design classes be can thought of as being bounded by the creative-routine spectrum which they divide into four regions: creative design,innovative design,redesignand routine [3]. Duhovik calls the four classesnew design, innovative design, variation design and adaptation
design [5]. Waldron only recognises creative, innovative and routine design,

( viewing redesign or variation design) as a special case of routine design [6]. Brown and Chandrasekaran do likewise but simply use the terms class 1, class 2 defined below. and class3 design [2]. Thesethree classes are

2.2.1

Class 1( Creative) Design

Neither the sources of knowledge required to solve the problem nor the design be known. This type to of requires employed are problem-solving strategies divergent thinking and will often result in totally new inventions. Very little design few designers into falls this are given the opportunity to undertake classand activity is important design design design. However, the the part of most creative creative [7]. be it because to totally generated enables new solutions process

2.2.2

Class2( Innovative ) Design

The sources of knowledge have been identified but the problem-solving strategies in design be knowledge Existing to new applied a new way will are still unknown.

componentsor techniques. Some creativity is required for innovative designto be successful. This classof design is more common than class 1.

2.2.3

Class 3( Routine ) Design

A plan specifyingthe knowledge and problem-solving strategiesto be used already designer The is exists. simply looking for a solution amongst a set of well understood alternatives. This classof designrequires convergentthinking only and design fall into this category. most problems

2.3

Engineering Design Process

As might be expected,there is also debate about what the process of engineering design involves. At present there is not one generally agreed model of the design [8]. process The reason for this is that "models of design are subjective

descriptions of the design process" [9]. Each person'sview of the design process However, depend the models of the their own experienceand opinions. on will design process proposed by Maher et al [10], Sriram et al [3], Ohsuga [11]. Kinoglu and Riley [12], Brown and Chandrasekaran[13], Dixon et al [1] and Smithers [8] all agree that the activity can be divided into stages. The number of in but "consensus" Figure is disputed their shown model a nomenclature stagesand 2.1 contains the stagesthat are generallyagreedupon. These stageswill shortly be discussedin detail. As the process progresses,an increasing proportion of the design be in the class. routine activity will

An aspectof the designprocessthat is widely acceptedis its iterative Some nature. stages or even the whole of the process may need to be repeated before an acceptable solution is found. Each time a stage is repeated the problem to be be different and so the design is actually progressingtowards the final solved will solution. Colton and Dascanio describethis as an upward spiral rather than cyclic iterations on a plane [9]. Additional knowledge and experienceare gained during the iterations. It is important to note that the design processboth begins and ends with the customer.

Specification Concept Design

Detailed Design I

Optimisation

Time

Figure 2.1

Stagesin the engineeringdesignprocess.

2.3.1

SpecificationStage

Also called requirementsdescription stage. The earliest stagein the design process involves translating customer needs or requirements into a formal specification. This should be a list of all the constraints that the designer must work within. It function, information detailed cost, aesthetics, size, weight, on could contain durability, ergonomics, maintainability and anything else that will reliability, safety,

affect the customer'sacceptanceof the design. The specificationshould be part of a design brief that would also give the designer targets for the cost and timing of the design programme.

2.3.2

Concept Design Stage This stage involves

Also called initial, rough or preliminary design stage.

ideas for designs that could possibly satisfy the design specification. generating Several alternativesmay be generatedin which case a selection procedure should be followed to determine which is most promising. This would normally involve
evaluating each concept against key criteria from the design specification. Even if

only one possible solution is generatedit should still be evaluated. It may be that the outcome will be a decision not to proceed to the next stage of the design
is important This design process. an method of avoiding wasted effort.

2.3.3

Detailed Design Stage

After a preliminary design has been selected,the next stage is to gradually add detail to the designuntil it is fully defined. This is often achievedusing a top-down into is into decomposed then sub-problems and approach where a problem individual tasks. The objective is to arrive at a designthat will fully meet all of the design specification and be as easyto manufacture as possible. This is where the bulk of the design effort will be concentrated. The output from this stagewill be a design. "first-cut" the of complete

2.3.4

Optimisation Stage

Also called re-design stage. The detail design is evaluatedagainstthe specification which could possibly have changed. (Some design models treat evaluation as a ) Any deficienciesin the design are corrected separatestage in the design process. and re-evaluated. This is where much of the previously mentioned iteration occurs. This part of the design processwill include work that is often called development, i.e. the evolution of the design through simulation, testing of prototypes and manufacturing trials. The end result will be an optimised design that will be distributed to the customer. manufacturedand

2.4

Use of Computers for Engineering Design

Computershave many roles to play in the engineeringdesignprocess. They can be for data (such finite numerical analysis used element analysis), storage, as word-processing of engineering reports and other non-graphical applications. However, only one aspectof computers in engineeringdesignwill be covered here, that is computer aided design (CAD). CAD has been defined as "the use of a

in to computer system assist the creation, modification, analysisor optimisation of a design" [14]. The characteristicof CAD that distinguishesit from other computer Interactive interactive in design is its graphics allow graphics. use of applications by designer design be the to the product using a created, viewed and modified display unit. visual

The development of CAD has gone through four major phasesroughly coinciding interactive four decades [15]. 1950s During the computer graphics with the past 10

were conceived but could not be implemented due to the poor performance of computers. In the 1960s interactive graphics became a reality with the

development of the Sketchpad system [16]. Soon afterwards the term CAD was by the end of the decade several commercial two-dimensional (2D) coined and systemswere available. The 1970s was the period that saw the introduction of three-dimensional(3D) modelling. This took CAD beyond the field of electronic draughting. Various organisations and standards were initiated to support the
growing CAD number of users. Throughout the 1980s the most rapid

development of CAD occurred.

Surface and solid modelling were developed

leading to a wide range of new application areas. Integration between CAD and computer aided manufacturing (CAM) became a reality. CAD systemsmoved from mainframes to workstations and micro-computers. The 1990s have seen increasinguse of parametric modelling and a drive towards integrating CAD and
CAM with non-engineering functions to achieve computer integrated

(CIM). in history brief Many this the terms of manufacturing of CAD are used below. explained

2.4.1

Electronic Draughting

This term is synonymouswith 2D CAD and emphasises the fact that when used to is CAD BS 708, 2D to simply a conventions such as representgeometry according is in drawing for drawing board. The created exactly the same a manual substitute is Curves then the to product and annotation are used create severalviews of way. Electronic draughting for information the to manufacturing. required added relate improved including less draughting clarity, gives many advantagesover manual

11

repetition of effort and a potential link into 2D CAM. However, they both suffer from the same fundamental weakness i.e. 3D shapes cannot be unambiguously representedby 2D drawings. This is why the emergenceof 3D CAD was such an important development. For the first time the designercould "draw" in 3D.

2.4.2

Geometric Modelling

The heart of a CAD system is its ability to create a computerised model that representsthe shapeof the product being designed. This is known as geometric [14] Since most engineering products are 3D, it follows that CAD modelling . models also need to be 3D (only objects with constant thickness or rotational be symmetry can adequatelyrepresentedusing 2D drawings). Once the geometric it is availablefor all mannerof model has been stored in the CAD systemdatabase, downstream activities. The three types of geometric modelling are described below.

2.4.3

Wireframe Modelling

Wireframe modelling is when the edges of the product being designed are by from Typical equations. curves used curves generated represented mathematical from lines, polynomial equations of are straight conics and splines generated is low degree. The the relatively storage advantageof wireframe modelling varying disadvantage is from Its demanded the that computer. and processing capability the design is not fully defined since there is no representationof the shapeof the faces between the edges. This may be satisfactory for flat faces but is totally inadequatefor complex shapes. By constructing several views of the 3d model, drawings. be detailed It to can can also support used generate wireframe modelling 12

some types of finite element analysis (FEA) and numerical control (NC) part programming.

2.4.4

SurfaceModelling

Surface modelling overcomesthe drawback of wireframe modelling by providing a mathematicalrepresentationof the faces of an object. There are several different representation schemes available but most work on the principle of parameter transformation as shown in Figure 2.2. The values of two parameters (usually input to a set of polynomial equationsto define the values of the called u and v) are Some that the schemes position on x, y and z co-ordinates at surface. use a single define to the whole surfacewhile others use multiple sets, each set set of equations defining a "patch" on the surface. Surface modelling completely describes the in FEA, be to all shapeof an object and can used support shell and plate elements types of NC part programming, renderedimage generationand rapid prototyping.

V-1,1
U V V_

X=f (U,V)
Y=f (U, V) Z=f (U, V)
polynomial eq'ns

X
Y z
1 ,

0,0 U =,

Figure 2.2

Parametertransformation in surfacemodelling.

2.4.5

Solid Modelling

Solid modelling effectively createsa fully defined surface model that is filled with it if lies in 3D For the can ascertain system within space, any point solid material. is [17]. This designed, its being the most the object the object on surfaceor outside

13

form complex of geometric modelling and requires much greater computer capability both in terms of storage and processing. Solid modelling offers all of the benefits of surface modelling and some extra ones. It makes the designer'stask by easier providing "building blocks" which can be combined to create the model. Typical building blocks are boxes, cylinders and sphereswhich can be combined
boolean the using operations of union, subtraction and intersection. Volumetric

attributes such as mass, centre of gravity and moments of inertia can be readily calculated. Preparation of detailed drawings is facilitated by automatic

cross-hatchingof sections and hidden line removal. Solid models can be used to drive automatic 3D meshing routines for FEA although the elementscreated are often tetrahedral and therefore of limited use. Finally, solid models are the
CAD type preferred of model to support the use or rapid prototyping (RP). This is

becausethe "watertight" nature of the model ensuresan unambiguousdefinition of the RP transfer data.

2.5

Feature-based Design

In relation to engineering design, there is no agreement about the meaning and definition of the term "feature" [18]. One simple definition is "a bounded volume be [19]. void" which may contain material or However, the features of a

during design the to specific activities and component are often related bolt, hole is be to accommodate a and may needed manufacturing process e.g. a definition feature drilling by For this reasona more useful of a operation. created a is "a geometric form or entity whose presence or dimensions are relevant to design and manufacturing functions" [20]. 14 This definition also allows non-

features volumetric which are more useful for certain applications. Feature-based design (FBD) refers to the creation of a CAD model using a combination of these geometric forms. FBD differs from solid modelling in that non-graphicalattributes such as surfacefinishesor tolerancescan be attachedto eachfeature.

A FBD package will provide a range of standard features such as holes, slots, bosses and grooves. It is possible to model quite complex components using standardfeaturesalone. However, there will be occasionswhen the standardrange features be of will not sufficient. To overcomethis problem, usersmust definetheir own features. One way of defining a feature is to construct a 2D profile and then it for instance, be This technique to sweep along a vector. a could used create, T-shaped slot. Another method is to modify a standardfeature e.g. create a hole in library bottom. features These are stored a where with a conical user-defined they can be accessedfor future designs. Using a combination of standard and features it is possibleto createvirtually any component shape. user-defined

A logical extension to FBD is to combine it

with parametric modelling.

This

fixed in features be described than to terms of parameters rather enables but be individual features Not the parameterised, also their only can attributes. location with respect to the model. Some commercially available FBD packages flexible [21,22]. tools this to modelling alreadyuse combination provide extremely

The benefits of using FBD include the ability for the designer to create a model function be These the to the of according shapes. shapes can selected natural using 15

component and its likely method of manufacture. When combined with parametric modelling, the advantagesof easy modification and part families are added. A benefit is the integration of CAD and computer aided process planning potential (CAPP), eventually leading to the automation of the planning procedure [23]. The CAPP systemwould match each feature (with its manufacturing related attributes) to a particular manufacturing process. This would facilitate the technique known feature-based as machining.

2.6

Integration with Downstream Activities These are

Much mention has already been made of downstream activities.

software applicationswhich can use a CAD model as one of their inputs. Some of them, such as NC part programming and CAPP, are regarded as being CAM technologies. When CAD and CAM are linked, the term CADCAM is used. When other technologies, such as FEA, are also integrated, the term computer (CAE) is is CAD is It aided engineering only when often used. used as part of thesewider activities that its full benefits can be realised. In general,the higher the level of data stored in the CAD model, the greater the potential for integration with downstream activities. Thus, feature-based design exhibits the most promising prospects.

The core of CADCAM integration is a common database. As the CAD data is is it be in format to other software applications. that accessible created, can stored a Some commercially available CADCAM systems use this method to provide totally integrated design and manufacturing software. A less desirablemethod is to 16

use conversion routines to transform the CAD data into the formats used by other applications. This is a method very often used to achieve integration between software packagescoming from more than one vendor. The possibility exists to integrate non-engineeringapplicationsusing the samedatabase.An examplewould be the,provision of parts lists to aid inventory control. A common database by used all computer applications in an organisation is one of the aims of computer integrated manufacturing.

2.7

Conclusions

Due to its origins in electronic draughting, CAD has concentratedon assistingthe detail design and optimisation stagesof the design process. The designeris forced to think in terms of specific dimensions and configurations. As a result,

CAD [24] do CAD to systems conventional nothing encourage creativity and using

dream [25]. design to as an aid concept was, until quite recently, considered a Conventional CAD systemsgeneratedata of a geometric nature i. e. regarding the in inherently limited This their ability to support the them product. makes shapeof the whole engineering design process [8]. Feature-baseddesign goes a stage

further by enabling function-related attributes to be attached to geometry. The has information in CAD the to model particular relevance ability embed additional including rapid prototyping. to supporting manufacturingprocesses,

17

CHAPTER THREE

RAPID PROTOTYPING

TECHNOLOGY

3.1

Introduction to Rapid Prototyping

Rapid prototyping (RP) has been defined as "a processby which a solid physical is model of a part made directly from a three-dimensional(3D) CAD drawing" [26]. To take account of other sourcesof 3D data (such as medical imaging), the author believesthat the term "CAD drawing" needsto be replacedwith "electronic that provide representation". This definition can be appliedto a range of processes first RP The to tooling. an alternative conventional manufacturing processesand become to system commercially availablewas stereolithography,first sold in 1988 [27]. Several commercial RP systemsare now available, most of which use the
in Figure 3.1 shown methodology

STL File Creation

Slicing STL of Facets

RP Build Process

Layered

RP Model

Contours

Figure 3.1

Methodology behind commercially availableRP systems. 18

The geometry of a computer aided design (CAD) surface or, preferably, solid is model approximated using triangular facets and this faceted format is transferred to the RP machinecomputer where it is sliced into a seriesof two dimensional(2D) contours. These contours are used to drive a layer by layer fabrication process which constructs the 3D physical model. A slight variation to this methodology that is supported by some RP systemsis for the CAD model to be sliced directly and the contours transferredto the RP system.

3.2

Leading Commercial RP Processes

The processesused in the five most common [28] commercial RP systems are describedbelow, together with some of their strengthsand weaknesses. A list of for in Table 3.1 [29]. major strengthsand weaknesses eachprocessis summarised

Process

Strengths

Weaknesses

Stereolithography Unattended Operation Good Accuracy Good SurfaceFinish Solid Ground No Post-curing Curing Nested Components No Su ort Structures SelectiveLaser No Post-curing
Sintering Laminated Object Variety of Materials

RequiresPost-curing Limited Materials RequiresSupport Structures ExcessMaterial Waste Attended Operation Limited Materials Not Fully DenseModels
Rough Surface Finish

Limited Support Structures

No Post-curing

Some Support Structures

Rough Surface Finish

Manufacturing

FusedDeposition Modeling

No Support Structures No Warpage No Internal Stresses No Post-curing Variety of Materials Fast for Hollow Models

Delamination of Models Removal of Interior Excess Material RequiresSupport Structures SeamedSurfaceFinish Slow for Solid Models

Table 3.1

Strengthsand weaknesses leading RP processes. of

19

3.2.1

Stereolithography

Stereolithography was the name given to the first RP process to become commercially available. The original "Stereolithography" systemwas developedby a company called 3D Systems but the term is sometimes generically applied to other, similar systems. The processused by stereolithography is the solidification of a photoreactive polymer upon exposureto an ultra-violet (UV) laser beam (see Figure 3.2).

Wiper

Mac

Liquid Photoreactive Poi yner

ld

Blatforn

RP Model

Figure 3.2

Principle of operation for Stereolithography.

The laser is directed by a pair of movable mirrors to scan across a vat of liquid is initially build distance, below The t, a set platform which polymer. vat contains a the surface of the liquid. The surface of the liquid in the scannedarea is solidified build the to platform. and adheres The platform then descends,allowing the

has just been layer liquid flow The that the to created. over solid resin surrounding its lower height t than then to original position and a wiper ascends a mm platform blade traverses across the vat to level the surface of the liquid. The laser once This layer to to the previous one. again scans create a new solid which adheres 20

processcontinuesuntil the whole model has been created as a seriesof layers, each one t mm thick. The platform is then raised completely out of the resin, the model is allowed to drain and then is ready for is is This the post-curing. where model saturated with LJV light to solidify any remaining liquid polymer trapped in the This model. post-curing can be effected in the RP machine itself or the model can be removed to a separatecuring oven.

Stereolithography models can be built from several different photoreactive resins different with material properties. The minimum layer thickness which can be achievedwith 3D Systems' apparatusis 0.05 mm [30] and it is this parameterthat largely determinesthe accuracy of models. If a model has overhanginggeometry, be support structures must used. Generation of these can be achieved

but they can be difficult to remove and will worsen the surfacefinish automatically build The for Systems' 3D the of model. maximum envelope stereolithography
is apparatus currently 508 mm X 508 mm X 584 mm, although larger envelopes

from [30]. are available other manufacturers

The other vendors supplying

include: CMET, Sony, Meiko Tejin Seiki stereolithography-typesystems and of Japanand EOS of Germany.

3.2.2

Solid Ground Curing

This process is similar to stereolithographyin that it employs the solidification of is by light. However, UV the polymer solidified through photoreactive polymers laser (see lamp Figure instantaneous 3.3). UV than to scanning a exposure a rather The lamp is shone through a series of masks that are generatedfrom the profile

21

contours. After each lamp exposure, non-solidified resin is removed by a vacuum head and replacedwith wax. The resin and wax are then cooled, machinedflat and coated with a fresh layer of polymer. The process is repeated using a different for layer mask each until the model is completed. The finishedmodel is surrounded
by wax which can be removed by washing with hot water.

v
Mask

UV Light

Wax

Support

Resin

Model

Figure 3.3

Principle of operation for Solid Ground Curing.

No additional supports are required with solid ground curing since the surrounding

inherent build is It to several parts nested within support. possible wax provides
is layer 0.06 Minimum thickness mm and the one another such as an assembly.

[30]. X 500 is X 350 build 500 mm mm mm envelope approximately maximum Solid ground curing is a complex process with several operations and therefore is by Cubital The large the sold under process and expensive machine. a requires trade name of Solider.

3.2.3

SelectiveLaser Sintering (SLS)

This processusesa scanningCO2 laser. The laser is used to selectivelysinter a thin layer of powdered material (see Figure. 3.4). The powder can be wax or a A thermoplastic material such as nylon or polycarbonate. piston within the 22

cylindrical build chamber which contains the powder is lowered, a fresh layer of powder applied using a roller, and the processrepeated. Each layer is also fused to the one below. The completed model is surrounded by unsintered powder from it be which must removed. This is undertaken using a brush or compressedair. Chemical wiping of the part is also used to improve surface finish. Although the surrounding powder gives support to the model as it is being built, additional supports are still sometimesnecessary.Minimum layer thicknessis 0.076 mm and
the maximum build envelope is 300 mm diameter by 380 mm deep [30]. Much

is being research work conducted into using SLS with metals, composites and ceramic materials.

Sear,
Ri

Unsinte Powder

Piston

Build Cylinder

Figure 3.4

Principle of operation for Selective Laser Sintering.

SLS has been commercially availablesince 1990 and is sold by DTM of the USA. A similar systemto SLS has been developedby EOS of Germany.

23

3.2.4

Laminated Object Manufacturing (LOM)

This processuses a CO2 laser to cut profiles from sheet material (seeFigure 3.5). The surrounding unwanted material is crosshatchedby the laser. The sheet of material is then indexed by rollers and bonded to the previous layer. The processis then repeated. When the model is finished, it is removed by breaking away the surrounding crosshatchedmaterial. This can causeproblemsfor enclosedvolumes. The materials available are paper and polyester film. Models made from paper have a wooden appearanceand are susceptibleto moisture ingress if not treated
with a waterproof coating. The surrounding material acts as an inherent support.

The minimum layer thickness for LOM is 0.05 mm and the maximum build
is envelope approximately 550 X 810 X 500 mm [30].

Bonding

CrosSho. Mo.thripl Builc PlCLf

Indexing Rollers

Figure 3.5

Principle of operation for Laminated Object Manufacturing.

LOM has been marketed by Helisys of the USA since 1992 and similar processes Sparx Sweden by Scale Kira Singapore, Japan, Kinergy-Hust of and of of are sold

24

Models Unlimited of the USA. The two latter processesrequire manual stacking layers. the of 3.2.5 Fused Deposition Modelling (FDM)

This process involves the extrusion of semi-molten thermoplastic material through heated a orifice onto a fixed base(seeFigure 3.6). The extrusion head is moved in the X and Y directions while each layer is being deposited and in the Z direction between layers. The raw material (which can be wax or plastic) is supplied in the form of a filament. Extra supports are sometimesrequired and the processis more is hollow build. layer Minimum to thickness to suited parts as solid parts are slow
0.051 mm and the maximum build envelope is 254 mm cubed. The FDM system is

by Stratasys of the USA. marketed

Fit ame nt

F-ilane Spool or Head

Model

Build Plat fora

Figure 3.6

Principle of operation of Fused Deposition Modelling.

jet Multi developed have 3D Systems a raster-based extrusion process called Modeling (MJM). This has the advantage of the build-time being almost

independentof the model geometry.

25

3.3

Other Rapid Prototyping Processes

There are many other RP processesbesidesthose described Some above. of them are commercially available,but still not very widely used, whereasothers are still at the researchand development stage. A selection of those which are of particular interest are describedbelow.

3.3.1

3D Printing

This process has been developed at the MassachusettsInstitute of Technology (IVIIT) and works using the principle of selective bonding of powdered material. The processis quite similar to SLS except that the powder is bonded with a liquid binder rather than a sintering laser. A layer of powder is deposited and a plotting head (similar to that used for inkjet printing) is used to apply the binder to the desired area. A new layer of powder is depositedand the processrepeated. When the model is completed, it is removed from the unaffectedpowder and fired to cure the binder. A commercially available application of this technology is Direct Shell Production Casting which createsceramic moulds for metal casting directly from a CAD model. Researchis being conducted at MIT into producing parts with micro
[31,32]. internal 3D textures printing structures using and micro surface

3.3.2

3D Plotting - SandersPrototype

An American company called SandersPrototype have developed an inkjet RP build for for heads, deposition the material and one one system that uses two from is build The the thermoplastic supports are made and supports. material a is flat deposited, layers has been After to the model machined one or more wax. is for building. Layer thickness approximately continued provide an even surface 26

0.08 mrn [30] When the build is completed, the wax supports must be removed by washing with kerosene.

3.3.3

Ballistic Particle Manufacture

This process deposits droplets of a molten build material through a small orifice using a piezoelectric pump. The deposition head is moved in 3,4 or 5 axes and builds the model in a layerwise manner. The fact that the droplets can be deposited from a wide range of directions means that the need for supports is virtually eliminated. The processis marketed by BPM Technology of the USA.

3.3.4

3D Welding

This is a techniquewhere a metal inert gas welder attachedto a robotic arm is used to deposit steel or aluminium in a layerwise fashion. The robot is programmed to
is 3D The the that create shape required. process uses a similar methodology to

FDM but is aimed at producing metal prototypes directly.

The work being

has been Nottingham University at able to produce models with similar conducted finish [33]. When to combinedwith a rotatable sand castings accuracy and surface build platform, 3D welding can be used to construct models using severaldifferent build orientations. This reducesthe needfor additional supports.

3.3.5

ShapeDeposition Manufacturing (SDM)

This technique, developedat CarnegieMellon University, usesa combination weldbased deposition, computer numerical control (CNC) machining and shot-peening deposited, being build Besides RP the a support material to create metal models. during is deposited (also the typically metal) construction. model around material

27

CNC machining is used to create a smooth surface on the deposited material and shot-peening is used to relieve stresses. SDM avoids the stepped surface finish exhibited by most other RP processes [34].

3.3.6

Laser Generation

This involves directing a high-power laser onto the model being built and, simultaneously,feeding a cladding material into the laser spot on the surfaceof the feed The part. system can use material in the form of wire [35] or powder [36].
Although typically used in a layerwise manner, it would be possible to construct build parts with a more complex pattern.

3.4

Applications of RP Technology

The ability to create 3D physical models directly from electronic data can be used design the to support several activities within and manufacturing engineering be for It non-engineering applications such as creating can also used process. imaging data. However, from only engineering applications are medical models discussedhere. Kochan statesthat RP models can be divided into three categories: design models, function models and manufacturing models [37]. Within each of (see Table RP different models applications of these categories, there are several 3.2). Each of these applicationsare describedbelow.
Design Models Function Models Manufacturing Models

CAD Model Verification Design Visualisation Proof of Concept Marketing Models

Form and Fit Anal sis Flow Analysis StressAnalysis Mock-up Parts Prototype Parts

Plastic Moulding Patterns Metal CastingPatterns EDM Electrodes

Table3.2

RP Classification the models. of of applications

28

3.4.1

CAD Model Verification

It is not always possibleto check that CAD is just by looking a model realisable at a computer screen. It is possible to construct CAD models that cannot exist in
reality, especially when using surface modelling (see Figure 3.7). Even with solid

modelling, some systems allow the user to create undesirable self-intersecting shapeslike the one shown in Figure 3.7. Making a "hard-copy" using RP is one method of checking for these problems.

Reo. lisalo (e
Figure 3.7

Non - realisable

Example of a non-realisableCAD model.

3.4.2

Design Visualisation

In terms of being able to assessshape, size and ergonomic efficiency, there is is RP hand. This in holding like can offer where one's an object nothing quite designersa tool for their own use and as a meansof communicationto others. A involved in be to the everyone presented and quickly created physical model can Although this introduction promises reality sort of virtual process. product [38,39]. it likely RP is to completely replace not capability

29

3.4.3

Proof of Concept

This involves using an RP model to determineif design is feasible a concept at an early stage in the design process. This is advantageous because it prevents excessive expenditure on concepts that are doomed to fail. An example of this is application the creation of a coffee pot RP model which was used to assess the pouring characteristics of the design [40]. Obviously, a coffee pot design that is cannot pour well not worth developingfurther.

3.4.4

Marketing Models

This application really takes design visualisation a stage further by introducing the RP model to potential customers. It can be used to solicit their opinions at an early is be desired The RP there time to changes. model can stage while still make finished, e.g. by smoothing and painting, to take on the exact appearanceof the final product. This is of particular value to products that have a high degree of images high Some CAD of quality, rendered systemscan produce aestheticappeal. do 2D However, brochures. for these and are only use within marketing products handled. be to the product not allow

3.4.5

Form and Fit Analysis

This is the simplest type of functional analysis that can be performed on a is the is the that The RP to correct component check used model component. An fit into its to parts. example of adjacent with to mate and given envelope shape RP the be gears, shafts and of models the this would assemblyof a gearbox using housing. Once again, CAD systemscan be used to perform assemblychecks and

30

clearanceanalysisbut there is no substitutefor the engineerbeing able to check that the gears actually meshcorrectly.

3.4.6

Flow Analysis

Certain products have a requirement for aerodynamic or fluid dynamic testing. Examples are car bodies, engine manifolds, ship propellers, missiles and shower heads. It is possibleto predict some performance values using computational fluid dynamics (CFD) but these packagesare not 100% accurate. RP models that are very close to the shapeof the finished product can be used for full-scale or scaleddown tests. These include wind tunnel experimentsand cold-flow enginerunning. However, if the test environment is harsh, either through high temperatures or fluids, into a more suitable the RP model may have to be post-processed corrosive Post-processing later. material. of RP modelsis discussed

3.4.7

StressAnalysis

It is possible to use RP models for stress analysis despite the fact that they are in final the seldom made production material. The stressanalysisresults obtained from tests upon the RP model can be utilised for the final part by allowing for differencesin material properties. The results obtained are by no meansdefinitive, but they do allow for comparisons between several design variants [41]. This

decision before be design to the the to manufacture real optimised part enables is components made.

31

3.4.8

Mock-up Parts

For some engineering applications it is of vital importance that a "mock-up" assemblyof the evolving product be availablefor everyone on the design team to inspect and evaluate. Examples this of are automotive under-bonnet models and aerospaceengine models. Normally such mock-ups would have to use modelinterpretations makers' of designs or wait for a long time until prototype parts became available. With the advent of RP, it is now possibleto have fast physical into the full-scale mock-up replications of the CAD model data, ready to assemble in a very short period of time. Problems exist where the mock-up is for a large RP build product. machine envelopesimpose practical size limitations. One way to overcome this is for a scale model to be made. An alternative,used by Boeing for the 777 aircraft, is to replace the physical mock-up with the electronic CAD representation [42]. This will work with a new product that has been designed totally on CAD, but physical mock-up still has an important role to play in
combining new components with older designs.

3.4.9

Prototype Parts

There is a growing number of "engineeringmaterials" that can be produced directly

FDM SLS. Examples ABS RP the system and nylon on machines. are with with

Although, the material properties may not be identical to the final product, they are functional for field As be tests. trials the to other and used often close enough it is likely functional RP that metal systems widens, material capabilities of from CAD directly be models. will also produced prototype parts .

32

3.4,10 Plastic Moulding Patterns An RP model can be used as the master to for tools create plastic moulding. One is method to surround the model with liquid silicone rubber, allow the rubber to set, into the model to divide the rubber into two parts, remove the model and mate cut the two rubber moulds to create a cavity. Vacuum casting can then be used with polyurethane based materials to create a limited run of parts with similar characteristicsto engineeringplastics. A secondtechnique is to use the RP model to create spray metal tooling. The part is mounted in a frame and thin layers of molten alloy are deposited until a shell of around 2mm has been created. This is then "backed-up" with a composite material into which cooling channelscan be incorporated if necessary. The RP model is removed and the process repeated from the opposite direction to create the other half of the mould. When the two injection for limited A be together, they run moulding. moulds are mated can used further possibility is to use the RP models themselvesas the injection mould tool, known as direct tooling.

3.4.11 Metal CastingPatterns build An RP model made of a suitable material such as wax or using a specialised investment for The be Quickcast' Systems' 3D casting. can used pattern such as is it. built into is dipped times up around until a shell a ceramic slurry several model
into be leaving burned is The model then melted or out a cavity which metal can

is A during RP this Note the that process. major problem model sacrificed poured. with this technique is the possibility of the shell cracking while the model is being in be RP Sand burned out. models exactly the created using casting moulds can

33

sameway as using conventional wooden patterns. The modelsare used as patterns which are placed in the sand and resin mixture. When the mixture has set, the patterns are removed and molten metal is poured into the resultant mould. Unlike investment casting, this processallows the RP model patternsto be re-used.

3.4.12 Electo-dischargeMachining (EDM) Electrodes EDM requires a conductive electrode that has the negative shapeof the tool it is being used to create. RP could be used in three ways to provide such tool. a Either a conductive RP model is created directly, or a non-conductiveRP model is coated with a conductive material, or the RP model is used as the masterto form a conductive electrode. Current researchwork is aimed at producing a commercially viable techniqueto createEDM electrodesfrom RP models [43,44].

3.5

Role of Rapid Prototyping

Design Process the within

Relating the applications of RP listed above to the stagesof design describedin Chapter 2, it is obvious that RP can play a major part in the total design process. As soon as a complete geometric model of the design is available on a CAD different functions. it is The RP to to many possible create models perform system, design is described detail RP the and optimisation stagesof within concept, role of below.

3.5.1

Use of RP within ConceptDesign

The main task of concept design is to generate and evaluate ideas. RP can help by design ideas providing proof of concept and visualisation with evaluation of designs be design The complete, e.g. only exterior surfaceson or will not models. 34

a telephone. The computer model required for RP could be created using an engineering CAD package. Alternatively, a "styling" or industrial design system could be used [45]. A computer model created using such a system would not contain as much detail as a CAD model. There are also developmentstowards an interactive virtual reality system which can be used to create an electronic representation [46,47]. Whatever route is used, as soon as the computer model

hasbeen completed, a physicalprototype can be createdfor by evaluation everyone in the design team, including prospective customers. In this way, RP can help to engendera concurrent engineeringapproachright at the start of the designprocess.

3.5.2

Use of RP within Detailed Design

The aim of detailed design is to transform the selected concept into a complete "first-cut" design. This will typically involve many design engineersworking in parallel on different aspectsor componentsof the whole product. RP can help here by providing models for CAD verification and by creating a "mock-up" of the is important It to point out that RP could be misusedin that complete assembly. designers could use it as a "safety-net" to catch errors that should have been detected on the CAD system. There is a fine balance to be stuck here between it it "cheap" be RP designers that to easyaccess will used without making so giving for a trial-and-error approach [48].

3.5.3

Use of RP within Design Optimisation

This part of the design process takes the first-cut detailed design, evaluates it iterative loops, design through the arrives at specification and, a series of against involves design be Much this then of the optimum which will manufactured. 35

simulation, testing and analysis. RP has an extensive role to play both in the provision of models for various types of analysisand in the creation of prototype parts either directly or through the use of a secondarytooling process. The major benefit of using RP in design optimisation is that the time and money neededfor individual iterations is greatly reduced. This resulting in a better finished design [49] iterations that means more are possible

3.6

Impact of Rapid Prototyping upon the Designer

The range of uses to which RP can be put indicatesthat it is a key technology that be into taken should considerationthroughout the designprocess. However, this is
only part of the argument for integrating RP into the design process. As RP

increase, it will offer more possibilitiesfor the designer. It is no longer capabilities just a quick way of making models, but rather a new set of manufacturing be indirectly (through directly that processes can secondary or used either impact have finished future RP The to that processes) produce will upon the parts. [50]. designer described below to possibilitiesavailable a are

3.6.1

Increasein Part Complexity

The fact that RP is a freeform process meansthat complexity of shape is not a limiting factor. Therefore, more complex parts which perform several functions lead lead This in increase time be or cost. will created without any significant can in in to a reduction the number of parts required an assemblywith a subsequent in design and manufacturingcosts. decrease

36

3.6.2

Reduction in MateriaVWeight

RP enables strength to weight ratios to be optimised using variable internal structures, large thin walls to be created and parts to be made without the requirementfor wasteful machiningfrom stock.

3.6.3

Reduced needfor Design for Manufacture

Parts can be designed without the need for draft angles, parting lines, fixture location holes and other constraintsimposedby particular manufacturingprocesses. This will reduce design effort and the need for time-consumingdiscussionsabout manufacturabilityrequirements.

3.6.4

Increasein Customer Acceptance

Parts can be designed to meet customer demands that would otherwise be impractical due to manufacturing considerations. For example,sculptured surfaces feasible. aesthetically are pleasing are entirely which

3.7

Future Potential of RP

RP has yielded many benefits including quicker model making, improved design

less tooling and an overall prototype evaluation, expensive and visualisation it has As design in length a result, and manufacturingprocess. of the reduction the been readily adopted by many manufacturers leading to an impressive rate of ) [51]. (see 3.8. Figure in RP machine sales worldwide growth This has been

in RP by increasing a appearing on articles of research number an ever matched in With the journals activity so much and conferenceproceedings. wide range of

37

field, it is not surprising that new RP systemsand applicationsare constantly being developed.

In future years, RP machineswill become faster, more accurate,easierto use and less expensive[39]. This should also lead to them becoming more commonly used. Already they are competing with many conventional manufacturingtechniquesfor prototype and tooling production. It is also likely that they will soon be used for batch small production [52]. Consequently, RP should receive the same

consideration during the design process as more traditional manufacturing techniques. Furthermore, RP is already changing the way the design process is flexibility designers learn how As increase RP to the techniques executed. of and RP Therefore, fully designs this, the themselves will change. exploit nature of more inherent but be "bolt-on treated part rather as an simply as a useful extra" must not its it be Only design to then the maximum effectiveness. can used process. of

800 600 100 200 0

No of RP
System s

Sold

1992 1993 1994 1995 1996 Year

Figure 3.8

Growth of RP sales.

38

3.8

Conclusions

The field of rapid prototyping technology is growing in severalways. More people it, are using the processesand materials available are increasing and the range of its it be for is be It that then, to applications can used expected are widening. impact upon engineeringdesignwill also grow. It is essentialthat the link between design and RP is fully understood and that any weaknessesin this link are identified. This is the subjectof the next chapter.

39

CHAPTER FOUR

LINKING

DESIGN AND RAPID PROTOTYPING

The current link between design and rapid prototyping has two elements. The first is the procedure or methodology followed by the designerwho is considering the use of RP. The detailed procedure followed when consideringthe use of RP will
from company to company and even from one designer to another. However, vary this author has identified four key steps which should be part of any procedure

in described designers. by These RP the aimed at ensuring effective use of stepsare the first section of this chapter. Failure to follow any one of these stepswill result in a weakenedlink between design and RP.

The second element of the link between design and RP is the computer software (or least designer followed by the the to at some parts of procedure used support it). The aim of such software is to assistthe designerin the decisionsand processes be A before be RP the of software which review can used. must undertaken which is currently available or being developed is given in the second section of this
chapter. It will be seen that although a large amount of software has been

developed,there are inadequaciesin both the integration of software tools and the in leads This the to data they of weaknesses source a second use. which range of link between designand RP.

40

Having identified the weaknessesin the link between design and RP, this chapter ends with a "requirement statement" which first of all highlights the needfor better integration between design and RP addressthis need. and then goes on to state what is required to

4.1

Ensuring the Effective Use of Rapid Prototyping

RP has an important role to play in shortening lead-time, reducing costs and

improving quality. However, RP may not always bring improvements to the design

be It process. must used effectively. Therefore, whenever the use of RP is being four key followed be by designer there the contemplated, are steps which must for be is Each the to and/or other support personnel. step a pre-requisite next be The RP the appropriate performed. end result will a model, created using technique with the optimum build parameters,which fully satisfiesthe designer's discussed below. These stepsare requirements.

4.1.1

Determining if RP is Appropriate

It would be foolish to assumethat RP is always the most appropriateway to

building Indeed, the a physical model even necessityof produce a physical model. images CAD The be and rendered advent of photorealistic questioned. must
immersive virtual reality systems has encroached on some of the traditional roles

has build However, [38]. to the that for physical models a model need assuming include its be RP been establishedas genuine, must evaluated against rivals which CNC machining. and model-making manual

41

RP has both strengths and weaknesses. On the positive side, it is totally freeform, requires no tooling or fixtures and can make almost direct use of CAD data. However, its range of materials is limited, its layers use of gives rise to a "stairstepping" effect (seeFigure 4.1) and it is relatively expensive. Therefore, before its be justified, it is necessaryto draw a comparison between RP and the use can available alternatives. This must be done on a part-by-part basis as the balance between the pros and cons will be different in eachcase.

Surface

of Cl

RP Model Layers

Figure 4.1

Stair-steppingeffect causedby use of layers.

for Kochanhasdeveloped a methodology comparison which usescost andtime as

its main criteria [50]. For both RP and CNC milling, cost models have been

in involved into tasks the take each technology. account various produced which These cost models require input about the geometry of the specific part in question, is decision Ideally, CAD the the that on model already available. and assume has been be before CAD RP taken the model created. should whether or not to use impose (most RP because is the the This may some restrictions on model use of for it surfaceor solid modelling) or alternatively may create notably the requirement

42

new possibilities e.g. the creation of several alternative CAD models to undergo concurrent prototype evaluation. This presents a quandary as it is difficult to produce reliable time and cost estimates until after a CAD model has been produced. However, it is possible to estimatewhether build time and cost would be less for RP than CNC simply by asking some pertinent questions about the component being designed, e.g. how complex is the design, what size is the component, what material must the model be madefrom.

The decision to use RP or some other model-making process should be jointly taken by the designer, who knows the model's requirements, and support know the capabilitiesof RP and the other processes. If RP is being personnel,who find "stand-alone" it be difficult to technology, operated as a support may in knowledge. if it is being However, this personnelwith used a model shop that
be has CNC facilities, The to this a problem. access manual and should not also

if hindered "over-the-wall" for joint be decision an a could also requirement is likely if functions. This between design RP the more an and relationship exists facility. in-house is being but bureau could also exist with an used outside

The objective of this first step is for the designerto ascertainif RP should be used. Assuming that its use has beenjustified, the CAD model can now be created or has been Once into the RP this completed, next step taking consideration. modified is to selectthe most suitableRP technique.

43

4.1.2

RP System Selection

The range of commercial RP systemsavailable is always expanding. This means that a choice has to be made as to which RP technique to use. For some organisationsthis will be a one-off decision concerningwhich machinethey should buy. Several one-time-purchase system selection methodologies have been

suggested. Most of these methodologieswork on the principle of following several steps. In the methodology developed by Kepner and Tregoe, [53] the first step is to list the organisation'sabsolute requirements. Then, a list of system capabilities that would be desirable but not essential is generated and importance factors item. is Each to that the assigned each system offers all of absoluterequirements is desired The 1 10 then assessed to the rating on a scale of capabilities. against by for item The the the optimum summed. each and products multiplied weighting be highest the the one with overall score. systemwill

The methodology described by Burns [54] begins with a decision on whether or by is This buy RP the costs of running a to comparing achieved an system. not is follow identified. The have been benefits to that a next step system againstthe benchmarkingprocess similar to that of Kochan (see below). The final stageis to in they RP the terms offer. support the of systemvendors assess

RP [55] Prototyping Rapid OSTEM of a range recommends The 1993 report For different for be companies who example, to applications. considered systems Solider. SLA LOM, large and consider should castings of sand number a make

44

The problem with this kind of prescriptive approach is that it soon becomes outdated as new technologiesbecome available.

For Kochan, [37] the selection methodology takes the form of benchmarkschosen from one of three categories: 1. 2. A single benchmarkpart that is representative of all a companiesproducts. Several benchmark parts that represent the families by part produced a company. 3. A "challenge-testparts"benchmarkwhere complex parts are designedwith the expresspurpose of testing the capabilitiesof RP systems. The benchmark approachhas also beenfollowed by Chrysler [56] who undertook a detailed cost comparison of several systems for a small automotive part. Challenge-part benchmarks have been developed by the SLA User Group [57], Lart [58], Juster and Childs [59], luliano et al [60], Jayaramet al [61] and as part Intelligent Manufacturing Systems Test Case on Rapid Product the of Development [62]. The weakness with all such benchmarks is that they are

different benchmarks to the characteristics of one part and restricted will give different results.

A problem common to all the above methodologies is that they are aimed at facilitate do RP decision. They the selection not of an making a once-and-for-all increasingly is becoming desirable basis. This as capability systemon a part-by-part large companies and RP service bureaux have made deliberate decisions to

45

purchase several systems to increase their flexibility. choosebetween these systemsfor eachpart they build.

Therefore, they have to

Some work has been done in the area of RP for individual parts. systemselection Optimat have developed an RP feasibility [63] assessmentmodel which gives consideration to individual part characteristics. This model consists of a table
lists RP systems against part parameters relating to function, size and which

geometric features. For each part parameter,the RP systemsare rated as having distinct feasibility or limited feasibility. Another attribute of this model is that it comparesRP technologies againstCNC and sheetmetal forming. In this way it is decision also acting as a aid on whether to use RP or a conventionalprocess. The decision the combining of about using RP with the choice of RP system seemsto be logical. However, one problem with this strategyis that the cost and timing data be be RP to to compare required processesneeds very accurate and will not decision has CAD been Ideally, the on whether or created. model calculableuntil a CAD be before RP the to model. made creation of a will not use

Once again, the choice of RP process should be a joint decision between the designerand RP support personnel. The outcome of this step will be a decisionto is The build to RP step prepare to next model. particular a system a particular use for the transfer of the model data from CAD to the RP system.

46

4. L'

Data Exchange for RP

There are several different ways of representinggeometry within a CAD system, Bezier e.g. surfaces,B-spline surfaces,B-rep solids and CSG solids [64]. Also, different CAD system uses its own proprietary data format. As each it is a result, desirableto have some sort of common format to transfer data to RP or neutral
systems. One such format has become dominant within the RP community and is

known as STL [65].

This format allows for the surface of an object to be

by approximated a series of triangular facets. It is a simple but rather inefficient

method as there is duplication of data for the co-ordinatesrepresentingthe corners STL of eachtriangle. A thorough treatment of the strengthsand weaknesses the of format is given by Jamiesonand Hacker [66].

STL is by no means the only available format for transferring data from CAD to

RP. Other faceted formats are available[67,68] and there is also the possibility of CAD format IGES VDA. In or all of exchange such as using a general purpose these cases, the aim is to pass a complete representation of the component from CAD RP. to geometry

A second,radically different approachis to make use of the fact that all commercial RP systemsare currently based on layered manufacture. This meansthat at some be into has With 2D to 3D converted contours slicing. via the representation stage has from data been CAD done is transferred the the STL, this system. after
involves is that However, there an alternative strategy creating the slices within the

is This RP these to the CAD model and transferring systemas a seriesof contours. 47

often referred to as direct slicing and RP exchangeformats have been developedto support this methodology [69,70]. It is a much more efficient method of

transferring data as only the geometry that is actually required is passedto the RP system. However, it dictatesthe direction of build for the RP processwhich can be disadvantageous (see section 4.4) and it causes some difficulty with support generation.

In terms of transferring CAD geometry, the designer must choose between the different alternativesavailable. The choice will very often be determinedby what formats are supported by the CAD systemand the target RP system. An exchange file is then generated and transferred to the RP system via a direct link, e.g. the Internet, or through a "hard" medium such as a 3.5 inch diskette. In theory, the RP be built model can now using the appropriate build parameters. However,
files be degree exchange must always of repair validated and sometimes require a

fundamental is does A the that more work. problem exchangeof geometry alone determine information build RP to the the operator all optimum required not give by drawing file be The a or textual accompanied exchange should parameters. document that defines other attributes of the part such as the material to be used,
finish, factors, tolerances, etc. surface shrinkage required Even then, dialogue

between the designer and the RP operator will be necessary to ensure continuity of

design intent.

This should prevent the "over-the-wall" attitude to RP that is

designers. by displayed sometimes

48

4.1.4

Selection of RP ProcessParameters

This step of RP implementation is normally undertaken solely by the system operator who is totally familiar with machinecapabilities. It hasbeenlikened to the process planning function for conventional manufacturing operations [71].
However, Kruth states that there is not the same for defining requirement operation sequences as the part is produced in one operation [72]. This would be true if the

itself model was the end result. However, when one or more secondaryprocesses are used, operation planning does come into play [73]. Starting with the STL (or file format) other exchange representationof the CAD model, the optimum process
for the chosen RP system must be selected. Some of these parameters parameters

individual be is build for RP there to specific systems,e.g. stereolithography will Others during (ACES QuickcastT"1), laser depth, recoating. are or style cure z-wait build layer include RP orientation, systemsand choice of material, common to most thickness and critically, choice of secondaryprocessing.

The selection of process parameterscan have a profound effect on the quality of the RP model. The material properties, model accuracy, surface finish and

dimensionalstability will all dependon the parameterschosen. Therefore, although has designer it is by RP be that the the essential they will operator, chosen be the fully to that parameters. select the used will requirements model conveyed for build determine the time the cost and Also, the choice of parameterswill partly
RP model. For example, build time largely depends on the maximum vertical

by build determined in be This the turn orientation. dimension of the model. will

49

For these reasons relating to quality, cost and timing, it is essential that the optimum process parametersare selectedfor each RP model that is built. When left to the judgement of the RP operator, the quality of the decisions made will depend heavily upon individual knowledge long as as there is an adequatesupply and past experience. This is satisfactory

of personnelwith theseattributes. However,

as the RP market continues to expand and the variety of systemsgrows, it will becomemore difficult to find operators with all the necessary skills.

4.1.5

Using Rapid Prototyping at Different Stagesof the Design Process

The four steps described above should be followed every time a model is required during the design process. Severalmodels may be neededat different stagesof the design process, each with its own requirements. During the concept design stage, model accuracy and material are not important issues as the requirement is for proof of concept and visualisation models. Detail design requires models for CAD "mock-ups", hence becomes verification and accuracy now essential whereas is flexible. design For choice of material still optimisation models, both accuracy and material selection are crucial as the models will be used for analysis and be secondary processes where particular material properties will required. Therefore, it is likely that different RP techniques and build parameterswill be for different The types complexity of this procedure and the of models. selected development lead be have decisions the to that of many must made number of in discussed RP. These the next section. to tools support are software

50

4.2

Software Tools for Rapid Prototyping

The fact that RP is a computer-based technology it makes especially suited to application-enhancing software tools. Many such tools have been developed

through researchprogrammes and some are commercially available. These be can categorised into several areas, each one dealing with a particular problem in the below. application of RP. These areasare discussed

4.2.1

System Selection

Although several selection methodologies have been suggested, only a few software tools to aid in this process have been identified. As part of a computer aided process planning (CAPP) package for RP, Muller et al have developed a databaseof systemcapabilitieswhich can be evaluatedagainstpart definitions [73]. Additionally, the databasecontains information on NC milling and consequentlyis decision use database RP. The to also acting as a approach aid on whether or not to system selection has also been used by Phillipson and Henderson [74]. In this for build the time to and cost several system,a seriesof algorithms are used predict RP techniques. Narayanan et al [75] have developed a rule-based expert system is RP The advisor. user asked a series of questions and the which acts as an inference mechanism of the system uses the answers to recommend which techniqueto use or buy.

4.2.2

STL File Manipulation

The problems associatedwith the use of the STL format have long been recognised [76]. As a result software tools are required to help overcome these problems. One such tool is the standardsoftware packagewhich is supplied by 3D Systemsas 51

part of their Stereolithography system. This package enablesthe RP operator to view the STL file, scale it and reposition and reorient it. It also checks for any in errors the file e.g. missing facets, and will attempt to repair any such errors so that slicing of the file can commence. This package is rather limited in its capabilities and so other tools have been developed to provide improved functionality [77,78,79,80]. These software tools enable STL files to be split, A particularly imaginative

merged, shaded, converted to other formats, etc.

example of software for STL file checking is the use of virtual reality to move the inside STL file [81]. Software has been developed at Colorado the user around State University which will import STL files and then allow the user to "sculpt" the functionality [82]. is STL It triangles the that mesh of of resulting clear is manipulation software advancingrapidly.

Some RP systemsrequire support structuresto be generatedbefore a model can be built. Once again, 3D Systems supply this facility in their software package

[83] "Bridgeworks" third-party and through a choice of two modules called "MAGICS" [84]. These modules will automatically create the required supports for a given STL file. Sometimes, more supports than absolutely necessaryare for Other be tools these. to software created and manual editing can used reduce Stevens by Institute been developed have the of the generation of supports includes POGO [86]. The International [85], POGO software Technology also and
STL editing capabilities.

52

Once an STL file (or files) has been checkedand any necessary supports

generated,

the next stagefor the computer software is to slice the file. Again, there is standard software available for this from the RP system vendors but alternative tools have been developed by others. Clemson University's CIDES software [87,88] allows STL files to be viewed and modified, supports to be generatedand slicing to be undertaken. A similar workbench of tools has been developedby DeskArtes under the name of Rapid Tools [89]. In addition this software can also create STL files from neutral CAD formats such as IGES, perform boolean operationson STL files from STL files. and create offsets

4.2.3

Direct Slicing

The principle behind all of the tools in Section 5.2 is that becauseSTL has become the current "de facto" standardfor RP, the industry'srequirementis for better ways direct files. is handling An that these slicing of the original alternativeviewpoint of CAD data offers major benefits over using STL and that software tools should be directly CAD benefits The to this model are of slicing a strategy. provided support listed by Jamiesonand Hacker [66]: 1. 2. 3.
4.

Reduced file size Greater accuracy(no approximation with facets) ReducedRP machineprocessingtime (no slicing)
Elimination of repair routines (assuming correct contours)

direct been have the to capability of slicing Several software tools created provide [66,90,91,92,93]. In each case,the software also incorporates the facility of

between be This (or the to tailored spacing slices enables slicing. variable) adaptive

53

for the changing geometry of the part. Where the geometry is fairly constant,large spacing is used, where it is changing rapidly, smaller spacing is required. This further improves either the accuracy or speed offered by direct slicing. As

mentioned in section 4.3, direct slicing requires a slice format to be used and an alternative method of support generationis also needed. Materialise have provided a software tool [94] which can use severaldifferent slice formats, allows for slice checking and repair, automatically creates build supports from slice data and facilitates conversion from STL to a slice format.

4.2.4

Process Planning

In terms of the automatic selection of optimum process parameters,much of the research has been in the area of optimising build orientation. Orientation

algorithms have been developed by Allen and Dutta [95], Frank and Fadel [96], Kim et al [97], Cheng et al [98] and McClurkin and Rosen [99]. (The software developed by McClurkin and Rosen also considerslayer thickness and laser hatch ) density as part of an overall build style optimisation. The objective of these

is following to the objectives: support optimise algorithms one or more of finish build time, and trapped volumes. accuracy, part surface structures,

Optimum build orientation has also been consideredas an integral part of the total
in being developed Germany [73]. BIBA The at system process planning strategy

developed aims to accept and verify neutral format CAD data, to create for from CAD data RP the model, to accept and verify slice triangulated and sliced data in various formats, to automatically generatesupport structures and to select

54

the most appropriate RP system and secondaryprocess(es). A different integrated process planning strategy is being followed by the University of Michigan [100]. This system can accept several formats data, of slice create a solid model through the slices, optimise orientation, generatesupports and output either a solid model STL file. It can be seenthat thesetwo systemsaim to integrate many of the or an capabilities of the other software tools which have been described in the above sections.

4.2.5

Process Simulation

The cost of creating a model using RP can be very high. If a model is built using poorly selectedprocess parameters,the outcome can be an overlong build time, or flawed is helps One this that to avoid simulation of model. strategy worse still, a the build process. If the computer simulation indicates a problem, process is This be parameterscan reselecteduntil an acceptablesimulation result obtained. hence At builds time. the one cut costs and number of poor and should reduce level, simulation can involve build-time estimationwhich will help to determinethe for for Build-time information estimators scheduling. model cost and provide Stereolithography have been developed by Yu and Noble [101] and Chen and Sullivan [ 102]. At a more sophisticated level, finite element analysis and

distortion, be dynamics fluid the to amount of predict can used computational build [103,104,105, flow during distribution the process and material temperature
106,107].

55

4.2.6

Integration of Software Tools

Many developments are afoot that have the aim of providing software tools to support the use of RP. Some of these tools are commercially availablebut most are at the researchstage. Of course, the most important software tool to support
RP has been commercially available for some time, i. e. surface and solid CAD

modelling systems. The relationships between these various software tools are in Figure 4.2. It can be seenthat these relationshipsare quite complex with shown
several possible alternatives in their order of usage together with a requirement for

feedback and iteration between different tools. There is a requirement for data
in The the use of representation and storage at many stages overall process.

independent data stores would lead to duplication and possible translation problems.

The use of the "stand-alone" software tools currently being developed (as leads by Figure 4.2) to a number of problems: represented 1. Designers and/or RP operators are presented with a complex suite of these. through route process and an uncertain programs 2. J. Data is duplicated and data translation errors can occur. There is no automatic feedbackto designers.

in is integrated is it terms taken that approach To avoid these problems, an essential

from This tools to the work data of software enable all would representation. of "umbrella be interface A to database. as an act should used user single the same in As lead the through them the to order. for correct user tools the and system" database. In be it data this the to way, is added the creates would each tool used, 56

feedback is automatically provided to the designer have who would continual access to the database. Such a collection of integrated software tools which provided a range of RP support facilities for the designerand RP operator could be termed a "design support systemfor RP"

CAD Moclel

STL File Generation

Neutral Format File Generation

I STL

File

Neutral

File

Direct Slicing

STL File Verification

STL Fite Slicing

Slice File

A[ernative

Representations

of

Part

Geometry

RP Process
Simulation

Model Reo's RP System Selection

RP Build optimisation

IPerornance

ISvsten ChoiceI

IOriented PartI

KEY TO SYMBOLS EXt, Entity

RP Process

Supports Generation

Process

Build Data

DotQ store
Figure 4.2 Relationshipbetweenvarious RP software tools. 57

The closest that any of the previously discussed tools get to being a complete design support system is the overall process planning system being developed at BIBA [73]. A common data structure has been for this system which envisaged will represent the geometry of the part(s) to be made using RP. This highlights a failure common to all of the tools discussed i. above, e. concentrating upon the geometric aspects of the design alone. During the design process, many other types of information besides geometry are created and used. These will have an impact on how RP should be used. Therefore, a truly integrated design support for RP, should cater for the use of non-geometric information as well as system part geometry.

The need for this "product model" driven approach has been recognised by Carleberg [108] who has proposed using STEP application protocol AP204 [109] to representboth a part's geometry and the processparametersrequired to drive an RP system. The use of a STEP application protocol has also been recommended by Steger et al [I 10] who comment that this would enableproduct data including features, be for to tolerances topology, materials and used planning and geometry, production of prototypes. It is this author's opinion that a product model

incorporating all relevant design information must be used to support the whole from RP to tools system selection processsimulation. software range of

4.3

Requirement

Statement

design is iterative has that Chapter 2 engineering a complex and processthat shown by immensely the adoption of computerisedtools. Integration of helped been has 58

design with downstream processes is desirable and feature-baseddesign offers great potential in this area. Chapter 3 described the expanding range of RP technologies and applications. This expansionis enablingRP to play a greater role in the design process, and the future impact designer be the upon will substantial. For these reasons,it should be treated as an inherent part of the design process. In this chapter, the need for an RP systemto be used effectively has been discussed. This requires RP to be used only when appropriate,the most suitableRP systemto be selected,a comprehensivedata exchangefrom CAD to RP and the optimisation of RP process parametersfor individual models. It is likely that RP will be used during times the design process, each time with different model several Failure to recognisethe necessarystepsfor effective use of RP will requirements. in result a weakened link between design and RP. The review of the RP software tools shows that major failings common to nearly all of them are their lack of integration and their concentrationupon geometric data alone. A "product model" by these the to approach offers potential overcome weaknesses using all relevant designinformation to optimise the use of RP.

Therefore, a "requirement statement" was formulated to provide an end goal for this research: -

The effective use of RP requires the use of a "design support for rapid prototyping". system Such a system must be used

is design RP the the process, whenever throughout use of design The that system must ensure all relevant contemplated. 59

information is used to optimise each task in the application of RP, from deciding whether or not to use RP to the selectionof process parameters. Therefore, a product model which can incorporate both geometry and other designinformation is required. A featurebased design approach has the potential for providing this comprehensiveproduct model.

The development, implementation and evaluation of a design support system for rapid prototyping are the subjectsof the remainderof this thesis.

60

CHAPTER FIVE

DEVELOPING THE SPECIFICATION FOR A DESIGN SUPPORT SYSTEM FOR RAPID PROTOTYPING

There have been severalattempts to promote the needto take the proposed use of RP into account during the design process. CAD considerations such as the in detail requirementfor a fully enclosedsurfaceor solid model havebeendiscussed [I 11]. The impact of using RP on the total design processhas also been discussed [112,1 13]. These authors conclude that RP can be used to improve the whole development product process. Therefore, the designershould not simply produce a CAD model which meets the design specification and then passthis on to the RP design integral This RP "bolt-on" treats than to the service. as a an processrather it. part of Besides, this "over-the-wall" mentality runs totally contrary to the

is This RP to that often used support. author concurrent engineeringphilosophy for "design (DSS) design that rapid a support system which embodies a argues into design integrate For is RP the to process. prototyping" methodology required been [114,115, has for the time recognised such a methodology need now some 116] but remainsunsatisfied.

5.1

The Need for a Design Methodology which Supports the Use of RP

into designer RP during to take be the that a It could easily argued account asking design process is imposing an unnecessaryconstraint upon the design. Many design design functionality to besides to exist, e. g. already cost, constraints design design for for To for design manufacture, reliability, etc. quality, schedule, 61

add yet another, simply to facilitate the model-making process, must surely be undesirable.

If RP was still limited to producing prototypes then this attitude would be justified. However, rapid prototyping has developed into rapid tooling and even rapid manufacturing. RP is no longer just a quick way of making models, but rather a new set of manufacturing processesthat can be used either directly or indirectly (through secondaryprocesses)to produce finished parts. Consequently,it should receive the same consideration during the design process as more traditional
manufacturing techniques. The most effective way of achieving this is to give

designersa methodological approach to follow when consideringthe use of RP. Such a methodology must overcome all the problems associated ignoring the with downstream use of RP. It must also allow for the fact that the use of RP can design the sometimeschange process, e.g. the use of a physical model to replace detailed drawings or CAD images during design review meetings. The most design is through the a computerised effective way of applying methodology support system.

5.2

Problems Caused by Not Considering RP during the Design Process

If RP is treated as a "bolt-on" process,i. e. the CAD model is first completed and little discussion, RP the a range of problems can service with then passedon to occur.

62

5.2.1

Unsuitable RP System Selected

One potential problem is that the most suitable RP technique will not be recognised. The range of commercially available RP techniques is continually widening. Each technique has its own particular characteristicsand capabilities. An essentialpart of a DSS will be to guide the designer in the choice of which technique to use. This decision will depend on several factors including the size and shapeof the product, its desired material properties and the uses to which the RP model will be put. Therefore the DSS must enablethe designerto match the needsof the designwith the capabilitiesof the RP techniquesavailable.

5.2.2

Design is Difficult to Produce Using RP

A design could be created in such a way as to make it unnecessarilydifficult to incur during RP. This RP time the construct using process extra cost and will its its benefits. has Each RP tends to technique own specific negatesome of which "trapped have For producing great problems requirements. example,some systems have i. the restricted access. model which volumes", e. volumes of space within This can sometimes be avoided by building the model in several stages. Alternatively, the design can be altered to eliminatethe trapped volume. The CAD is being If RP RP be the that to the tailored used. process model must therefore be it have is CAD to is then that either will unsuitable model service given a Both by RP these the for system operator. altered or modification returned lead-time into because the they extra add product are undesirable remedies lead loss latter Also, to the an accidental of the alternativemay developmentcycle. be less is design If then than intent. the a optimum model will not altered designer's

63

created.

Therefore the DSS must ensure that the designer is aware of the process.

implication of the designupon the RP

5.2.3' RP Model does not Meet Requirements Designers normally use RP when they have a specific application for the RP model

to perform e.g. it may be for designverification, test and analysisor to facilitate the production of tooling. In each case,the RP model must be capableof meeting the requirements that the designer has in mind. If it does not, this will result in a RP increased wasted model, cost and time, and dissatisfactionwith the RP process. Possible causes of an unsuitable RP model are poor surface finish, insufficient accuracy and wrong material specification. It is desirablefor the RP model to be first information on model time. This can only be achievedif all the necessary right is from designer to the RP service. Therefore, the the requirements communicated DSS must ensurethat this information is produced and madereadily available.

5.2.4

Designer is Unaware of New Possibilities

The final problem that could be encounteredis that an opportunity to improve the
design process using RP is overlooked. New manufacturing possibilities are being

be final Not by RP tooling techniques. and products only can prototypes, realised difficult but CAD from directly that were previously also products a model, created feasible. becoming A trivial this impossible to example of are now manufacture or is the internal staircasein the model castle shown in Figure 5.1. A more useful described by [117]. Burns Using in is the as realm of microfabrication application systemsthat are only RP techniques,microfabrication can create electromechanical

64

few in a microns size Designers must be encouragedto make use of these new possibilities.

Figure 5.1

CAD model of a castlewith internal staircase.

Increasingly, designerswill have accessto severalRP techniquesand will need to be aware of the design implications of using each one. In order to avoid potential follow DSS them to a which enables problems and exploit new capabilities, a full is RP to consideration a necessity. methodology giving

5.3

Determining

Designers' Requirements for a DSS for RP

developed, it determine designers' be DSS to the Before the was necessary could for have it then to these to meet and create a specification use requirements would designers by This function. it conducting a survey of how would was achieved

65

have who made use of RP.

The survey was designed to ascertain how the

designers used RP, what benefits they had seen, any problems that had been
encountered and what they would like to see happening in the future. Many of the

potential problems (as listed in section 5.2) had already been identified but it was important that thesewere corroborated by evidencefrom the "field". Likewise, the

author had his own ideas about what future improvementsneededto be made in the link between design and RP but these had to be supportedby wider opinion to
make a strong argument for implementing them.

The process used to determine which questions should be asked during the survey

was one of conception, evaluation and improvement. It was decided that the designers follow RP four deemed for the survey of using should steps necessary the from in discussed Chapter RP This 4. the effective use of results as would enable the survey to be used to support the developmentof a DSS which embodiedthese four steps. Therefore, an initial seriesof questionswas conceivedby applying the four Some design the to general questions steps. experience each of author's own by These design the questionswere evaluated environment were also produced. on in Several had RP. the the use of more experience author's colleagueswho one of
improvements These identified suggested. and were used to shortcomings were

develop a new series of questions which were used as the basis for a structured interview with the CAD SystemManager of a company which was making regular interview, During [118]. this RP any questions which were ambiguous, of use identified. be to expanded were needed which or unnecessary Also, several

list the the of questions of modified. structure was and added were questions 66

Finally, to check the validity of the survey questions,another structured interview was conducted using the revised series of questions [119]. This progressed

without any problems and so it was decidedto go aheadwith a larger scalesurvey designers of using RP. The survey took the form of a questionnairewhich was sent out to 55 people in 49 companies. The companies were selected from three Centre for in Rapid Manufacturing Prototyping the sources: a) members of at Nottingham University, b) members of the UK Stereolithography Users Group and c) members of the UK Rapid Prototyping and Manufacturing Association. The

final form of the questionsused for the questionnaireis shown in Appendix A.

5.4

Results and Analysis of the Survey of Designers

Completed questionnaireswere receivedfrom 26 people in 23 companies. One of these was unusable since the company had not yet used RP although they were intending to do so in the near future. Therefore, the usableresponserate was just but in B Appendix full The 45%. the a summary survey are shown results of over is given here under the headings used in the questionnaire. Throughout this highlighted have been from identified the survey summary, all specific requirements
by using bold text.

5.4.1

GeneralInformation

in designers these There was a wide range of company sizes and number of but different industry The over sectors of many represented companies companies. Therefore, from the aerospace,automotive or electrical goods sectors. 50% came diversity for have this to cater the DSS will The vast majority of respondents

67

(over 80%) were practising designers or engineeringmanagerswho were closely involved with the use of RP models.

5.4.2

Use of RP

Most respondents (56%) had heard of RP through Magazine articles or from colleagues. Almost half (48%) had started to use RP becauseit could help them reduce lead-times. Therefore, it is important that the DSS does not slow the design process in any way, preferably the opposite would be true. However, many other reasonsfor adopting RP were also stated. The most common criteria by far in the decision to use RP were project timing, cost and model complexity. However, once again, many other criteria were stated. There were many reasons for but RP (part justify RP) these too to stated not using only one of simple use of by was quoted more than three respondents. The DSS needs to guide designers RP. for types to as what of component are suitable

There was a wide variation in the number of models made in the last year from five leading functions 100. The to that these models were put to were over under form and fit analysis, sales/customerstudies and design approval. Only 24% of lack have from RP that them a of availability of would stopped respondentsstated is in fact (68%) This the that reflected most respondents still producing models. This techniques. that shows most companiesare model-making other of make use in design is However, RP the the this to of role models process. expand not using lot has RP to to offer, especially a as an aid concurrent engineering. one area where The DSS must enable designers to identify every possible use of RP in the

68

design process and then they can it. go on to evaluatethe appropriateness of using There was a wide range in the proportion of models madeusing RP from lessthan 10% to 100%.

The most common criteria for assessing the quality of RP models were accuracy (quoted by 64% of respondents)and surfacefinish (52%) with severalrespondents quoting both. However, many other criteria were also quoted. These criteria need to be stated by the designer and values placed upon each one. A majority of respondents (60%) had been satisfied by the quality of over 90% of their RP models. Every respondentwas at least able to say that they had been satisfiedwith "most" of their models. There were many reasonsstatedfor poor model quality. The only two which were quoted by more than three respondentswere poor finishing be RP The to model needs accuracyand poor of model. quality of each evaluated against the criteria the designer. by specified In this way,

designer, from be thus the reaching models prevented would unsatisfactory increasingconfidencein the RP process.

The most important benefits of using RP were quoted as speed by 76% of by design Many by 16%. 20% other and verification respondents,cost reduction benefits were also quoted and severalrespondentsstated more than one benefit. durability RP The most often quoted problems encounteredwhen using of were (12%). However, (16%) accuracy respondents most unsatisfactory and models he in Part these of problems may had had no major problems. unrealistic

69

expectations by the designer. The DSS must indicate to designers what capabilities they can expect with confidence from RP.

5.4.3

SecondaryProcessingof RP Models

Every one of the respondents had used RP models for secondary processing. Vacuum casting was the most commonly used process (72% had of respondents it) investment used with casting the next most popular (44%). Spray metal tooling direct tooling for wax parts had also been used. The majority (68%) of and had respondents not used RP models to create prototype tooling but over half of these had considered doing so. All designersshould at least consider using RP to aid the creation of prototype or even production tooling.

5.4.4

Effect of RP upon Design Process

The responsesto the survey indicated that RP is being used throughout the design
it for for development layout design, 20% 20% after process with using concept

design had been completed, 16% using it prior to committing to tooling, 8% at the it design 36% the to using at severalof these end of process prior production and
is further DSS This to the must aim encouraging and stages. promote the

in While 52% design RP the of process. only at all stages effective use of (96%) designs, RP had the to vast majority evaluate alternative used respondents had used feedback from RP models to make design modifications. Again, it seems its full RP Once designers to of potential. making use a concept are not that many it likely be it is If to been to has design are minor. not the chosen, modifications be incurred lead-time design then and cost will and profits reduced. extra optimum designer design. in the be the selecting optimum concept RP should used to aid 70

Similarity, 60% of respondents stated that RP had changed their design process while 76% saw it as an essentialpart of the design processfor at least some their of products. The only changeto the designprocessthat was quoted by more than one respondentwas an increasein the usageof 3D CAD (20%). Other changeslisted included earlier verification of design, earlier discussionswith toolmakers and the ability to design more complex shapedproducts. Each designerneedsto determine how RP can increase the overall effectiveness the design of process, rather than simply speedingup parts of it.

5.4.5

Choice of RP System

Every one of the respondents had access to Stereolithography while no other systemwas availableto more than 32% of respondents. This meant that in 20% of cases,no choice of RP systemwas available. In 40% of cases,the choice of system by designer, in 36% of casesby someoneelse in the company (e.g. the was made CADCAM Manager, Project Manager) and in the remaining4% by the RP service for RP The the commonly selecting systemwere the supplier. most stated criteria

(24% (16%). the enduseof model of respondents) and cost/timingconsiderations

Other criteria included surfacefinish, accuracy and strength of the model. In 68% for finish RP the the model were and material required accuracy, surface of cases decided upon by the designer. In a further 12% the decision was made jointly between the designer and the RP service supplier or model shop manager. The designer needs to be provided with a method of selecting the most suitable RP be to the selection criteria used. particular system given

71

5.4.6

CAD System

Many CAD systems were listed with several respondentsusing more than one system. However, the most commonly used systems by a sizable margin were AutoCAD and Unigraphics (both used by 32% of respondents)and Pro-Engineer (used by 28%). 80% of respondentswere CAD for design 90% their using over of Almost work. all of the respondents(92%) had accessto a CAD systemwith solid modelling capability but there was a wide range in the proportion of work undertaken using this capability (from less than 20% to 100%). Most respondents (64%) had accessto a CAD system with FBD capability and of these, all but one were making use of this capability. Only 28% of respondentswere sure that their CAD systemscould attach non-geometric information to featuresand lessthan half of these were using this capability. Therefore, the DSS can start with the

have designers FBD that that assumption most many capability. are using solids and However, the use of a CAD model to represent non-geometric data cannot be assumed.

5.4.7

Transferring Data to RP System

STL was the leading file format for transferring data from CAD to RP (used by 40% of respondents). The remaining 60% used either a neutral CAD exchange format, an actual CAD file or a 2D drawing. Over half of respondents(52%) had listed data but that transferring were problems experienced no problems with included poor STL files, incompatability of systems,lost data and over-large file data. by the These overcome some sort of reworking of were problems sizes. in be These problems must avoided every caseto avoid time-consumingrepetition.

72

Only 24% of respondentsstated that they information sent no other than geometry

to the RP servicesupplier. Other informationtransferred included2D drawings,

number of models required, material requirements,timescalerequirements,required finish, surface required accuracy, layer thickness and preferred orientation. This information was sent in a variety of ways including by telephone, email, post and modern. Designers must be encouraged to send this data in the most suitable
format and using the most reliable medium.

5.4.8

Relationship with RP Service

64% of respodentswere using a RP bureau service,20 were using facilities owned by their parent company and the remaining 16% had their own in-houseRP facility. The length of time that RP had been used varied from less than one year to over five years. Most respondents (76%) received feedback from the RP service being if but this encounteredand when the was usually only a problem was supplier had been be However 12% that they of respondentsstated ready. model would facilitate be RP. 60% how to to the of changed model could given advice as RP the that their service supplier was one of relationship with respondentsstated The DSS it described "over-the-wall" Only 20% relationship. as an partnership. RP forming in designers their service a partnership with needs to assist supplier.

5.4.9

Future use of RP

foresee RP applications of within their company new 72% of respondentscould in increase RP. Rapid the tooling to of was an usage see them expected of all and

(44% Designers of respondents). need application new quoted commonly the most
73

to be encouraged to actively look for new applications of RP. Most respondents (64%) saw no barriers to wider use of RP but those barriers which were listed included cost of RP models (quoted by 28% of respondents), lack of solid modelling capability, unacceptablelead-times, lack of managementunderstanding and poor quality of RP models (all quoted by 12% of respondentsor less). These barriers must be broken down. The use of RP could be made easier for the

designer by reducing its cost (20% of respondents), introducing a desk-top RP

(12%) system and through a number of other actions. Finally, all but one of the respondents aims to keep abreast of new RP developments, mainly through literature and conferences. This function may also be provided by the DSS by giving designers an up-to-date list of currently available RP technologies.

5.4.10 Conclusionsfrom Survey Most designersconcluded that RP had both changedthe design process and had become an essentialpart of it. However, the changesto the design process that fairly listed have Those trivial. companieswhich were were mainly used RP to design benefits. is because has have This RP their process seenmost radically alter been used to open up new possibilitiesrather than replicate existing practices. It is desirablethat RP is used in an imaginativeway to aid as many stagesof the design is One RP such usage of as a new means of models process as possible. is in 3D by nature and understandable a wide range of communication which [28]. personnel

74

In terms of the future use of RP it interesting to note that most designers was foresee both a greater usage RP of and a wider range of applications. It is logical to assumethat in future years, more designers be will making use of RP for a wider range of applications. Designers need to be made aware of what applicationsRP can be used for.

The overall conclusion to draw from the survey is that most designersare happy with their current use of RP. It could be arguedthat there should be no attempt to "fix what is not broken". However, many problemswere identified including: 1. 2.40% 3. 4. 5. 6. RP is not being used to expandthe role of physicalmodels of designerswere not satisfiedwith at least 10% of RP models RP is not being used to its full potential by many designers Most designersdo not attach non-geometricdata to CAD models by almost half of designers Data transfer problems havebeenexperienced Feedbackfrom RP servicesis very limited

All of these are attributable, at least in part, to a weak link between design and RP, in for Designers in RP the or supporting software. who procedure using either have beenusing RP for some time may have learnt how to overcome many of these for be in first be it but the them to case. preferable avoided would problems Therefore, there is a strong casefor new RP users to be given the ability to learn from the experience of existing users, i.e. to adopt best practice. Also, it should improve RP to the they the be any user continuously the way of use aim always technology. The designers' requirements which have been derived from the

75

questionnaire indicate the necessityof improving the link between design and RP
by eliminating existing problems and by increasing integration.

5.5

Aim of the DSS for RP

Drawing upon the previous sections,it is define to the overall aim of the possible DSS as follows: -

To integrate RP into the design process in order to maximise its effectiveness.

This raisesthe question as to what the terms "integrate" and "effectiveness"mean. The precise definition of these terms will be clarified within the remainder of this chapter.

5.6

Characteristics of the DSS for RP

The overall aim of the DSS was expandedinto ideal systemcharacteristics(which be can used as performancemeasures) using a matrix approach,similar to that used function deployment (see designers' Figure 5.2). The the quality process within from listed down left-hand the the questionnaireresults are requirements obtained listed DSS to them the needed meet are along matrix and characteristicsof a side of between by The top. requirementsand characteristicsare shown the correlations in crossesentered the matrix.

76

System Characteristics
v1 U ^d dJ y

o
as

v
cn QA

r-

[0

Ch
Q

NQ
W

Q.

Designers' Requirements
Must not slow design process Suitability of component for RP Identify possible uses of RP
Quality assessment of RP models Indicate RP system capabilites Promote use of RP at all stages of design

"z;; CIS 2
b

Q0

o.

"
v

p., 8

x x x

x x x x

Enable selection of most suitable RP system Encourage use of best data format
Assist partnership with RP service

x x
x

Provide up to date list of RP technologies

Figure 5.2

Correlation matrix between designers' requirements and system characteristics.

The characteristicswould provide both the starting point for system definition and be measured. a yardstick against which system performance could is described in detail below. characteristic Each

5.6.1

Make RP an Integral Part of the Design Process

An ultimate aim of some RP researchersand vendors is to provide the designer is dimensional This "three "desktop printer". manufacturing" system or a with a in the to the the systems and of available given some commercially names reflected CAD involve designer, This the them. sitting at a would companies that sell hard 3D the invoking that of model create a copy option would menu a terminal, be The 2D in today. the can requested same as a plot way much the screen, on downstream for for be as any process such then or visualisation used could model

77

analysis or tooling manufacture. This would be the ultimate in integration RP of into the design process. Indeed, severaldesignersreferred to this possibility within the survey.

However, current RP systems are from some way providing this service. One for this is that the operation of an RP machineis a skilled task, requiring an reason experiencedperson to select the optimum process parameters. Another reason is that the transfer of data from CAD to RP may require some manual intervention to check and repair exchangefiles. A third reason is that no single RP machine can offer an unlimited capability in terms of material, accuracyor speed. Therefore, the conversion of a 3D CAD model into a 3D physical model is by no means fully automated. The challengefor the DSS is to make the transition from CAD model to RP model as easyand reliable as possible. This could be referred to as providing designer. it desktop is how "virtual" RP to the this systemsince a would appear

5.6.2

Enable the Designer to Consider the use of RP at any Stage in the Design Process

The use of RP can yield benefits at many stagesin the design process. However, the function of the RP models and the amount of design information availablewill be flexible Therefore, DSS to the the cater enough must throughout process. vary for any function which the model may be used for and to make use of whatever design information is available. If a totally inadequate level of information is be information designer the to required. must prompted provide extra available,the be decided RP of using Only then can the appropriateness upon.

78

5.6.3

Avoid the use of RP in Unsuitable Circumstances

RP process time is a scarce commodity like any other and should not be used in a profligate manner. Its unnecessary be use must avoided by ensuringthat RP is the most suitable method of creating the physical model. Other model-making

techniques exist such as hand-crafting, conventional and CNC machining. The most appropriate technique in any specific casewill dependon the geometry of the component, the material requirements, the desired accuracy and so on. It is

essentialto determine if the use of RP will savetime or money, or indeed if its use is practical. A fundamental characteristicof the DSS must be that it enablesthe designer to identify when other model-making techniques would be more appropriate.

5.6.4

Ensure Right First Time RP Models

Creating RP models is a costly processfor severalreasons: RP machinesare often build depreciation RP be to costs; materials are expensive and costs must added be RP processesmust supervised often expensive e.g. photosensitiveresins; some henceadding the cost of operator'stime; if pre-processingof CAD data is required, this is a skilled task and adds considerablyto the cost of the model. Also, although hours, is for RP build the total turn typically time a matter of the actual an model is CAD from to typically to two time model one physical model weeks around [120]. Therefore, the result of an RP model which does not meet the designer's importantly, is a waste of requirements a waste of money, and perhaps more One of the key characteristicsof the DSS must be to ensure lead-time. valuable first i. fit for designer in has time, the the RP e. purpose mind. models are right that

79

Several factors must be in

place for this to happen. The most suitableRP process

be must used, all relevant information must be made availableto the RP operator and the correct build parameters must be selectedfor the RP process. Each of these factors lead directly to the following characteristicswhich the DSS must also have:-

5.6.5

Ensure the Correct Choice of RP Process

The number of RP processesis growing continuously, offering a wider range of possibilities to the designer. The DSS must provide the designer with a robust procedure whereby the optimum RP process for a particular component can be established.

5.6.6

Improve Communication betweenthe Designer and RP Operator

The DSS must ensurethat all relevant information is transferredfrom the designer to the RP operator in the most suitable format, and at the right time. This will designer's RP has the that the operator ensure a clear understanding of is data from RP the time that the pre-processing requirements model and minimised.
5.6.7 Optimise RP Build Parameters

Most of the build parameterswill be selectedby the experiencedRP operator. However, there are some, such as model orientation and layer thicknesswhich the designer may want to specify. The DSS must provide designerswith a list of build for RP for them to they and allow select process values any which each parameters The RP these the the to model. effect of choices of performance critical see as if designer. designer For to the to the be example, wishes conveyed must 80

investigate the effect of a certain build orientation on the surfacefinish of a critical feature, the influence upon build time and cost must also be evaluated. This should

allow the optimum compromise between the factors of cost, time and surfacefinish to be reached.

5.7

Conclusions

The survey of engineering designers using RP confirmed the need for a design for RP which would improve the link betweendesignand RP. The support system designers'requirements,identified through analysisof the questionnaireresults,had been translated into ideal system characteristics. The next stageof the project was to define a system which would meet all of the requirementsidentified during the survey.

81

CHAPTER SIX DEFINING THE DESIGN SUPPORT SYSTEM FOR RAPID PROTOTYPING

6.1

Objectives

The survey of designersusing RP identified the overall aim of the DSS as being to integrate RP into the design process in order to maximise its effectiveness. Furthermore, the ideal characteristicsof the DSS were derived from the designers' requirementsobtained from the questionnaire: 1 2 Make RP an integral part of the designprocess Enable the designerto knowledgeably consider the use of RP at any stage
in the design process

3 4 5
6

Avoid the use of RP in unsuitablecircumstances Ensure right first time RP models Ensure the correct choice of RP process
Improve communication between the designer and RP operator

Optimise the RP build parameters

For a DSS to display all of these characteristics,it would have to give the designer information be design information, RP-relevant this to to enable applied access all information RP RP decisions this to the to serviceto and communicate about using Only designer have high RP then the quality models. will allow them to provide it become integral be RP that should used effectively and an can confidence that design the process part of

82

The questionsthat need to be answeredin order to define the DSS are as follows: 1. What design information is required to support RP?

2.
3.

What format shouldthis informationbe in?

How will the information be used to decisions make about the use of RP?

These questionsare answeredin turn in the following sections.

6.2

Design Information

Required to Support RP

Chapter 4 identified four steps which must be followed if RP is to be used effectively. Three of these are the responsibility of the designer,i. e. deciding if RP is appropriate (Step A), selecting the optimum RP process (Step B) and transferring the necessarydata to the RP service (Step Q. The decisionsmay be
taken in consultation with the RP service. The survey of designers using RP was

(see identify information Table these three to the type able of steps of used at each 6.1). It can be seenthat some of the information is used for all three steps, some for two and some for only one of the steps. Most of the information is either part information it. it from design However, be derived the some or else can of product is not related to the design, e.g. the availability of RP resource. The design information required for RP could come from one of severalsources. Much of the information would be contained in a comprehensive design specification as describedby Pugh [121]. Additional cost and timing information would be in the design brief However, tolerance and surface finish requirementswould have to

be drawing. This detailed from not would available after until component a come designer In these the design could make of all practice, use stage. the concept drawbacks. designer have Firstly, have the to two but would this would sources 83

decide what information is relevant to RP and, secondly,the non-unified structure of the information would make it difficult to computerise the decision-making process.

Lead-time requirement uantit of models Cost of making model Complexity of model

Project riori rating

x x x x
x

X X X

Model required for communication Customer requires model Overall dimensionsof model Surfacefinish requirements Required accuracy/tolerances Required model materials Build time for model Availability of RP resource Strength requirements Advice from RP operator Project cost code Outline drawing of model Preferred build orientation Project brief Seconda processrequirements Facetting accuracyfor STL file Solid or hollow part required
_ Colour(s)

x x x x x X x x x x x x x x x x x x X x
x

X X X

Laver thickness Critical dimensions

Datums

Table 6.1.

Information required to support the use of RP.

84

6.3

Format of Information

Required to Support RP

A method of storing all the information about a product throughout the whole of its design and manufacture process is known as "product modelling" [122]. The

concept of product modelling is gaining acceptanceand forms the basis for the development of the STEP international standard[123].

For some researchers,product modelling must feature-based design make use of (FBD) [124,125]. This enablesnon-geometric information to be attached to the

product model at a sub-component level. This would clearly be of benefit in supporting RP as some of the information in Table 6.1 could then be related to individual features of the model to be built, e.g. surface finish requirement, build preferred orientation, required accuracy. This would involve the use of feature-baseddesign.

Kruth has claimed that RP eliminatesthe needfor FBD [72]. The argumentused is that the RP processdoes not needto be adaptedfor the type of feature being built. While it is certainly true that RP models can be built without the use of FBD, the be features RP information to optimise can associatedwith used non-geometric DSS Therefore, [126,127]. the the product modelling make of will use processes FBD. incorporate the of use concept and will

6.3.1

for Rapid Prototyping Design Feature-based

features inherent as part of a product There is an problem associatedwith using different functions different have to the throughout Features meanings model. 85

design and manufacturing

process. For example,a designerwill use featuresas the

building blocks for a CAD model whereasa manufacturing engineeruses features to determine which processesto use. "Features in a geometric model become ..... for. used "

context specific and highly dependenton the application that the model is

[ 128]. In other words, since features are used for different purposes,they need to contain different information. A product model tries to be comprehensivewhereas features are application specific. How can this apparentcontradiction be resolved?

A solution to this dilemma is to use multiple feature views and feature conversion [ 129]. Different versions of the samefeature are held in the product model. Each function has its own view of the product model which gives it access to the features relevant to a specific application. Features for one application are related to features for another through conversion or "mapping". In this way the product design to the the model continues meet all requirements of and manufacturing function having features. DSS Part the the to of of processwithout use generalised determine features be define RP to to to and what mean with respect must what format should be used to store RP-related feature information. It would then be in features into defined DSS. CAD features the the to a within system possible map

6.3.2

Definition of Features for RP Applications

have the same shape as the electronic CAD is RP used to create models which features be important the to Therefore, aspect of which will used an model. different There how types is they RP shape. a of of represent are number support These be to create shapes. are closely related to the main features which can used

86

representationsused in solid modelling CAD systems,i.e. boundary representation (B-rep), constructive solid geometry (CSG) halfspace and representation[129]. Most RP models are created from CAD systemswhich use B-rep solid modellers,
e.g. Pro-Engineer and Unigraphics. With B-rep, the solid object is defined by a set

of faces, edges and vertices (called topology) which map onto surface and curve geometry (see Figure 6.1). This gives the advantageof being able to use complex freeform surfaceswhile still having the ability to construct models from volumetric incorporated into a B-rep format. Also, using B-rep it is then primitives which are have for is features. This to either possible surface or volumetric particularly useful RP where some information will relate to surfaces,e.g. surfacefinish, while other information will relate to volumes, e.g. required material. Using B-rep features, it is possible to attach non-geometric information to collections of surfaceswhich form a surfacefeature or a volumetric feature.
Topology

B-Rep

Geometry

Surfaces

?rtex

Face

Figure 6.1

topology. Example of B-rep geometry and associated

87

It has been determined that the definition feature for RP applications must of a allow for either surface features or volume features. This allows total flexibility for the designer in deciding which shapes to attach information to. Examples of

features surface would be chamfers, datum faces and styling surfaces. Volume features could include bosses, surface protrusions, stiffening webs and even the whole part itself Negative volume featuresare not required sinceRP does not use material removal operations. The range of featuresthat could be used to describe component geometries is extremely large. This means that it would be almost impossibleto restrict the designerto a set of standardfeatures. Rather, freedom to create user-definedfeatures must be provided. Some commerciallyavailableCAD for this e.g. Pro-Engineer. systemsallow

Once the designer has decided upon which feature shapeswill be used in the design, the next stage is to create their geometry and attach non-geometric information to them. This information could include most of the items listed in Table 6.1. However, the designerwould also be given the flexibility to attach other types of information which will be of benefit to the application of RP processes. An example of how this would be of benefit is described by Palm and Shafiee [130]. Not all the information may be availableinitially but more can be added as its feature An design attached nonexample of a with the process proceeds. in items information known is Figure The 6.2. information of are shown geometric has be textual or a a could numeric, attribute value which each and attributes as both. combination of

88

Attribute Feature ID Feature Name Feature Type Parent ID Loc Face Dat_X Face Dat_Y Face Orientation Position Diameter Depth

Value 10002 HOLE I BLIND HOLE

10001

Description Unique identifier User-definedname Standardclassof feature

Unique identifier of parent part

Surf Fin

1 2 3 0,0,1 20,20 10 mm 20 mm 0.05 mm

Location face on parent feature X Datum face on parent feature Y Datum face on parent feature X, Y, Z axis values Local X, Y co-ordinates Hole diameter Hole depth Surfacefinish (RA)

Figure 6.2.

Example of a feature with its associatedattribute list.

6-33 J

RepresentingFeaturesfor Supporting RP

implemented, it is be formally define how RP DSS to Before the necessary can features will be represented. A formalised method for representingdata structures is the EXPRESS language, used to define the STEP international standard. 89

EXPRESS is a textual conceptual schemalanguage[ 131] and is defined within ISO 10303 [1 32]. Essentially,EXPRESS can be used to define the structure of the data which is to be used to support a STEP application. Indeed, all of the application protocols (AP's) which have been published as part of ISO 10303 have been
defined using EXPRESS. The EXPRESS data structures can be mapped into

actual data files following a set of rules also defined in STEP. Furthermore, they be can represented in a diagrammatic manner using a formal graphical notation

subset of EXPRESS called EXPRESS-G. EXPRESS can be used to define any data structure required as part of a product model and many such structures are
already available within STEP AP's.

It would be most fortuitous if an AP for RP had already been developed which

features. Unfortunately, has been developed for is RP AP made use of no nor there for features AP the an which allows use of with associated non-geometric information [133]. However, an EXPRESS specification for data structures to [134]. layered developed by Kennicott Many of the support manufacturing was from Configuration AP203 he "standard" taken entities entities used were Controlled Design (the STEP AP most widely supported by mechanical design CAD vendors) [135]. In addition, severalentities unique to layered manufacturing had to be developed, e.g. "slice_model", "layer", "scan" and "layer thickness". RP techniques. An entity These were specifically tailored to supporting laser-based for inclusion the "specification" of material specification and which allowed called information Moreover, be this finish also was used. could requirements surface in (called "shape_aspects"). features This EXPRESS individual a part attached to 90

although non-standard and non-approved, provided a useful starting from point which a complete EXPRESS definition of data structures for RP be could developed.

To create a complete feature-based data structure definition for RP it was first of all necessaryto consider the type of geometric and non-geometric information that be to needed represented. The geometric data required for RP would be stored in the form of B-rep solids. A part could be represented as a single solid or as several solids relating to different volumetric features. Groups of surfaceswithin the part be linked together and defined as surface features. could Non-geometric

information could be attachedto either type of feature. Other geometry that could be represented would include triangular facets (to allow compatibility with STL files) and contours created from slicing through the original solids and/or their facetted approximations. A further extensionwould be the capability of assigning
RP process parameters to entities within the database.

A simplified EXPRESS-G diagrammatic representation of the required data does limit EXPRESS-G in 6.3. The Figure is the not of use structure shown definition of the DSS to being a STEP-basedapplication. This is simply a useful is However, data the the advantage required. structure which way of representing become likely is it the is that this STEP of standardised that will way of using future. in data the representingengineering

91

par

S[ll]
po -t one Function STRING quantity project -priority L[3: 3] req'd_date
INTEGER dimensions L[3: 3] cost estimate time estivate

S[0:? ]
surF feature
name L-q STRING 11 surface finish

S[1?] S[I? J surface ID

vol-feat ure
narre r)aterial colour STRING yield- strength elasticity inpact_resistonce

INTEGER
S[0:?]

INTEGER S[l: l]
STL file

slice curve

ID INTEGER

S[0:?]

INTEGER

REAL

narre STRING

-r-S[4 ?] SEI:? ]
facet z_value z_volue REAL REAL L[3: 3]

Key to

Svnbols

entity

pATA TYPE Entity-attribute relationship

normals

vertices REAL ID

L[9: 9]

S[0? S[0 J ?J vector x_start, y_start, x_end,y_end REAL

S[0:? )

Set >[r)in:r')ax] LLnin, nax] List

with no. oP elements with no. of elements

INTEGER II

Figure 6.3

EXPRESS-G representation of the DSS for RP data structure.

6.4

Using Design Information

to Support the use of RP

Having determined what design information is required to support RP and having decided to use EXPRESS-G to represent this information, it is now possible to factors have important be information A how the an will used. number of consider bearing upon this matter. Firstly, the data will not all be availableat the start of the design process but will be added to as the design progresses. Secondly, the data (or subsetsof it) will be used to make severaldifferent decisionsregarding the use be by least different (the data two Thirdly, RP. the at accessed will people of is Thus RP there and several designer and the possibly others. operator) a data data-sharing be to environment where can to a added and create requirement 92

selected from. This is where STEP again proves to be a useful tool. Within the ISO 10303 standard there is a class of parts referred to as Implementation Methods. An implementation method describesthe way in which STEP data is to be sharedamongst applications. Four levels
to date [131]: -

implementation have beenidentified of

1. 2. 3. 4.

Physical file exchange Software assistedfile exchange Shareddatabase Intelligent knowledge-basedsystems

Currently, only the first three of these implementation methods are supported ISO 10303. The most advancedof these, a shared database,offers many within benefits compared to file exchange[136]. A single repository of product data can be accessedby several different applications, each one using all or part of the information. initial (e. An CAD) can generatethe original available application g. information to populate the databaseand then this can be added to by subsequent improved data "master" Multi-user and access, a single copy of applications. further benefits. security are

6.5

Components of the Design Support System

it database for DSS, been had the decision to taken use a shared Once the approach determine the the components of complete system various to what necessary was (see ideal Figure 6.4). This time, the Again, be. was used approach matrix a would for ideas in 5.2 Figure from to the were used generate matrix characteristics system leftlisted down The the system characteristics are components. system possible 93

hand side of the matrix and components of the DSS are listed along the top. The correlations between characteristicsand componentsare shown by crossesentered in the matrix.

System Components
U. a

v [
ty 'b

00 0 H

O
p/1 td

cd 'O -D

y
G '

F'_

V_

C.

E N
y

t"' V

7; > V
Q

0
-

System Characteristics
Make RP an integral part of design process Enable designer to consider use of RP at any stage in design process Avoid the use of RP in unsuitable circumstances Ensure right first time RP models
Ensure the correct choice of RP process Improve communication between designer and RP operator

Cn U

ti

C4

a N Q

a., ci:

x x x x
x x

X X X X X X
x x X

Optimise RP build parameters

Figure 6.4

Correlation matrix between system characteristics and system components.

Obviously, the shareddatabaseis the most important part of the system and has a RP design integration in the other and all the and to of play central role CAD be data design initial the Much created using would of the characterisitcs. system. This would include geometry and possibly some non-geometric

information as well. Extra design information relating to material specification, by be finish, desired the other could added etc. surface timing constraints, cost and identification of individual features could The 6.5). (see Figure software modules However, this CAD a potential problem with done the system. be within also for (the AP203 STEP application protocol used is most commonly that approach

94

mechanical design) does not currently support features and transferring the information to the databasemight require a non-standardapplication protocol e.g. AP214 - Core Data for Automotive Design Processes. However, is work progressingto incorporate featureswithin AP203 [133].

Figure 6.5.

Using a shareddatabase to integrate designand RP.

Since STEP has not yet been developed to a stage where it can directly support files. However, definitions the STEP it decided RP, was or to make use of not is to is implementation-independent database suited very and concept of a shared been has RP information design entered required to support this area. Once all the RP-related for data into the database,other applications could use this various it data the how use would description A and work would application each tasks. of following in discussed the sub-sections. and create are

6.5.1

CAD Data Input

boundary be modelling. solid representation of capable CAD The systemused must including Pro-Engineer, these CAD criteria systems meet Several commercial be CAD The CADDS. standard created using will model Catia Unigraphics, and FBD the may system selected capabilities any and/or commands modelling solid

95

have. Some CAD systems enable non-geometric information to be attached to features and this information could possibly be incorporated into an exchangefile.

More than one solid model may be required if the RP model is to consist of several
volumetric regions with different properties such as colour or material.

Other functions which could be performed CAD the on systemare the tessellation and/or sectioning of the model to create STL facets and slice contours respectively. The facets and contour segmentswould have to be associatedwith the surfaces from which they were created to be compatible with the EXPRESS definition describedin Section 6.3.2. Once the CAD model is complete,it will be exported as file an exchange which in turn will be input to the shareddatabase. This will create the geometric data in the database to which further information will be added.

6.5.2

Feature Editor

If the data which is transferred from the CAD system to the shared database is

it identify features be to and attach non-geometric purely geometric, will necessary information to them. For features, this will volumetric be relatively

have been in Different transferredas volumetric regions a part will straightforward. be in database. Attaching the this reflected will separate solid models and information to these features would simply involve accessingthe databaserecord in fields. associatedwith each solid model and enteringvalues additional

features be This information to would more problematic. would surface Attaching be identified linked is feature This in to together. and each the surfaces require

96

possible in some CAD systemswhere severalsurfacescan be "sewn" or "stitched" together to form an open shell. Such an open shell could be transferred via an exchangefile. However, with other CAD systemsthis is not possible. It would be necessaryto "tag" each surface which belongs to a surface feature and transfer a list of these surfaces to the database. Once this had been done, non-geometric information could be attached to the surface features in the for same way as volumetric features.

Therefore, the feature editing module would not only have to accessthe database but records, may also have to give the user accessto a graphics package which display in the would surfaces the CAD model. After using the feature editor, all the design information relating to RP will have beenenteredinto the shareddatabase.

6.5.3

RP Usage Advisor

The role of the RP Usage Advisor would be to perform a "first pass" analysison the suitability of using RP for a particular model in preference to other modelfrom in form its input The the the take technologies. of user module would making answers to a series of questions about the requirements and application of the
model. These would include the number of models required, the importance of

These finish the the and general of complexity model. and accuracy, good surface

it is form how likely be that the to a qualitative assessment of used answerswould being beneficial low, If RP beneficial. be the RP probability of was no would use of further analysiswould be conducted.

97

6.5.4

RP Build Time and Cost Estimators

It is possible to arrive at a build time estimatefor RP models using an algorithmic approach [137] or through the manipulation of system-specificbuild files [101, 102]. In both cases,certain information about the RP model requirements and the system capabilites is used to calculate the time neededto build the model using a particular RP technique. Build time has a large influenceupon model cost and so a logical extension of a build time estimator is to predict the likely cost of building one or more models. Much of the information required for theseestimationswould be included in the product model for the component in question. The output from this module or modules would be a build time estimatein hours and a cost estimate in local currency.

6.5.5

RP System Selector

This module would be a more detailed "second pass" analysisof the comparative built. be for RP to the model processes suitability of particular It would use

detailed information on model requirements together with the time and cost for RP from to each available the previous module arrive at a score estimates be different importance The catered would requirements of user relative system. for using a "weighting and rating" approach.

6.5.6

Adaptive STL Generator

feature for to tolerance surface the each values of required This module would use larger for This STL tessellation. the would allow determine the meshing accuracy facets tolerance a smaller where and be specified was a wide facets to used where

98

tighter tolerance was required. In this it way, would be possibleto create STL files which would meet the user's accuracy requirementsand yet have less facets than when using a constant meshingaccuracy.

6.5.7

SurfaceFinish Optimisation Module obtained on any surfacefeature

Given a minimum slice thickness,the surfacefinish

of the RP model will depend heavily upon model orientation [138]. Therefore, for different model orientations, the required surfacefinish and predicted surfacefinish for each feature could be compared. The user could then select an orientation
which gives the best overall achievement of surface finish requirements or one that

best finish in a particularly critical part of the component. The the ensures surface output from this module would be X and Y rotation angles from the original orientation of the CAD model.

6.5.8

Adaptive Slicer

This module would use the values of required tolerance and surfacefinish for each be features feature determine Different thickness. to could created slice surface different slice thicknessesyet still achievingthe required accuracyand surface with finish. The actual thicknessesof slicesused would be limited by the RP technique being used. The thickness calculations would have to take the model orientation (as determinedby the previous module) into account.

6.5.9

RP Data Output Module

input both design data data the original and new The database would contain information be Much this by of modules would application the various generated

99

useful to the RP operator to help in the selectionof processparameters. Therefore, it would be beneficial to have a module which would output the required information in a useful format, either as a databasetable, a text file or some other common format. The module would allow the user to executea range of standard queries on the databaseand then output the resultsto one or more files. Thesefiles could then be made available to the RP operator. Moreover, if the RP operator had accessto the database,specialisedqueries could be created and executed. Either way, the DSS would enablethe relevant information to be transferredfrom the design processto the RP process,resulting in better integration.

6.6

Conclusions

This chapter has defined the DSS by answering three questions: what design information is required to support RP?, what format should this information be in? how information be decisions RP? The the to the about of and will make use used laid down by has been defined the the to characteristics seven satisfy system design is in described Chapter 5. The which requires result a system specification the designerto identify all the designinformation pertinent to the application of RP, features in information to a product model and to attach this volumetric and surface key information help decisions this to make to use software modules which access its has been in detail described DSS The RP. to enable sufficient about the use of description is designed be deliberately However, the to implementation. full hardware The the or software. independentof any specific next chapter addresses
DSS. the implementation of partial

100

CHAPTER SEVEN IMPLEMENTATION AND EVALUATION OF THE DESIGN SUPPORT SYSTEM FOR RAPID PROTOTYPING

The previous chapter describedthe definition of the DSS which from was created the specification developed in Chapter 5. The central pillar of the definition was the use of a feature-basedproduct model database. The next task was to implement the DSS for RP incorporating as many as possible of the modules defined in Chapter 6. The aim of this implementationwas not to createa fully functional DSS but rather a "demonstration package" which would illustrate both how the featurebasedproduct model could be used and the benefitsit would yield.

7.1

Choice of Hardware and Software

An IBM compatible personal computer (PC) was chosenas the hardware platform for the DSS. This was becausethe author had permanent accessto a PC and becauseit would allow the system to be more portable sincePCs are abundantin both industry and academia.

The choice of software for implementing the system was made using several criteria:1. 2. 3. Ability to support product modelling department Availability within author's industry Level of use within

101

No single software package fully satisfied all the requirementsdictated by these criteria. Rather, it was necessary to use a combination of two packages,Microsoft (MS)-Access [139] and AutoCAD [140]. MS-Access is high level a relational database management system (RDBMS) facilitates definition the which of the

complex data structures required for product modelling. AutoCAD is a computer aided design (CAD) package which provides both a graphical user interface and a high level programming language for data manipulation. Both were readily availablewithin the department and are in common use in industry.

7.2

Shared Database

The DSS defined in Chapter 6 envisageda computerisedsystemwhich would be based on a shared database approach. Therefore, the most critical part of the implementation database to system create was a with the correct structure to enable data to be readily sharedby various software modules. The required data structure for the system had already been defined using EXPRESS-G as shown in Figure 6.3. This data structure was mapped into the MS-Access RDBMS with

EXPRESS-G entities being represented by tables whose fields related to the being hierarchical EXPRESS-G The relationship of an entity entities' attributes. facet) (e. being was represented an attribute of another entity's attribute g. vector field in having by table the MS-Access child a which contained a reference within its hierarchical In in identifier table. this structure was parent way, a to the unique for is in Figure 7.1. the An entity shown vector of mapping example created. in database. Some tables the the to other extra Similar mappings were used create file STL the entities vol_feature and were combined since fields were added and 102

they had a one-to-one relationship. The database tables, their fields and the
hierarchical links between them are shown in Figure 7.2.

EXPRESS-G

DIAGRAM

MS-ACCESS

TABLE

vector

Vectors Facet number Layer_z_value

facet

x_st art ystart finish x

yf finish

layer

x_start,

y_start,

x_end, y_end

READ

Figure 7.1

Mapping an EXPRESS-G entity into an MS-Access table.

Figure 7.2

Structure of the shareddatabase.

103

7.3

Other System Components

The full requirement for system componentshad already been defined as discussed in Chapter 6. However, not all of these neededto be implementedto demonstrate how the DSS would work. The componentswhich were essentialto the operation of the system were those which enableddesign information relating to individual features to be entered into the database.Also, to demonstratehow this information be could used to optimise the use of RP, at least one "application module" needed to be developed. Therefore, besides the shared database,the following system componentswere implemented.: 1. 2. 3. CAD data input Feature editor Surfacefinish optimisation

By referring to back to Figure 6.4, it can be seen that these system components, together with the shared database, cannot satisfy all of the ideal system be fully However, they to characteristics. should able satisfy one of the (i. integral design RP the of process) and partly characteristics e. make an part improve first (i. RP time three models, communication others e. ensureright satisfy between the designer and RP operator, optimise the RP build parameters). The implementation of the three chosensystemcomponentsis describedbelow.

7.3.1

CAD Data Input

individual features be information their The top-level components and about must designer. facilitate data-entry forms To by DSS this, the into the were entered feature data feature data. data, These input surface volume and part created to

104

forms are shown in Appendix C. The data input forms these via serves two purposes. Firstly, it defines the features which belong to a particular component, and secondly, it allows non-geometric information to be entered against either the whole component or any of its features. Examples of this information are the number of parts required, the part's material(s)and its required surfacefinish(es).

Once the component and its features are defined, it is then possible to enter the geometry design data. To enable design data to be enteredfrom a wide range of CAD systemsand not just those supporting FBD or even B-Rep modelling, it was decided to use STL files to enter the geometry into the shareddatabase. The STL files normally generated by CAD systemsare in a binary format rather than the human-readable ASCII format which is also available. Therefore, a software files load "loadstl" designed binary STL to through module called and was parse the numeric values for each triangle into the "Facets" table. The name of the STL file from whence it originated was recorded against each facet in the table. This language is Access Basic the programming which module was written using
incorporated into MS-Access. A program listing for this module is given in

Appendix D.

7.3.2

Feature Editor

Volumetric features within a component were handled simply by using several STL

files, one for each feature. This was representedwithin the "Volumetric_features" "STL_file_name". link from This field the the entitled provided a table using its facets. Surface features feature to of constituent more of each were volumetric

105

a problem since STL files are designed to represent solid models only. It was decided that the best strategy to deal with this problem was to provide the user with a mechanism for identifying the constituent facets of each surface feature. This neededto be done using a graphical interface facets the since making up each feature shapecould only be identified if they could be seenby the user. There are no graphical interface functions provided within MS-Access and so it was to use a different package. This is where AutoCAD cameinto play since necessary it has a programming language(AutoLISP) that enablesdirect manipulation of the
graphics entities shown in the CAD drawing.

To enable the individual facets in each surface feature to be identified, it was follows: to through three necessary go a stageprocedure as Stage 1. Stage2. Stage 3. Load all facets representinga componentinto AutoCAD. Identify and list the facets containedwithin eachsurfacefeature. Create a link in the MS-Access databasebetween each surface feature and all facetswhich belong to it. A separatesoftware module was createdto perform the tasks in each of thesethree stages.

into be loaded the facets the to Firstly, the component needed representing files do loading been by STL have It to this the AutoCAD system. would possible had done, facets if have However, this the AutoCAD. was would not directly into database. Therefore in MS-Access the link an alternative those to stored any described "load_stl" The that above so was modified module approach was used. 106

in addition to inputting facets into the MS-Access database,it also output the facet co-ordinates, together within their unique identifying numbers,to a neutral file. An example of such a neutral file is given in Appendix I. This neutral file was then ready for loading into the AutoCAD system. An AutoLISP program called "facetsin" was written to load the neutral file into AutoCAD and to display each facet as a triangle with its unique identifying number shown it the centre of the triangle. The program listing for this module is shown in Appendix E and an example of the display seenin AutoCAD is shown in Figure 7.3.

Figure 7.3

Example of the facets display seenin AutoCAD.

listed. A identified be feature in to facets and Secondly, the needed each surface "selfacet" to the enables user which was written AutoLISP called program, second (either CAD facets the the screen on required digitise the numbers of all identifying The these then unique program writes all individually or in groups). be file One is by file the ASCII must into text such user. which named an numbers 107

created for each surface feature in the component and the file namesshould reflect the features they belong to. Appendix F. The program listing for "selfacet" is shown in

Thirdly, the link in the MS-Access databasebetween each surface feature and all the facets belonging to it had to be established. This was done using another Access Basic module which reads in a facet list file and then, for facet the all listed, inserts the file name againstthe "surface_feature_name" field in the numbers "Facets" table. The module is called "assign_facets" and the program listing is shown in Appendix G. Finally, the name of the facets list file is entered into the "file name" field in the "Surface features" table. This createsthe link betweenthe individual facets and the surface feature they belong to. A diagram showing the between different feature files the the relationship exchange editing modules and in data between is to transfer them used shown Figure 7.4.

Figure 7.4

Relationshipbetween different feature editing modules. 108

7.3.3

Surface Finish Optimisation

The application chosen to illustrate the benefits of using the system was that of calculating optimum stereolithography model orientation to achieve required surface finish values. This application was selectedbecauseit relies heavily upon non-geometric design information, i. e. surface finish, and becausemuch work has been done within the author's department on the relationship between model
orientation and surface finish [138,141,142,143,144,145]. Also, the ability to

optimise part orientation to achieve required surface finish is an essentialpart of for RP tooling applications. Therefore, this applicationwill be of real using models
RP the to value multitude of users working in this area.

The definition of surface finish (also called surface roughness)used for this work illustration (Ra) is based the the arithmetic meanvalue on schematic of a was which in Figure 7.5. The through a rough centreline shown cross-section surface shown is located so that the area above the line is equal to the area below. A number of The from to the taken the surface. centreline perpendicularly measurementsare following by is for finish the then equation:given surface arithmetic meanvalue Ra=(a+b+c+.... )/n

is b, the number of measurements. and n where values absolute are etc. c, where a,

N-/

'11

4L'1

1J 111

I. tii

Figure 7.5

Schematicdiagram usedto define surfacefinish [26]. 109

A software module called "calc_surf fin" was written in Access Basic for this application. The principle behind the software module was that surface finish for the faces on a stereolithography model is highly dependent upon the normal vector of each face. The actual surface finish which can be achieved has been The

mathematically predicted and experimentally verified by Reeves [138].

relationship between normal vector angle to the horizontal and surface finish for a particular set of stereolithographybuild parameters is shown in Figure 7.6.

SLA 250 using ACES build style

60

50

40

-:

'30

r ,. i..
'J^

20

10

0=---r---r-1"

It

I"

1D

TI t0 f0

IIII11IIIII 11mmC! Ql OONN

f-r-i-I PI QYNN

tf0 f0

T 1tm

ONONON0NONONONON0NONONONONONON0N0N0N0 NN Pf P) OQNN

Angle of normal vector (measuredfrom verical)

Figure 7.6

Relationship between normal vector angle and surface finish [138].

Once the normal vector for each face is known, the surfacefinish for that face can be predicted. The method for achievingthis is describedbelow.

Assuming that the stereolithography model is being built from an STL file, the facet for for The each are readily available. normal vector vectors normal original

110

each facet will have three orthogonal components,Vx, Vy and Vz. If the model is re-orientated before being built, these componentvectors will also be re-oriented. Assuming that the model is first re-orientedby a rotation around the X axis through

an angle Ax, the effect upon the component vectors, Vx, Vy and Vz, is seen in Figure 7.7. Following the rotation, the newly positioned component vectors are once again resolved into the X, Y and Z directions, resulting in the following equations:Vx'=Vx

Vy' = (Vy * cos (Ax)) - (Vz * sin (Ax))

Vz' = (Vz * cos(Ax)) + (Vy * sin (Ax))

(note that the Vx component vector is unchangedby a rotation around the X axis)

Z axis

VZ

I xis

X axIs

Figure 7.7

Effect of rotation around X axis upon Vx, Vy and Vz.

Y by the this time is axis around a second rotation If the model now re-oriented the this Ay, vectors the new component of rotation upon effect through an angle

111

Vx', Vy', and Vz' is seen in Figure 7.8.

Once again, the newly positioned

component vectors are resolved into the X, Y and Z directions, resulting in the following equations: -

Vx" = (Vx' * cos (Ay))+ (VZ'* sin(Ay))

Vy" = Vy"
Vz" = (Vz * cos(Ay)) - (Vx * sin (Ay))

(note that this time, the Vy' component vector is unchangedby a rotation around the Y axis)

z axis

Y axis

axis

Vxl
Figure 7.8 Vz'. Vy' Vx', Y Effect of rotation around axis upon and

first from be Vz' the taken Vy' set of equations can now The values of Vx', and following the to into equations: give the set second and substituted
* * (Vy * + sin(Ay)) (((Vz sin(Ax))) + cos(Ax)) * (Vx Vx '_ cos(Ay)) * (Vz * sin(Ax)) (Vy Vy- = cos(Ax)) * (Vx * * sin(Ay)) (Vy cos(Ay)) sin(Ax))) * + (((Vz Vz" _ cos(Ax))

112

where Vx, Vy and Vz are the orthogonal componentsof the original facet normal vector and Ax and Ay are the anglesof rotation around the X and Y axes (rotation around X axis is followed by rotation around Y axis). Vx", Vy" and Vz" are the three orthogonal componentsof the re-oriented facet normal vector.

Thus, for any given re-orientation, the new normal vectors and hence surfacefinish for each facet can be obtained (using valuestaken from the graph in Figure 7.6).

The operation of the "calc_surf fin" software module is illustrated by the flow diagram in Figure 7.9. Within the MS-Access database, facets have groups of will been assignedto surface features for which the designercan input a surfacefinish requirement. The software program loops through all the possible build

(at intervals) for 5 facet, between the the orientations and, each calculates ratio achievablesurface finish with the required surface finish. For each orientation, the is for facets if the there to ratio greater than one. which are any program checks see These orientations are labelled as having a problem. The user can then select an from the remaining problem-free alternatives. orientation If none of the

the that the select one comes normally user would problem-free are orientations for To the the this end, averageratio program also calculates closestto a solution. has lowest indicates the the orientation which overall average. each orientation and The average ratio is weighted to take account of the differing surface areasof all file the to the the number of For facets. program also outputs each orientation, the finish A this the facet ratio and value of ratio. program the with the worst surface
in Appendix is H. for listing this module shown

113

Start
Declarations
List facets Set Set 'Move
Se t a ll

with

surf ace to to 0 0

finishes

X angle Y an first le

to

Facet
values

in list
to 0

calculated CURRENT finish, to

FOR Read surface Use equations Calculate Look

FACET; normal vector

and area

calculate of normal

angle up actual

new normal vector to vertical vector

finish for angle surface Calculate ratio between required and actual Record if facet has worst ratio so far Calculate runnin average of ratios

Move

to

next

facet

in list

End of facet list yes Write Keep out of results lowest for avera

No

this

orientation e ratio so far

note

Increment

Y angle

bS

Is No Y angle = 180 , Yes Increment

No

X an
Is \ X angle = I80 Yes

le

b5

/rite

out

angles

with

lowest

average

ratio

End

Figure 7.9

Flow diagram of the "caic_surf fin" software module.

114

7.4

Using the System

To give the reader a better understanding how the DSS of system works, the application of the system to a particular component design is now described. The component selected for this purpose was a simple test piece originally designedto illustrate the use of features within a CAD is in The model. component shown
Figure 7.10. This was thought to be an ideal demonstration part since it has several

different form features, each of which can be assigneddifferent surface finishes. The fact that the features have varying orientations as well as different surface finish requirements makes it difficult to estimate what the optimum build be. This component will demonstrate how the feature-based orientation would is build DSS RP the to the approach of orientation to achieve the able optimise finishes. following The required surface sectionsgive a step-by-stepguide to using the various modules within the system.

Clear

Slot

Figure 7.10

Component selectedto demonstrateusing the DSS.

115

7.4.1

Starting Procedures

Before the user starts to use the system, certain facts must be determined and noted. The first of these is the name or identifying number of the componentbeing considered. This will be used to uniquely identify the component within the MSAccess database. Secondly, the user must have an understandingof what features

there are within the component.. These may be volumetric features, surface features or a combination of both. If the is designer the user who created the component this should be no problem. However, if this is not the case,then liaison
designer be the with will necessary. This will avoid the pitfalls of an "over-the-

wall" approach to using RP and the lack of communicationthis can cause. In the caseof a component that has severalvolumetric features,e.g. for different material colours, a single STL file must be availablefor each feature. This is best achieved designer's CAD system where the component can be split into several the within file STL solid models and an created for each one. Finally, unique namesmust be identification features for in in to the the purposes the assigned all part, again database.

7.4.2

Initiating Software Packages

The user must have accessto a PC which has MS Windows, MS-Access and
AutoCAD for Windows loaded onto it. The user must enter the Windows Within AutoCAD, the

AutoCAD environment, start

MS-Access. and start

be loaded, if has DSS files the this to system AutoLISP must not run needed is by "Load This during done the been clicking achieved on start-up. already highlighting "File" facetsin. lsp the the and menu, Applications" option of

116

lsp files in the list of applications selfacet. provided and then clicking on "Load". The AutoCAD window should then be minimised. Within MS-Access, the

database containing the DSS must be opened. This is done by clicking on the "Open Database" option of the "File" menu, entering the name of the mdb file for the databaseand clicking on "OK". used MS Access then opens a window

which shows the different types of objects within the database,e.g. tables, forms, queries, modules, etc. Clicking on any one of these object types will cause all objects of this type to be listed in the window.

7.4.3

Entering Part Data

To assignattributes at the top level of the databasestructure, i.e. whole component attributes, the user must click the "Forms" object button in the database window. This is followed by a double-click on the "Part_data_input" list item. The form for data is in text then the and values each of entering part opened and user can enter the slots. Only one of the slots must be completed, i. e. "Partname", the others being optional (some slots have default values). At least one part must exist in the databasebefore any feature attributes can be assigned. The demonstrationpart is length, function, "features" the quantity, overall attributes and requires called 47.5, be "demonstration", 1,50,30 to to and set overall-width and overall-height respectively.

7.4.4

Entering FeatureData

i. database feature level the To assign attributes at the next of structure, e. in "Forms" button have the the the clicked object previously must user attributes, database window. There is then a choice of double-clicking on either 117

"Volume_feature_input"

or "Surface_feature_input". Each part must contain at

least one volumetric feature before any STL data can be entered and so for a new part, volumetric feature data would normally be entered first. When the form for entering volumetric feature data hasbeenopened,the user can enter text and values in each of the slots. Again, some of the slots must be completed, i.e.

"Volume feature" and "Part-name". If the name of the STL file which contains the geometry of the feature is alreadyknown, it can also be enteredat this stage. If the part is to have only one volumetric feature then obviously this feature must representthe whole of the part's volume and should be given a nameto indicate this fact. If more than one volumetric feature is defined, each one should have a
meaningful name perhaps related to its shape or function. The demonstration part

has only one volume feature and this will be named "features whole". The other be to attributes entered at this stage are part_name (features), STL_file_name (c:\acadwin\features. (SL has 5180). been selected This stl) and material material becauseit is one of the standardresinsused with the StereolithographyACES build style.

Although surface features are optional within the system, most of the envisaged implemented (and Entering them. the one) make use of certainly applications described for is feature to the process attributes very similar volumetric surface features. The form used is called "Surface_feature_input"and the compulsory slots "Part-name". "Surface-feature" and are As with volumetric features, the

"Partname" slots creates a hierarchical link with the part to which the feature belongs. It is at this level that attributes such as surface finish are entered. The 118

surface features within the demonstration part are "boss", "hole", "slot" and "chamfer". Each of these is entered in turn its with required surface finish. The four features are arbitrarily assignedsurfacefinishes of 25,30,35 and 40 microns respectively.

7.4.5

Reading in STL File(s)

Up to this stage, only non-geometric information about a part and its features has been entered. It may be possible that some future applications will use this information only. However, for the orientation application which has been

implemented, it is also necessaryto enter the shapeof the part. This is done by in file STL for eachvolumetric feature in the part. To do this, the user reading one database form the must open entitled "convert_STL". Two slots need to

"STL_file_name" The "output_file_name". completed, and user then clicks on the "Run Macro" button which executesthe "load stl" module described in Section 7.3.1. The result of this is that all the facets in the STL file are loaded into the databaseand also listed in a neutral file format. The user can now select another STL file and repeat the process. The names of all neutral files that are created is for file be STL The by this be to part the user. converted only noted should "c: \acadwin\features. stl" "c: \acadwin\features. neu". file the neutral and will be given the name

7.4.6

Assigning Facetsto SurfaceFeatures

by the is system,made so the needto work This the most complicated part of using in both MS-Access and AutoCAD. However, providing the user follows the

MS-Access Firstly, be difficulties the encountered. should no procedure, correct 119

window should be minimised and the AutoCAD window opened. Then, in order to visualise the facets which define the geometry of the component, the command "facetsin" should be typed. This will simply ask the user for the nameof the neutral facets file to be loaded. Upon hitting return, the facets will appear on the drawing window as triangular strings with a text number in the centre of each triangle. Execution of this command is repeatedfor each neutral file until the geometry of the entire component is visible. For the demonstration component, the resultant AutoCAD screenwill be similar to that shown in Figure 7.3.

Secondly, still within AutoCAD, the "selfacet" commandmust be typed. This will for the the name of the surface feature file to be created followed by a ask user facets to that belong to this feature. This can be done by requirement select all individual facets by clicking on or using a window. To aid this process, the user direction, in can change viewing zoom and erase unwanted geometry. The

"selfacet" command is repeatedfor every surfacefeature within the component, in this case,four. The result is a seriesof text files containing a list of facet numbers. The user must note the name of eachfacets list file. The files for the demonstration flf'. flf' flf', "slot. "chamfer. flf', "hole. be "boss. and called component will AutoCAD window can now be closed. The

Finally, the MS-Access window must be maximised and the form called "assign_facets" clicked open. Once again, two slots must be completed,

first "input-file'". The "surface_feature_name"and slot contains a pull-down menu defined during been have features the step the previously which surface showing 120

described in Section 7.4.4. Once the user has selectedthe desiredfeature, the name of the associatedfacets list file created in AutoCAD be must enteredin the second slot. The user then clicks on the "Run Macro" button which executes the

"assign_facets" module describedin Section 7.3.2. The result of this is that, within the "facets" table of the database, all the facets listed in the input file have the name of their surface feature entered into the "surface feature" field. The user can then select another surfacefeature and repeat the process. This processwill be repeated four times for the demonstration part, once for eachfeature.

7.4.7

Running the Surface Finish Calculation Procedure

The steps described above will have resulted in the geometry of the component having been entered into the database in the form of volumetric and surface features with associatednon-geometric information. The next stage is to use this design information to optimise a particular aspect of the RP process, i. e. model in orientation regard to surface finish requirements. This is done by clicking open the "calc_surf fin" form within MS-Access, completing the two slots "Part-name" "output-file" and and clicking on the "Run Macro" button. This executes the "calc_surf fin" module describedin Section 7.3.3, resulting in the creation of a file lists for different finish 1,300 the average almost orientations, surface which, ratio, the x and y rotation angles,the facet at which worst surfacefinish ratio occurs and finish the worst surface ratio together with the optimum orientation angles (in lowest file An is the averageratio). terms of achieving exampleof an output given in Appendix J. The optimum orientation angles are also displayed in the

"calc surf fin" form.

121

When the orientation optimisation module was run using the demonstration part, two orientations were found which satisfied the surface finish requirementsfor all four features. These were with an X angle rotation of 55 and aY angle rotation of
150, or an X angle rotation of 125 and first The 30. of aY angle rotation of these orientations is shown in Figure 7.11. The two orientations gave identical

finish surface values as they were actually mirror imagesof one another. This was to be expected sincethe part is symmetrical. The averagesurfacefinish ratio across the whole part was 0.408 showing that for most of the part, surfacefinish was well have is This within requirements. not an obvious solution and would probably not been selectedintuitively by an experiencedRP operator.

71

Figure 7.11

Optimum orientation of demonstration part.

has in DSS the a single system The orientation optimisation algorithm used features finish the a where i. all on to surface acceptable an achieve e. objective, There designer. by orientation which been criteria other the has are specified value for build the the into time part, the take e. g. account, could optimisation 122

requirement for support structures,the stability of the part during the build process, the number of layers required, the build height for the part, the avoidance of trapped volumes and part accuracy. Some of theseare inter-related,e.g. number of

layers and build time. Others are applicable to only some RP techniques, e.g.
requirement for support structures (some RP techniques require no additional

supports). There may also be a conflict between two or more of the criteria, e.g. building the part with a low overall height may not be compatible with good accuracy. Therefore, if the orientation optimisation algorithm was to be expanded to be multi-objective in nature, it would be necessary to incorporate some method
of achieving an acceptable compromise between the different criteria. One method

doing is be by factors this to to of give weighting eachcriteria which can altered the based for be These the to orientation would used calculate an overall score user. it has. is An how the example of the type weighting on well each criteria met and is be used of multi-objective orientation optimisation program which could describedby Cheng et al [98].

The optimisation orientation has only been applied to one RP technique, i.e. for RP be technique It where any used equally-well could stereolithography. for is This the dependent is finish case upon surface orientation. mainly surface "stair-stepping" layers, distinct based the RP on techniques which are most fact finish. The that the being between layers the greatest source of poor surface it does more makes actually structures support consider not algorithm optimisation laminated for RP ground solid manufacture and techniques object as such suitable during build for inherent the the process. model support provide which curing 123

Likewise, the fact that neither the model height nor number of layers used are considered makes the orientation algorithm more suitablefor RP techniques where build time is highly dependent on model volume and much less so on the delay time between layers. An example this is 3D is little interof there welding where very layer delay time. Therefore, the build time does not vary nearly so much with orientation as it does for stereolithography. The only reasonwhy stereolithography was chosen was the readily available data on surface finish versus surface
orientation.

7.4.8

Sensitivity Analysis

The demonstration part was now subjected to a two-way sensitivity analysis. Firstly, to see how much a change in orientation would alter the achievable surface

finish, and secondly,to seehow much a changein surfacefinish requirementwould first The the the alter results of orientation optimisation. part of the sensitivity in designer flexibility be level the there to analysiswould allow of would seewhat be from different This the might recommended ones. using orientation angles in the previous subdesirableto help cater for other orientation criteria as discussed section.

The results form the orientation optimisation were imported into a spreadsheet
finish X dimensional ratio against graphs of surface package and used to plot three

finish done for both This the Y averagesurface ratio was and orientation angles. in finish for facet for (see 7.12) Figure the for the whole part any worst surface and designer looking By 7.13). (see Figure the these plots can quickly at the part

124

.. -

Qo. 7-o. s
M 0.5-0.6 0.4-0.5 Q 0.3-0.4 Q 0.2-0.3 0.1-0.2
0-0.1 E3 0 15-\ 30 45 60
j

o.s 0.7 bo. 6 o.s

\ 105 120 135 Y angle 150
C)

0 4.0
0)
a,

75 90

F-0.4 -0.3 -0.2 0.1 0

o (D co

165 =' 180

ON O O O CO

CO X angle

oo

o cv

Figure 7.12

Result of sensitivity analysis for average surface finish ratio.

indication obtain a qualitative of what effect a change in X or Y angle will have upon the surface finish ratios. For average surface finish, changing the
Y angle has a much smaller effect than changing the X angle, as indicated

by the elongated shape of the two minimum regions shown in light blue in
Figure 7.12. For the worst as the minimum surface finish ratio, the picture is more

complicated

in regions shown orange-brown

in Figure 7.13

"kidney-shaped". are actually

However, it would seem that for this variable,

Y X is to the than the angle angle. there more sensitivity

125

1.8-2

1 \
0, 10
20 30 40 50 60 70 80 90 100 ` 110- + 120 130 140 150 160 -k 170 180-0 N
O00gC: )
O OO

1.6-1.8 Q 1.4-1.6 1.2-1.4 01-1.2 0.8-1 Q 0.6-0.8 Q 0.4-0.6 0.2-0.4 00-0.2

2 1.8 1.6 1.4 F1.2 1 0.8 0.6 0.4 0.2 0

rq CO

0 Co

X angle

Figure 7.13

Result of sensitivity analysis for worst surface finish ratio.

The second part of the sensitivity analysis allows the designer to see what finish (or the tightening) of specified surface requirements effect a relaxation be if initial have This the the orientation results. would useful upon would finish be for that the could not achieved required surface a results showed be determined if feature. for It the could relaxing requirement critical part finish feature As this the problem. an example, would rectify surface another

126

requirement for the boss feature was varied up to +/- 6 microns from its current value of 25 microns, whilst leaving the requirements for the other three features unchanged. The optimum orientation (actually one of the two optimum

orientations) was identified in each case by searching through the output file produced. The results obtained form this sensitivity analysisare shown in Table 7.1.

Boss surf. fin. re 't

19

21

23

25

27

29

31

X angle rotation Y angle rotation Av. surf. fin. ratio

Worst facet ratio

45 145 407 .

1.07

45 145 397 .
1.07

45 145 388 .
1.07

0.96

55 150 408 .

0.93

55 150 400 . hole yes

0.96

60 145 375 .

0.93

60 145 370 . hole yes

Worst facet location Solution found?

hole no

hole no

hole no

boss yes

boss yes

Table 7.1

Results of sensitivity analysison surfacefinish requirements.

It can be seenthat a decreasein the value of boss surfacefinish requirement made a in The difference to the optimum orientation. an part was rotated slightly small finish This the ratio worst surface caused attempt to meet the tighter requirement. for by determined (this facet boss facet hole be searching was to rather than a on a it feature to). in "facets" facet table to assigned the was what see the worst number be met. However, even with this slight rotation, the tighter requirement could not began but due first improved then finish this to rotation The averagesurface ratio at for boss. When the the in line tightening the specified requirement with to worsen boss for the finish the part was relaxed, progressively was requirement surface
finish improve in the the to in of surface attempt the an manner opposite rotated

As 30 demanding had the secondmost microns. requirementof hole feature which

127

a result of this, and due to the relaxed requirementfor the boss,the averagesurface finish ratio improved.

The two-way sensitivity analysisshowed that the relationship between orientation, average surface finish ratio and worst surface finish ratio is very complex and therefore difficult to predict without the aid of a computer program. When it is not possible to achieve the surface finish requirementsfor all the features in a part, there is a trade-off to be reachedbetweenachievingthe best overall averagesurface finish for the whole part or the best possible surfacefinish ratio for any particular feature. This trade-off could perhapsbe automatedby assigningdifferent priorities to different featureswhich would be incorporated into the surfacefinish calculation.

7.4.9

Finishing Procedure

Once the user has finished using the system,MS-Access and AutoCAD can both be exited by clicking on the "Exit" option of their "File" menus.

7.5

Evaluation

System the of

to evaluate Once the DSS for RP had been implementedand used, it was necessary in Chapter 5. listed ideal its performance compared to the system characteristics implemented been have in 7.3, will As stated section the systemcomponentswhich Therefore, four the evaluation was the towards characteristics. of only contribute is four. The the each characteristic against these evaluation to of result restricted below. presented

128

7.5.1

Make RP an Integral Part of the Design Process

The current method most often used for creating an RP model from a CAD model is for the STL file to be created by the designer and sent to the RP operator. The STL file may sometimesbe accompanied with other written or verbal instructions. This is an "over-the-wall" approach leads being RP "bolt-on to and used as a option" rather than an integrated designtool. The implementationof the DSS for
RP allows the designer to take the STL file it design and combine with other information to create a product model. Any information is which relevant to the be RP use of can added to this model at a component or feature level. The

designer is encouraged to think about the design requirementsfor the RP model is and able to record these in a structured manner which reflects the way the
has been designed. component In this way, the RP model requirements are

by be RP The the then can accessed specified more clearly. product model information RP to the to process. optimise extract and use all relevant operator Therefore, the designer can have greater confidencethat the RP model will meet the specified requirements. This means of achieving the transition from CAD integral RP designer help RP to the part of as an use persuade model will model to the design process.

7.5.2

Ensure Right First Time RP Models

DSS the the The comprehensivenature of ensuresthat product model used within be RP information the to can made of support use needed the non-geometric all in do fields database if Even tables RP the the not the current to operator. available be The fields easily. requirement added very can additional requirements, all cover

129

for geometric data to be provided is currently met using triangular facets. This is not ideal but it is as accurate as using STL files. The availability this of complete product model allows RP systemselection,data accessby the RP operator and the correct choice of build parametersto be achieved. This ensuresthat right first time RP models become the normal expectationfor the designer.

7.5.3

Improve Communication between the Designer Operator RP and

The information held in the product model within the DSS is not enough, on its
own, to improve communication between the designer and RP operator. What is

is also needed the meanswhereby the RP operator can access the data input by the designer. This could be achieved in two ways. Firstly, tailor-made output files be by designer Secondly, RP the to the the RP could generated and sent operator. operator could accessthe DSS directly and interrogate the database. The current implementation of the DSS allows for either of these and does actually produce an file to help the RP operator decideupon optimum build orientation. output

7.5.4

Optimise the RP Build Parameters

There are many build parameters which need to be optimised for the various be Algorithms RP could written systemswhich are currently available. commercial for any of these which would accessthe data in the DSS product model and use has been Only one such algorithm this to calculate optimum parameter values. finish build i. to the meet surface orientation created, e. the optimisation of different for The for RP the the of requirements consideration model. requirements different features within the component is an essentialpart of this algorithm. It

130

demonstratesthat the DSS can help to optimise RP build parametersin a way that non-feature-basedsoftware tools cannot.

7.6

Conclusions has been created from the

A design support system for rapid prototyping

specification derived in Chapter 6. It is not fully functional in that it does not contain all the specified modules but it does demonstrate how a feature-based product model can be used to support the designer's use of RP. It allows both geometry and non-geometric design information to be combined in the same database. It can store this data at both a component and feature level, with a
hierarchical link between the two. The database structure has been designed to

future allow expansion,e.g. the use of surfaceand curve geometry. A number of have been software modules created which enabledata to be entered, manipulated and used to support an orientation optimisation procedure. The use of all these been demonstrated has This through the an component. modules example use of feature-based how the approach can solve a problem which would showed for i. be difficult, the optimised orientation to achieve e. requirement otherwise very different surface finish values on different parts of the component. A sensitivity demonstrated the orientation optimisation was also performed which analysis of how the system provides designers with essentialfeedback on the effect of their decisionsupon the prototyping process. Finally, the DSS was evaluatedagainstthe in identified 5. ideal chapter systemcharacteristics relevant

131

CHAPTER EIGHT FURTHER WORK

This chapter contains recommendations how the work undertaken during this on research project can and should be continued. Indeed, some further work has already been undertaken by studentsworking under the supervisionof the author. Where this has happened,referencehasbeen madeto the students' work.

8.1

Enabling the Direct Transfer of Non-geometric Design Information from CAD Systems

The insertion into the databasesystemof designdata related to whole components

features is and currently achieved using input forms. This is quite time consuming

and susceptible to human error. If the designer was using a CAD system that supported FBD and had the capability of attaching non-geometric information to the features, then some of this manual data entry could possibly be avoided. The design information contained within the CAD system product model could be directly transferred to the databasesystemby meansof an exchangefile. However, format for is in Chapter discussed 6, there currently no standard representing as been by Chrisp Geldart has data. Nevertheless, and conducted some work such AutoCAD implemented has FBD together with the ability to create a within which feature-based data exchange file [126]. The software, which was developed in AutoLisp, enablesthe designer to create severaltypes of form feature and attach CAD is When information the these. the to complete, model non-geometric designer can ask for an exchangefile to be created. An exampleof such a "feature

132

attribute list" file is shown in Figure 8.1. The feature-basedexchange file and AutoCAD solid model would provide the design information required for rapid prototyping. This work needsto continue so

downstream processessuch as

that a direct link between the DSS this CAD system is proved feasible and beneficial.

Feature ID: POS PROFILE-1 Volume Type: POS NAME: Vol-feat-1 PROFILE: PLINE_pos_profile_1 POSITION: 120.00,110.00,0.00 DIRECTION: 0.00,0.00,1.00 ROTATION: 0.00 HEIGHT: 30.00 HEIGHT TOLERANCE: 0.1 Feature 111 POS BOSS 2 Volume Type: POS NAME: Vol_feat_2 POSITION: 145.00,205.00,30.00 DIRECTION: 0.00,0.00,1.00 RADIUS: 15.00 HEIGHT: 30.00 DIAMETER_TOLERANCE: 0.1 OLERANCE : 0.25 HEIGHT _T POSI TI ONAL_T OLERANCE : 0.1 MATERIAL: Blue 3 Feature ID: NEGHOLE Volume Type: NEG NAME: Surf feat I POSITION: 200.00,145.00,0.00 DIRECTION: 0.00)0.00, -1.00 RADIUS: 10.00 DEPTH: 30.00 DIAlVETER_T OLERANCE : 0.1 POSITIONAL TOLERANCE: 0.1 SURFACE_FINISH: 50 FUNCTION: location EOF Figure 8.1 Example of "feature attribute list" file createdin AutoCAD. 133

8.2

Increasing the Range of Data in the System systemat present is restricted to triangular facets

The geometric data stored in the

which are derived from STL files. However, if the system could support B-rep geometry and topology, then more accuratedata could be transferred directly from the CAD model. This would require either the mathematicaldefinition of surfaces and curves to be stored within the system or a high-level link into an actual CAD modelling kernel such as ACIS [146]. The individual surfacepatchescould then be grouped together to form volumetric and surface features. The structure of the databasehas been created with such an expansionin mind (see figure 7.2). The range of non-geometric data could also be extended. Attributes could be attached to geometry at various levels within the databasestructure; layer, curve, surface, feature or part. This would enablea much wider range of designinformation to be included within the product model.

At present, the only information relating to RP process capability is the list of for be finish be This could obtained stereolithography. surface values which can include RP the to systems once their all other commercially available extended larger On has been finish a scale, the range of capability measured. surface include be for RP information to accuracy, widened system could each capability be designer The then able to specify would material properties, running costs, etc. the model's requirements for these attributes. A process which endeavouredto designer the the model's requirementswith systemcapabilitieswould enable match has Some RP informed to already system use. work choice of which to make an been undertaken in this direction by Bernie [ 147] 134

8.3

Increasing the Number of Application Modules

Other application modules could be written to make use of the design information contained within the database. One exampleof this would be an RP usageadvisor. An undergraduate student, working under the author's direction, has developeda simple RP usage advisor which has been incorporated into the latest version of the DSS [148]. This advisor asks the user to answer several pertinent questions

regarding the use of RP (seeAppendix K). If the user answers"yes" to all of these, the use of RP is recommended. If any negative answersare given, the user is first of all advisedto consider the wider implications and then, if the answer is still "no", the use of RP is not recommended. This is obviously an over-simplification of the RP usageadvisor which is actually required but it illustratesthe principle involved.

Another application module which could be developedis an RP build-time and cost design The the requirementsand the geometry program would examine estimator. build-time be built to these estimates produce comparative and use of the model to for severalRP systems. The figures produced would not have to be very precise, i. e. within a few percent of the correct time. They would only need to be A be different to cost the compared. to processes enable sufficiently accurate build-time by be the running estimates with combining obtained estimate would building for to costings at comparative arrive overheads and costs material costs, be have Again to different these extremely would not systems. the model using be 20% the acceptable. would cost correct of within accurate,

135

A third application area would be a rapid prototyping system selector. Indeed, a simple RP selection module has been incorporated into the latest version of the DSS [149]. This module comparesthe design requirementsspecified for the RP model (accuracy, strength, surface finish, machinability, timescale, budget and maximum dimensions)to the capabilitiesof five RP techniques. It usesthe relative performance of the RP techniques and the user-defined importance of each requirement to calculate a percentagescore for each technique. The user can then investigate the composition of each score to see where a technique's performance is strong or weak. Once again, the RP system selector is not as sophisticatedas it could be but it does illustrate how a product model approachcan be used.

Two further examples of application modules which could be implemented are an adaptive STL generator and an adaptive slicer. These would work best with a Brep model of the component as described in section 6.2.1. In both cases,the different accuraciesrequired for different featuresin the component would be used to determine the accuracy of the RP model to be built. With the adaptive STL be done by this generator would varying the chordal accuracy of the triangular facets according to the accuracy required for the feature being meshed. The be between layers built RP to to the the used spacing adaptive slicer would vary features being the to the sliced. maximum required accuracyof all model according This assumes that the RP processto be used can use variable slice thickness. The STL or slice files created by these modules would be smaller than normal but layer The RP ensure an accuracy. of model of adequate use variable would still build RP the time. thickness would also speedup 136

The aim of creating an extended system would be to assist designersin all their decisions about the use of RP. It could be used each time a physical model was needed and would thus help to further integratethe considerationof using RP into the design process.

8.4

Enhancing

the User Interface of the System

The current user interface is very simplistic and involves quite a lot of memorisation and/or note-taking on the part of the user. The correct order of tasks must be followed and file namesmust be recorded. Two packagesare used simultaneously integration between them is far from "seamless". These weaknesses the and could be overcome by using a specially developed user interface. Such an interface lead in involved the through the tasks would user using the system and various help be be On-line tasks available at every stage automated. some would would be for The displayed file selection. on-screen possible user and names would interface would also act as an "umbrella" for MS-ACCESS and AutoCAD so that the system would appear as an integratedunit to the user. The systemwould then be at a stage where it could be releasedfor evaluationby a much larger number of it be functionality. They its to able use would people, to solicit their opinions on be doing One to this the would of the way author. supervisionof without requiring Web, Wide World or as code the as executable either the on available system make [149], by developed Jones DSS, latest The the interactive version of site. an incorporates a Visual Basic [150] user interface which supports many of these requirements.

137

8.5

Creating a Commercially Available Version of the System

Once the improvements described above have been implemented,the system will be at a stage where it can be shown to potential partners who could then begin to develop a commercial version. This would not necessarilyuse the samesoftware packages as the research system but it would use the same structure and methodology. One possibility might be to integrate the software with a turnkey CAD system. The product model databasecould then contain direct links to the CAD geometric database, hence avoiding the duplication of data. The

disadvantage of this would be to limit accessto the system to those designers be An to CAD route would that alternative package. particular making use of data transfer the with severalpopular of capability system with create a stand-alone CAD packages. The direction of future developmentwill dependvery much upon be what partnershipscan established.

138

CHAPTER NINE

CONCLUSIONS
In this chapter, the achievements of the project are assessedin regard to the objectives stated in Chapter 1. For each objective, conclusions are drawn as to how it has been met. Also, the original contribution to knowledge made by the researchis stated. Finally, to act as an overall conclusion, a brief progress review and validation assessment of the whole project is made.

9.1

To Determine What Links are Required Between the Engineering Design Process and Rapid Prototyping

In Chapters 2 and 3, engineering design and rapid prototyping technologies were described in detail. The iterative nature of the design process, with the need for frequent evaluations, provides an ideal opportunity for the use of physical prototype models. As the design process moves towards production, so the function, and hence requirements, of these models will change. The development RP technologies and materialsmeansthat many of the models of a wide range of be made more quickly using previously made using alternative methods can now RP. Also, the speedof creating models using RP has openedup new areaswhere, had been However, too or slow expensive. some previously, model-making RP to and the advancement of computer modelling not suited models are is Therefore, to there a sometimes offers a cheaper alternative physical models. it bring benefits in RP the to over manner, only when will an effective use need availablealternatives.

139

Chapter4 showsthat, to ensure the effectiveuseof RP, a series of stepsneedto be

executed to provide the optimum transition from design to RP model. It must be determined if RP is appropriate. If so, the most suitable RP system must be selected. All the information required to support the RP processmust be available in the correct format. The optimum combination of RP process parametersmust be established. These four steps must be consideredeach time the is RP use of being considered. Each one acts as a link between the design processand RP and each one must be optimised. The links between design and RP having been identified, Chapter 4 went on to describewhat software tools have been developed to support these links. A failing of nearly all thesetools is their concentrationupon
initial data The RP. this to part of the culmination of support using only geometric

formulation the of the requirement statement at the end of research project was Chapter 4.

9.2

To Design a Computerised System to Support the Designer's Use of Rapid Prototyping

Chapter 5 began with a justification of the needfor a designmethodology for using had formalised but by been had This RP. methodology no recognised others need the decided It been the of applying that way effective most was created. ever develop for RP. To design system the through support a of use was methodology both for were used. approaches positive this and negative system, a specification into RP by taking consideration not On the negative side, the problems caused had The identified. that these to design ensure system during the process were designers On the of requirements and the opinions side, positive were avoided.

140

using RP were solicited through a questionnairewhich received a high response rate of over 45%. The DSS had to meet the commonly agreed requirements of these designers. Using both these positive and negative requirements,the specific objectives of the DSS were drawn up and used as the starting point for the definition of the computerisedsystem,describedin Chapter 6.

The central pillar of the DSS was to provide the designerwith the ability to access, process and communicate all RP-relevant design information. A feature-based product modelling approachwas used to enableall information, both geometric and be by to the system. The product model was defined by non-geometric, used considering the designer'suse of RP-related information. Firstly, the information identified. format RP Secondly, to the required support required of this was information was specified. Thirdly, the processesrequired to use this information for supporting RP were specified in the form of an implementation-independent computer systemarchitecture.

9.3

To Implement the Design Support Benefits its Use will Yield

System and Demonstrate

the

Once the design support systemhad beendefined, the next stagewas to develop an implementation. Chapter 7 describes how a "demonstration system" was Several, but not all of the

implemented using MS-ACCESS and AutoCAD.

However, 6 in Chapter described the ones which were written. modules software into be information design the feature-based to entered were written enabled This the prototyping process. to rapid of aspect one optimise and used system

141

particular application used feature-based surfacefinish requirementsto calculatethe optimum build orientation to achievebest overall surface finish on an RP model. The ability to assign different surface finish requirementsto different parts of the is model useful both to the designerand the RP operator. The designer for can ask good surface finishes on functional surfaces and leave other surfaces with an unspecified value. This is a typical requirementfor engineeringcomponents. The RP operator can now use a build orientation which will take account of varying surfacefinish requirementson different parts of the model. This avoids the creation of models with unnecessarilysmooth surfaceswhere they are not required. This single application of the feature-based product modelling approach to RP demonstratesthat it does yield tangible benefits. Other potential applicationshave been describedin Chapter 8.

9.4

To Identify the Future Research and Development which is Required to Transform the System into a Commercial Package

The fulfilment of this objective is describedin Chapter 8. The areas of required further work have been identified and described. It is clear that a significant is in improved development the needed way of system amount of research and functionality and user interface. However, the basic structure of the system is in is improved implementation for information an required place and essential its in Chapter 6. This time and work will require many man-months of provided For is beyond the a single part-time researcher. capability of prompt completion be it is hoped team that able to continue work of personnelwill this reason, a small on the system.

142

9.5

Original

Contribution

to Knowledge

The use of a feature-based product model to optimise RP is an idea that originated within this research project. This is testified to by several publications during 1995/96 [126,127,149] which precedeany other publishedpaperspromoting this approach. Indeed the only relevant article which hasbeen found prior to this work was one which argued againstthe use of FBD for RP [72].

The originality of this work is seen through the demonstration that attaching non-

geometric design information to geometric features within a product model can bring tangible benefits to the RP process. This has been achieved by creating a computerised systembasedon a feature-based product model. Such a systemis far from unique but it is the first system of this type to cater for design information distinctively tailored to the support of RP, e.g. triangular facets and multi-material components. The availability of a fully-functional version of this systemwill enable the consideration of RP to be fully integratedinto the designprocess.

9.6

Review of Progress and Validation of Work

This research project has been successful in that it has identified a problem, investigated the weaknessesof current solutions and provided a novel solution to the problem. It has met or provided the potential to meet all of the research has has knowledge been It to made. objectives and an original contribution foundation in journal for future has a resulted several and researchand provided listed in Appendix L. publications are which conference

143

It should be noted that the project has in four year period and some run over a areashas been overtaken by recent developments. Therefore, if the project were to be undertaken in present circumstances,some things would be done differently. Most notably, the ability to embed non-geometric information in feature-based CAD models is now available on a PC hardware platform, e.g. within AutoCAD Release 14. This was not the case in 1995. Hence, some of the programming link AutoCAD and MS-Access to create a product model to which was undertaken be be longer this accommodatedwithin a standard necessarysince could would no CAD system. This does not detract from the work that was done but rather serves by followed been has direction a this now to validate the which was taken since CAD supplier. major

144

References 1 Dixon J.R., Simmons M. K. and Cohrn P. An Architecture for Application of Artificial Intelligence to Design. Proc. of 21st Design Automation Conference, IEEE, 1984. pp 634-640. Brown D. C. and Chandrasekaran B. Expert Systemsfor Class a of Mechanical Design Activity. in Knowledge Engineering in Computer Aided Design, Gero J.S. (Ed), Elsevier Science Publishers, 1985. pp 259-290.
Sriram D., Stephanopoulos G., Logcher R., Gossard D., Grouleau N., Serrano D. and Navichandra D. Knowledge-based System Applications in Engineering Design: Research at MIT Al Magazine, Fall, 1989. pp 79-95.

Liu C.R. and Trappey J.C. A Structured Design Methodology and Metadesigner: A System Shell Concept for Computer Aided Creative design. Advances in Design Automation, Vol 1- Computer Aided and Computational Design, ASME, 1989. pp 309-313. Duhovnik J. Systematic Design in Intelligent CAD Systems. Intelligent CAD Systems I: Theoretical and Methodological Aspects, ten Hagen P.J.W. and Tomiyama T. (Eds), Springer-Verlag, 1987. pp224-238. Waldron M. B. Design Processes and Intelligent Computer Aided Design (ICAD). in Intelligent CAD III, Yoshikawa H., Arbab F. and Tomiyama T. (Eds), Elsevier SciencePublishers, 1991. pp 51-75.
Yang J. and Rozenblit J.W. Case Studies of Design Methodologies: A Survey. Proc. of AT, Simulation and Planning in High Autonomy Systems, IEEE, 1990. pp 136-141.

Smithers T. AI-based Design versus Geometry-based Design or Why Design Cannot be Supported by Geometry Alone. Computer-Aided Design, Vol 21, No 3, April 1989. pp 141-150. Colton J.S. and Dascanio J.L. An Integrated, Intelligent Design Environment. Proc. of International Computers in Engineering Conference, ASMIE, 1990. pp 9-15. Maher M. L., Sriram D. and Fenves S.J. Tools and Techniques for Knowledge Based Expert Systemsfor Engineering Design. Advanced Engineering Software, Vol 6, No 4,1984. pp 178-188. Ohsuga S. Introducing Knowledge Processing Technology to CADCAM. Finite Elements in Analysis and Design 1, Elsevier Science Publishers, 1985. pp 255-269. 145

10

11

12

Kinoglu F. and Riley D. Artificial Intelligence: Expert System Model of the Design Process. Design News, 3rd March, 1986. pp 149-156.

Brown D. C. and Chandrasekaran B. Knowledge and Control for a Mechanical Design Expert System. Computer, July, 1986. pp 92-100. Groover M. P. and Zimmers E. W. CADCAM: Computer Aided Design and Manufacturing. Prentice/Hall International, 1984. Zeid I. CAD/CAM, Theory and Practice. McGraw-Hill, 1991. Sutherland I. E. Sketchpad: a Man-machine Graphical Communication System. Proc. of Spring Joint Computer Conference, AFIPS, 1963. pp 329-346. Keast J.E. Solid Modelling. Integrated Graduate Development Scheme CAD Module Course Notes, University of warwick, 1987. Dixon J.R., Cunningham J.J. and Simmons M. K. Research in Designing with Features. in Intelligent CAD I, Yoshikawa H. and Warman E. A. (Eds), North-Holland, 1989. pp 137-148. Patel R. M. and McLeod A. J. Engineering Feature Description in Mechanical Engineering Design. Computer Aided Engineering Journal, Vol 5, No 5,1988. pp 180-183. Dixon J.R., Libardi E. C., Luby S.C., Vagul M. and Simmons M. S Expert Systems for Mechanical Design: Examples of Symbolic Representation of Design Geometries. Engineering with Computers, Vol 2,1987. pp 29-46. CADDS S. Computervision, Argent Court, Sir William Lyons Road, Coventry, UK. Parametric Technology, Pro/ENGINEER. Bracknell Beeches, Bracknell, UK. Technology House,

14

15 16

17

18

19

20

21

22

23

Krause F., Ulbrich A. and Vosgerun F.H. Feature Based Approach for Systems. in Process Planning Product Design Integration the and of Modelling for Computer Aided Design and Manufacturing, Turner J. et Publishers, 1991. 285-298. Science (Eds), Elsevier pp al Tomiyama T. and Yoshikawa H. Knowledge Engineering and CAD. Proc. of International Symposium on Design and Synthesis, Japan Society of Precision Engineers, 1984. pp 1-6.

24

146

25

Tomljanovich M. Computer and Designer Integration. in Computer Applications in Production and Engineering, Kimura K. and Rolstadas A. (Eds), Elsevier Science Publishers, 1989. pp 143-151. Kalpakjian S. Manufacturing Engineering Technology (Second and Edition). Addison-Wesley, 1992. Wohlers T. State of the Industry Proc. of Rapid Prototyping and Manufacturing Conference `93, Dearborn, Society of Manufacturing Engineers, 1993. JacobsP. Stereolithography and other RP&M Technologies.ASME Press, 1996.
Miller J.F. Rapid Prototyping Fact Sheets. EDS Unigraphics, 800 Tower Drive, Troy, MI 48098, USA. Chua C. K. and Leong K. F. Rapid Prototyping: Principles Applications in Manufacturing. John Wiley & Sons, 1997. and

26

27

28

29

30

31

Sachs E., Curodeau A., Gossard D., Jee H., Cima M. and Caldarise S. Surface Texture by 3D Printing. Proc. of Solid Freeform Fabrication Symposium,University of Texas at Austin, USA, 1994. pp 56-64. Cima M., Sachs E., Cima L. G., Yoo J., Khanuja S., Borland S.W., Wu B. and Giordano R. A. Computer-derived microstructures by 3D Printing: Bio and Structural Materials. Proc. of Solid Freeform Fabrication Symposium, University of Texas at Austin, USA, 1994. pp 181-190. Dickens P., Pridham M., Cobb R., Gibson I. and Dixon, G. 3D Welding. Proc. of Ist European Conference on Rapid Prototyping, University of Nottingham, UK, 1992. pp 81-93.
Merz R., Prinz F. B., Ramaswami K., Terk M. and Weiss L. E. Shape Proc. of Solid Freeform Fabrication Manufacturing. Deposition Symposium, University of Texas at Austin, USA, 1994. pp 1-8.

32

33

34

35

Konig W., Celiker T. and Herfurth H-J. Approaches to Prototyping of Metallic Parts. Proc. of 2nd Euopean Conference on Rapid Prototyping Nottingham, UK, 1993. 303-316. University Manufacturing, of pp and
Luck L., Bauman F., Keller B. and Wiedemann B. Material Research Techniques IKP. for Prototyping Rapid Proc. Development the at and Rapid Prototyping Manufacturing, Conference Euopean 3rd on and of University of Nottingham, UK, 1994. pp 309-326.

36

147

37

Kochan D. Selection and Evaluation SFM Systems. Proc. of of Towards World Class Manufacturing, International Federation of Information Processing,1994. Gibson I., Brown D., Cobb S. and Eastgate R. Virtual Reality and rapid Prototyping: Conflicting or Complementary? Proc. of Solid Freeform Fabrication Symposium,University Texas Austin, USA, 1993 113of at pp 120. Hull C., Feygin M., Baron Y., Sanders R., Sachs S., Lightman A. and Wohlers T. Rapid Prototyping: Current Technology Future and Potential. Rapid Prototyping Journal, Vol 1, Nol, 1995. 11-19. pp Cheshire D. G., Harrison D. K. and Wormald P.W. Novel Techniques for Tableware Development. Proc. of 3rd European Conference on Rapid Prototyping and Manufacturing, University of Nottingham, UK, 1994. pp 181-190.
Allin K. J. Experimental Stress Analysis Rapid Prototype of Components. Proc of 2nd European Conference on Rapid Prototyping and Manufacturing, University of Nottingham, UK, 1993. pp 119-129.

38

39

40

41

42

Cowie J. Now Boeing's Digital Gamble Must Fly ProfessionalEngineering,Vol 7, No 10,1994. pp 10-12.

For

Real.

43

Arthur A., Dickens P.M. and Cobb R. C. Using Rapid Prototyping to Produce EDMElectrodes. Rapid Prototyping Journal, Vol 2, No 1,1996. pp 4-12. Stucker B., Bradley W., Eubank P.T., Norasetthekul S. and Bozkurt B. Zirconium Diboride/Copper EDM Electrodes from Selective Lase Sintering. Proc. of Solid Freeform Fabrication Symposium, University of Texas at Austin, USA, 1997. pp 257-264.
Coole T. J., Cheshire D. G. and Newman D. J. The Application of Rapid Prototyping to the Tableware Industry. Proc. of 4th European Conference on Rapid Prototyping and Manufacturing, University of Nottingham, UK, 1995. pp 41-53. Hirose M. and Sato S-I The Design of Virtual Worlds. Proc. of Design for Manufacturability Conference, American Society of Mechanical Engineers, 1994. pp 13-18.

44

45

46

47

Dani T. H., Fathallah M. and Gadh R. COVIRDS: An Architecture for a Virtual Design System. Proc. of Design for Conceptual Manufacturability Conference,American Society of Mechanical Engineers, 1994. pp 19-26. 148

48

Smyth S. Personal Communication GeneralMotors, Midsize Car Division Headquarters,30001 Van Dyke 2-10, Warren, MI 48090-9020, USA. Gee R.W. Case Studies in Rapid Prototyping Sunstrand Electrical at Power Systems. SME TechnicalPaper (Series)PE, 1993, pp 1-23. Kochan D. Solid Freeform Manufacturing Elsevier, Amsterdam, 1993. Wohlers T. State of the Industry Reports 1993-1997. Wohlers Web World, http://lamar.colostate. edu/-wohlers/, 1997.
Marcus H. L., Harrison S. and Crocker J. Solid Freeform Fabrication: an Overview. American Society of Manufacturing Engineers, Manufacturing Engineering Division, MED, 1996 Vol 4. pp3-9.

49

50 51

52

53

Smith-Moritz G. (Ed) Evaluating Rapid Prototyping Systems. Rapid Prototyping Report, March 1993. pp 3-6. Burns M. Automated Fabrication Prentice Hall, Englewood Cliffs, NJ, 1993. Waterman N. Rapid Protoding in the USA. Report on the Overseas Science and Technology Expert Mission to the USA, Institution of Mechanical Engineers/Dept.of Trade and Industry, London, 1993.
Schmidt L. D. Rapid Prototyping Technology: Benchmarking Comparison. Proc. of 4th International Conference on Rapid Prototyping, University of Dayton, 1993. pp 301-308.

54

55

56

57

Gargiulo E.P. Stereolithography ProcessAccuracy User Experience. Proc. Conference 1st European of on Rapid Prototyping, University of Nottingham, UK, 1992. pp 187-201
Lart G. Comparison of Rapid Prototyping Systems. Proc. of 1st European Conference on Rapid Prototyping, University of Nottingham, UK, 1992. pp 243-254. Juster N. P. and Childs T. H. C. A Comparison of Rapid prototyping Processes. Proc. of 3rd European Conference on Rapid Prototyping and Manufacturing, University of Nottingham, UK, 1994. pp 3 5-52. Iuliano L., Ippolito R. and De Filippi A. A New User Part for Performances Evaluation of Rapid Prototyping Processes. Proc. of 3rd European Conference on Rapid Prototyping and Manufacturing, University of Nottingham, UK, 1994. pp 327-339.

58 -

59

60

149

61

Jayaram D., Bagchi A., Jara-Almonte C.C. and O'Reilly S. Benchmarking of Rapid Prototyping Systems Beginning Set to Standards. Proc. of Solid Freeform Fabrication Symposium,University of Texas at Austin, USA, 1994. pp 146- 153.
Aubin, R. F. A World Wide Assessment Rapid Prototyping Technologies. of Proc. of Solid Freeform Fabrication Symposium, University of Texas at Austin, USA, 1994. pp 118-145. Expert Briefing. Optimat Ltd., Nasmyth Building, Scottish Enterprise Technology Park, East Kilbride, Glasgow, UK, 1994

62

63

64

McMahon C. and Browne J. CADCAM From Principles to Practice. Addison-Wesley, Wokingham, UK, 1993.
STL File Format Specification. 3D Systems Corporation, 26081 Avenue Had, Valencia, CA 91355, USA.

65

66

Jamieson R. and Hacker H. Direct Slicing of CAD Models for Rapid Prototyping. Rapid Prototyping Journal, Vol 1, No 2,1995. pp 4-12. Cubital Facet List (CFL) File Format Specification. Cubital Ltd., 13 Hasadna Street, P.O.B. 2375, Industrial Zone North Raanana, 43650 Israel. Wang Y., Dong J. and Marcus H. L. The Use of VRML to Integrate Design Solid Solid Fabrication Freeform Fabrication. Freeform Proc. of and Symposium,University of Texas at Austin, USA, 1997. pp 669-676.
Common Layer Interface (CLI), version 1.31. Brite-Euram Prototyping Techniques, Project No. BE5278,1994. Rapid

67

68

69

70

SLC File Format Specification. 3D SystemsCorporation, 26081 Avenue Hall, Valencia, CA 91355, USA. Chalasani K. L. and Bagchi A. Process Planning Issues in Freeform Fabrication. American Society of Mechanical Engineers, Production Engineering Division (Publication) PED, Vol 56,1992. pp 81-92. Kruth J.P. Materials Ingress Manufacture by Rapid Prototyping TechnologiesAnnals of CIRP, Vol 40, No 2,1991. pp 603-614.
Muller H., Joppe M. and Sheng X. Concept for an integrated Rapid Prototyping Process Planning System. Proc. of 4th European Conference UK, Nottingham, University Manufacturing, Prototyping Rapid of and on 1995. pp 81-94.

71

72

73

150

74

Phillipson D. K. and Henderson M. R. Rapid Prototyping Machine Selection Programme. Proc. 6th European Conference on Rapid of Prototyping and Manufacturing, University of Nottingham, UK, 1997. pp 291-303.

75

Narayanan V., Chua C.K. and Ang B. A Selection Expert Systemfor Solid Freeform Manufacturing. Proc. 2nd International Conference of on Computer Integrated Manufacturing, Singapore, 1995. pp 1416-1423. Smith-Moritz G. (Ed) Interfacing CAD and Rapid Prototyping. Rapid Prototyping Report, January 1992. pp 4-6. Martin A. D. ADMesh version 0.95. Home Page of Anthony D Martin, http://heart.engr.csulb.edu/-amartin/, 1997.
Bohn J.H. and Wozny M. J. Automatic CAD-model Repair: Shell-Closure. Proc. of Solid Freeform Fabrication Symposium, University of Texas at Austin, USA, 1992. pp 86-94.

76

77

78

79

Smith-Moritz G. (Ed) Imageware Introduces Rapid Prototyping Module (RPM). Rapid Prototyping Report, November 1994. pp 6-7. StlTools. Igor G. Tebelev, PO Box 4491, Palos Verdes, CA 90274, USA. Fadel G., Crane D., Dooley L. and Geist R. A Link Between Virtual and Physical Prototyping. Proc. of Rapid Prototyping and Manufacturing Conference `95, Dearborn, Society of Manufacturing Engineers, 1995. Alciatore D. G. and Wohlers T. T Importing and Reshaping Digitized Data for Use in Rapid Prototyping. Rapid Prototyping Journal, Vol 2, No 1,1996. pp 13-23. #10, Unit Crocker, Avenue Concepts Inc., 28231 Solid 3.0. Bridgeworks Valencia, CA 91355, USA
MAGICS Belgium. Heverlee, B-3001 60, Kapeldreef V., N. Materialise 3.01.

80 81

82

83

84

85

for Structure Design Support Computer-aided Webb D. and Gerdes V. Rapid Conference International 5th Proc. Models. on Stereolithography of 221-228. USA, 1994. Dayton, University pp Prototyping, of

86

Suite 242, Drive, Eastmark Inc., 7607 International POGO STZ Manager. College Station, Texas 77845, USA.

151

87

Kirschuran C.F., Jara-Almonte C.C., Bagchi A., Dooley R.L. Ogale and A. A. The Clemson Intelligent Design Environment for Stereolithography. Proc. of 2nd International Conference Rapid on prototyping, University of Dayton, USA 1991. pp 240-245. Kirschman J.S., Kirschman C.F., Fadel G.M. and Greenstein J.S. The Clemson Intelligent Design Environment for Stereolithography CIDES 2.0. Proc. of Solid Freeform Fabrication Symposium,University of Texas at Austin, USA, 1997. pp 569-576.
Dolenc A. and Makela I. Rapid Tools: A Workbench for Data Transfer Data Preparation for Rapid Prototyping. Proc. of 4th European and Conference on Rapid Prototyping and Manufacturing, University of Nottingham, UK, 1995. pp 95-103. Suh Y. and Wozny M. J. Adaptive Slicing of Solid Freeform Fabrication Processes. Proc. of Solid Freeform Fabrication Symposium, University of Texas at Austin, USA, 1994. pp 404-411

88

89

90

91

Dolenc A. and Makela I. Slicing Procedure for Layered Manufacturing Techniques.Computer-aidedDesign, Vol 26, No 2,1994. pp 119-126.
Kulkarni P. and Dutta D. Adaptive Slicing for Parameterizable Surfaces for Layered Manufactuirng. Proc. of ASME Design Automation Conference, American Society of Mechanical Engineers, New York, USA, 1995. pp 211-217.

92

93

Sabourin E., Houser S.A. and Bohn J.H. Adaptive Slicing Using Stepwise Uniform Refinement. Rapid Prototyping Journal, Vol 2, No 4, 1996. pp 20-26.
Rapid Interfacing Contour in J. Pauwels B. Swaelens W., Vancraen and Prototyping - Tools that Make it Work. Proc. of 3rd European Conference UK, Nottingham, University Manufacturing, Prototyping Rapid of and on 1994. pp 25-33.

94

95

Using Orientation Part Computation On D. the Allen S. and Dutta of Support Structures in Layered Manufacturing. Proc. of Solid Freeform USA, 1994. Austin, Texas University Symposium, pp Fabrication at of 259-269. Prototyping for Rapid Build Direction Preferred G. of Frank D. and Fadel Prototyping, Rapid Conference International 5th Proc. on Processes. of University of Dayton, USA, 1994. pp 191-200.

96

152

97

Kim J.Y., Lee K. and Park J.C. Determination Optimal Orientation in of Stere'olithographic Rapid Prototyping. Technical Report, Dept. of . Mechanical Design and Production Engineering, Seoul National University, Korea, 1994.

98

Cheng W., Fuh J.Y. H., Nee A. Y. C., Wong Y. S. Loh H. T. Miyazawa and T. Multi-objective Optimization Part-building Orientation in of Stereolithography. Rapid Prototyping Journal, Vol 1, No 4,1995. pp 1223.
McClurkin J. and Rosen D. Build Style Decision Support for Stereolithography. Proc. of Solid Freeform Fabrication Symposium, University of Texas at Austin, USA, 1997. pp 627-634.

99

100

Marsan A. L., Allen S., Kulkarni P. and Dutta D. An Integarted Software Systemfor Process Planning for Layered Manufacturing. Proc. of Solid Freeform Fabrication Symposium, University of Texas at Austin, USA, 1997. pp 661-668.
Yu G-B. and Noble D. The Development of a Laser Build-time Calculation Program using Stereolithographic Apparatus (SLA). Proc. of 3rd European Conference on Rapid Prototyping and Manufacturing, University of Nottingham, UK, 1994. pp 353 67. -3 Chen C. C. and Sullivan P.A. Predicting Build-time and the Resultant Cure Depth of the 3D Stereolithography Process. Rapid Prototyping Journal, Vol 2, No 4,1996. pp 27-40. Brown S. Simulation of Solid Freeform Fabrication. Proc. of Solid Freeform Fabrication Symposium, University of Texas at Austin, USA, 1993. pp 143-149.

101

102

103

104

Bugeda G., Cervera M., Lombera G. and Onate E. Numerical Analysis of Stereolithography Processes using the Finite Element Method Rapid Prototyping Journal, Vol 1, No 2,1995. pp 13-23. Klocke F., Celi I., Noken S. and Wirtz H. Stereolithographyand Selective Laser Sintering - Process Technologyand Simulation for Optimized Part Building. Proc. of 4th European Conference on Rapid Prototyping and Manufacturing, University of Nottingham, UK, 1995. pp 169-183.
Flach L., Klosterman D. A. and Chartoff R. P. A Thermal Model for Laminated Object Manufacturing (LOM). Proc. of Solid Freeform Fabrication Symposium, University of Texas at Austin, USA, 1997. pp 677-688.

105

106

153

107

Yardimci M. A., Guceri S.I. and Danforth S.C. Thermal Analysis of Fused Deposition. Proc. of Solid Freeform Fabrication Symposium, University of Texas at Austin, USA, 1997. 689-698. pp Carleberg P. Product Model Driven Direct Manufacturing. Proc. Solid of Freeform Fabrication Symposium, University of Texas at Austin, USA, 1994. pp 270-276. ISO 10303 STandard for the Exchange Product data (STEP) of Application Protocol 204 Mechanical Design Boundary using Representation. International Organization for Standardization, 1, de rue Varembe, Casepostale 56, CH-1211, Geneve20, Switzerland.
Steger W., Geiger M. and Haller T. Data Models and Information technology for the Production of Prototypes. Proc. of 4th International Conference on Rapid Prototyping, University of Dayton, 1993. pp 333341.

108

109

110

111

Smith-Moritz G. (Ed) Why is CAD so Bad? Rapid Prototyping Report, December, 1992. pp 4-7. Smith-Moritz G. (Ed) How Rapid Prototyping is Changing Product Development. Rapid Prototyping Report, December 1991. pp 5-7.
Ulerich P.L. Rapid prototyping: Cornerstone of the New Design Process in Computers in Engineering Vol 1, Gabriele G.A. (Ed), American Society Engineers, Mechanical 1992. of pp 275-281. Jara-Almonte C. C. Bagchi A., Ogale A. A. and Dooley R. L. Design Environment for Rapid Prototyping. American Society of Mechanical Engineers, Design Engineering Division (Publication) DE, Vol 29,1990. pp 31-36.

112

113

114

115

Crawford R. Computer Aspectsof Solid Freeform Fabrication: Geometry, Process Control and Design. Proc. of Solid Freeform Fabrication Symposium,University of Texasat Austin, USA, 1993.pp 102-112.
CAD Slicing Models for Rapid Direct H. Hacker Jamieson R. and of Prototyping. Rapid Prototyping Journal, Vol 1, No 2,1995. pp 4-12.

116

117

Smith-Moritz G. (Ed) Automating Microfabrication. Rapid Prototyping Report, January 1994. pp 2-4. Transcript of Interview with Graham Hately. Bespak Plc, Kings Lynn, Norfolk, UK. Plastic Rexam Packaging Ltd., Upcraft. Steve Interview Transcript of with UK. Derbyshire, Heanor, Estate, Industrial Gate Heanor 154

118

119

120

Coole S. Personal Communication. Staffordshire University, School of Engineering, Beaconside,Stafford, UK. Pugh S. Total Design. Addison-Wesley,Wokingham, UK, 1991. Kjellberg T. and Schmekel H. Product Modelling Informationand Integrated" Engineering Systems. Annals of CIRP, Vol 41, No 1,1992. pp 201-204. Ghodus P. and Vandorpe D. Standardization of Product Models using STEP. Proc. of Symposium on Emerging Technologies and Factory Automation, IEEE, 1995. pp 145-153.
Wingard L., Carleberg P. and Kjellberg T. Enabling Use of Engineering Terminology in Product Models and User Interfaces of CAD/CAM Systems. Annals of CIRP, Vol 41, No 1,1992. pp205-208. Gu P. and Chan K. Product Modelling Design, Vol 27, No 3,1995. pp 163-179. Using STEP. Computer-Aided

121 122

123

124

125

126

Campbell R. I., Chrisp A. G. and Geldart M. Using Features to Integrate Design and Rapid Prototyping. Proc. of 13th International Conference Production Research, Freund Publishing House, London, 1995. on Addendum pp3-6.

127

Campbell R.I. Using Feature-based Design to Optimise Rapid Prototyping. Journal of EngineeringDesign, Vol 7, No 1,1996. pp 95-103. Gindy N. N. Z. A Hierarchical Structurefor Form Features. International Journal of Production ResearchVol 27, No 12,1989. pp 2089-2103.
Bronsvoort W. F. and Jansen F.W. Feature Modelling and Conversion Key Concepts to Concurrent Engineering. Computers in Industry, Vol 21, 1993. pp 61-86. Palm G. and Shafiee A. M. Product Model Driven laser Sintering at IVFKTH. Proc. of 5th European Conference on Rapid Prototyping and Manufacturing, University of Nottingham, 1996. pp 1-13 .

128

129

130

131

Owen J. STEP An Introduction. Information GeometersLtd, Winchester, UK, 1993. Exchange Product data (STEP) Part 11 STandardfor the ISO 10303 of Manual. Reference International Organization Language EXPRESS The for Standardization, 1, rue de Varembe, Case postale 56, CH-1211, Geneve20, Switzerland.

132

155

133

Pratt M. Personal Communication. National Institute of Standards and Technology (NIST), Manufacturing SystemsIntegration Division, Building 220, Gaithersburg,MD 20899-0001,USA. Kennicott P.R. An Application Reference Model for Layered Manufacturing. Proc. of Intelligent Manufacturing Systems International Conference on Rapid Product Development, Fraunhofer Institute for Manufacturing Engineering and Automation, Stuttgart, Germany, 1994. pp 451-473. STEP Vendor Translator Announced Releases. PDES, http://www. scra. html, 1997. org/pdesinc/vendor. Inc.,

134

135

136

Doty R. The Final STEP. in STEP The Future of Product Data Exchange, Automotive Industry Action Group, Southfield, MI, USA, 1995. pp 16-21.

137

Osorio A. Estimating Building Times for the STEREOS 300. EARP (European Action on Rapid Prototyping) Newsletter No 7,1995. pp 1416. Reeves P.E. and Cobb R. C. Reducing the Surface Deviation of Stereolithography Using In process Techniques. Rapid Prototyping Journal, Vol 3 No 1,1997. pp 20-31.
Microsoft Access 2.00. Microsoft Redmond, WA 98052-6399, USA. Corporation, One Microsoft Way,

138

139

140

AutoCAD Release 12. Autodesk Inc., 111 McInnis Parkway, San Rafael, California 94903, USA. Cobb R. C., Spencer J.D. and Dickens P.M. Better Surface Finishing Techniques for RPT is a Must. Proc. of 2nd Scandinavian Rapid Institute, Denmark, Danish 1993. Conference, technological prototyping
Reeves P.E. and Cobb R. C. The Finishing of Stereolithography Models Using Resin based Coatings. Proc. of Solid Freeform Fabrication Symposium, University of Texas at Austin, USA, 1995. pp 96-107. Reeves P.E. and Cobb R. C. Improvements in the Surface Finish of Stereolithography Models for Manufacturing Applications. Proc. of 1st Tooling Prototyping Research, Rapid Conference National and on Mechanical Engineering Publications, London, 1995. pp 139-149.

141

142

143

144

Surface Cause, Finish Effect C. SL Cobb R. the E. P. and Reeves and Handsfree Solution. Proc. of Future of Model Making International Conference,University of Hertfordshire, UK, 1996.

156

145

Reeves P.E. and Cobb R. C. Surface Deviation Modelling LMT of Processes Comparative Analysis. Proc. of 5th European Conference on -a Rapid Prototyping and Manufacturing, University Nottingham, 1996. pp of 59-75.

146

ACIS 3D Toolkit. Spacial Technology Inc., 2425 55th Street, Building A, Boulder, CO 80301-5704, USA. Campbell R.I. and Bernie M. R.N. Creating a Database of of Rapid Prototyping Capabilities. Journal of Materials ProcessingTechnology, Vol 61, Nos 1-2,1996. pp 163-167.
Jones K. G. The Production of a Rapid Rrototyping Decision Support System. Third Year Project Dissertation, Dept. of Manufacturing Engineering and Operations Management, University of Nottingham, UK, 1997.

147

148

149

Jones K. G. and Campbell R.I. Rapid Prototyping Decision Support System. Proc. of Solid Freeform Fabrication Symposium, University of Texas at Austin, USA, 1997. pp 551-558.
Visual Basic 4.0. Microsoft Corporation, One Microsoft Way, Redmond, WA 98052-6399, USA.

150

157

APPENDIX

QUESTIONNAIRE

USED FOR SURVEY OF DESIGNERS USING RP

QUESTIONNAIRE ON DESIGNERS' USE OF RP UNIVERSITY OF NOTTINGHAti1

1. General information

What is the name of your company? How many employees work for the company'?
How many employees work in product or tool design? What products are manufactured by your company?

What is your position within your company?

2.

Use of RP
How did you First hear about RP?

Why did you start to make use of RP models?

How do you currently decide when to use an RP model?

Under what circumstanceswould you definitely not consider using RP?

How many RP models have you had made in the last year? What did you use these RP models for?

Would you still have had these models made if RP had not been available? Yes/No It yes, how would these models have been made?

Yes/No techniques? model-making Do you still make use of other

" If yes, what percentage of your models are made using RP?

A- i

What criteria do you use when evaluating the quality of an RP model?

What proportion of the RP models that you have had built have been of satisfactory quality?

What were the reasons for any models not being satisfactory?

What has been the most important benefit(s) of using RP?

What (if any) problems have you encountered when using RP?

3.

Secondary Processing of RP models

Have any of your RP models been used for secondary processes? Yes/No
If yes, which processese. g. vacuum casting, investment casting, spray metal tooling'?

Have you used RP models to produce prototype tooling? Yes/No

if yes, what secondary process(es) did you use to produce the tooling?

If no, have you ever considered doing this? Yes/No

4.
-

Effect of RP upon Design Process

At what stages of the design process have you made use of RP?

Have you used RP to evaluate alternative designs? Yes/No Have you used feedback obtained from RP models to modify designs? Yes/No Has the use of RP changed your design process in any way? Yes/No

-If

yes, how?

Would you consider RP to be an essential part of your design process? Yes/No

S.
-

Choice of RP system
Which RP process(es) do you have access to (either within your company or through a service bureau)? Stereolithography Selective laser sintering Laminated object manufacture Fused deposition modelling Solid ground curing

Other (please specify)

Who decides which RP process to use?

How is the choice of RP process made?

for be RP Finish and decides surface material will needed an Who what accuracy, model?

A-3

6.
-

CAD System
What CAD system(s) do you se? What proportion of your design work is done on CAD?

Does your CAD system have a solid modelling capability? Yes/No

If yes, what proportion of your CAD modelling? is done using solid work

Does your CAD system have a Ieature-based design capability? -

Yes/No

If yes, clo you make use of feature-based design? Yes/No

Can non-geometric information e.g. surface finish, be attached to features in your CAD system? Yes/No
If yes, have you ever used this facility? Yes/No

7. -

Transferring

data to RP system

How do you transfer part geometry from CAD to RP?

What problems have you encountered when transferring this information?

How have you overcome these problems?

Besides part geometry, what other design information do you provide for the RP operator?

information? do this transfer to additional What medium you USC

A-4

8"
-

Relationship with RP service

Where do you have your RP models produced?

How long have you used this RP service?

Do you receive any feedback from the RP service when the model is supplied or while it is being built? Yes/No If yes, what sort of feedback?

Would you describe the relationship between yourself and the RP service to be one of partnership or an "over-the-wall" relationship?

9.
-

Future use of RP
Do you foresee any new applications of RP in your company? Yes/No

It yes, what applications?

Do you expect the use of RP to grow within your company? Yes/No Are there any barriers to the wider use of RP within your company? Yes/No If yes, what barriers?

How could the use of RP be made easier for the designer?

in developments RP keep technology? intend of abreast to How do you

THANK

YOU VERY MUCH FOR YOUR TIME

A-5

APPENDIX

RESULTS OF SURVEY OF DESIGNERS USING RP

1.

General information

How many employees work for the company?

1-10

11-50
51-100 101-500 2 7

501-1000
1001-5000

4
5

5001+
Total 25

How many employees work in product or tool design? 1-5 6-10

11-20

4
3

21-50 51-100 101-200 201-500

501+

7 2 1 3
2

Total

25

What products are manufactured by your company? Aerospace Automotive Consultancy Electrical Medical Others Total 6 3 2 6 2 6 25

What is your position within your company? Managing Director Technical Director/Chief Engineer Senior/Design Engineer CAD/CAM Manager Development Engineer 1 5 16 2 1

Total

25

B-1

2"

Use of RP

How did you first hear about RP? Magazine Colleagues 10 4

CAD Vendor Television Other Don't know Total

2 2 4 3 25

Why did you start to make use of RP models? Reduce lead-time Reduce cost Improve confidence in design Visualisation Convenience for complex parts Other 12 2 2 2 3 4

Total

25

How do you currently decide when to use an RP model?

Timing constraint Cost constraint Part complexity Standard practice Customer request Other 9 9 7
J

11
42 (Some respondents gave more than one answer)

Total

definitely RP? do not consider Under what circumstances you using Simple part Not cost effective Minor modification to design Sheet metal parts Never Others Total 6 3 2 2 2 10 25

B-2

How many RP models have you made in the last year?

1-5 7

6-10
11-20 21-50 51-100

5
3 4 1

101+
Don't know

3
2

Total

25

What did you use these RP models for?

Form and fit analysis 10

Customer studies Design approval Process definition

Visualisation

7 5 3
3

Casting patterns
Vacuum casting

3
4

Other Total

8 43 (Some respondents gave more than one answer)

Would you still have had these models made if RP had not been available?
Yes No Sometimes 16 6 3

Total

25

been have how If yes, made? would these models

Conventional Manual6 10

CNC Total

3 19

techniques? model-making Do you still make use of other Yes No Total 17 8 25

B-3

If yes, what percentage of your models are made using RP?

1-10% 8

11-20% 21-3 0%
)1-40% 41-50% 51-60% 61-70%

0 1
0 1 1 0

71-80%
81-90% 91-100%

3
2 1

Total

17

What criteria do you use when evaluating the quality of an RP model?

Accuracy Functionality 16 6

Surface finish 13 Appearance

Strength Distortion

5
2 2

Stability Other Total

2 3 49 (Some respondents gave more than one answer)

What proportion of the RP models that you have had built have been of satisfactory quality? 1-70% 71-80% 81-90% 91-100% "Most" Total 0 4 5 15 1 25

What were the reasons for any model not being satisfactory?
Poor accuracy Poor hand finishing Distortion Poor surface finish 5 5 3 2

Incorrect file
Other

2
5

Total

22

(Not all respondentsgave an answer)

B-4

What has been the most important benefit(s) of using RP?

Increased speed Reduced cost Design verification Better communication Visualisation Other Total 19 5 4 2 2 6 38 (Some respondents gave more than one answer)

What (if any) problems have you encountered when using RP? Durability of models Lack of accuracy Poor surface finish High lead-time Other Total 4 3 2 2 5 16 (Not all respondents gave an answer)

B-5

3.

Secondary Processing of RP Models

Have any of your RP models been used for secondary processes? Yes No 25 0

If yes, which processes?

Vacuum casting Investment casting Spray metal tooling Direct tooling Total 18 1I 3 2 34 (Sonne respondents gave more than one answer)

Have you used RP models to produce prototype tooling?

Yes No 8 17

Total

25

If yes, what secondary process(es) did you use to produce the tooling Sand casting Direct tooling Spray metal tooling Other Total 3 3 2 2 10 (Sonnerespondents gave more than one answer)

If no, have you ever considered doing this?

Yes No Not yet 5 10 2

Total

17

B-6

4.

Effect of RP upon the Design Process

At what stages in the design process have you made use of RP? Several Concept After layout Development Before tooling Other 8 5 2 2 4

Total

24

(Not all respondentsgave an answer)

Have you ever used RP to evaluate alternative designs?

Yes No 13 1

Total

25

Have you used feedback obtained from RP models to modify designs?

Yes No 24 1

Total

25

Has the use of RP changed your design process in any way? Yes No 15 10

Total 25 If yes, how?

More 3D CAD work done Other Total 5 8 13 (Not all respondents gave an answer)

design be RP process to Would you consider an essential part of your Yes 17

No Sometimes Total 25

6 2

B-7

5.

Choice of RP system

Which RP process(es) do you have accessto?

Stereo lithoraphy Selective laser sintering Laminated object manufacture Fused deposition modeling Solid around curing 25 8 8 4

Other Total

1 49 (Sonnerespondents gave more than one answer)

Who decides which RP process to use?

Designer Company No choice Project manager RP bureau CAD/CAM manager 9 6 5 2 2 1

Total

25

How is the choice of RP process made?

Don't know 12 6 4 3 2 5 32 (Some respondents gave more than one answer)

End use of model Cost and time Company decision Accuracy Other Total

finish be for decides and material will needed an Who what accuracy, surface RP model?
Designer RP bureau Both Other 17 2 1 5

Total

25

B-8

6.

CAD System

What CAD system(s) do you use? AutoCAD Unigraphics

ProEngineer CADDS

8 8
7 4

Catia

4 2

SDRC
Alias

Other
Total

5
41 (Some respondentscravemore than one answer)

What proportion of your work in done on CAD? 1-70% 0

71-80% 81-90%
91-100% Don't know

3 1
20 I

Total

25

Does your CAD system have a solid modelling capability? Yes No Total 23 2 25

If yes, what proportion of work is done using solid modelling?

0% 1

1-10%

11-20% 21-30% 31-40% 41-50%

51-60% 61-70% 71-80% 81-90% 91-100% Don't know

1 2 1 0
1 4 0 2 9 2

Total

23

B-9

Does you CAD system have feature-based design a capability?

Yes No Don't know 16 6

Total

25

If yes, do you make use of feature-based design? Yes No Sometimes Total 15 0 1 16

Can non-geometric information e.g. surface finish, be attached to features in CAD system? your Yes No Don't know 7 8 10

Total

25

If yes, have you ever used this facility?

Yes No 3 4

Total

B-10

7.

Transferring

data to the RP system

How do you transfer part

STL IGES DXF CAD file 2D drawing 10 4 2 2 1

geometry

from CAD to RP?

Total

19

(Not all respondents gave an answer)

What problems have you encountered when transferring this information?

None Poor STL files Lost data Incompatibility-') Other

Total

26

(One respondent gave two answers)

How have you overcome these problems?

Tried again Rebuilt lost data Mesh repair software 4 2 2

Total

(Not all respondents gave an answer)

Besides geometry, what other design information to you provide for the RP operator? 2D Drawings
Number required Material

7
4 4

Accuracy Surface finish Tirnescale Layer thickness Orientation

Function

3 3 3 2 2
2

Other None Total

3 6 39 (Some respondentsgave more than one answer)

B-11

What medium do you use to transfer this additional information?

Verbally Email Courier Modem Drawing Fax Total 5 3 3 3 1 17 (Not all respondents gave an answer)

B-12

g"

Relationship

with RP service

Where do you have your RP models produced? Bureau Parent company In house 6 5 4

Total

25

How long have you used this service?

0-1 years 2-3 years 4-5 years 6+ years 8 10 6 1.

Total

25

Do you receive any feedback from the RP service when the model is supplied it is being built? or while

Yes No Total

19 6 25

If yes, what sort of feedback? Problems Delivery estimate l0 3 3

How model should be changed Other Total

5 21 (Some respondents gave more than one answer)

be RP between to service yourself and the Would you describe the relationship "over-the-wall" relationship? an or one of partnership Partnership Over-the-wall Neither Both 15 5 4 1

Total

25

B-13

9.

Future use of RP

Do you foresee any new applications of RP in your company? Yes 18

No Total

7 25

If yes, what applications? Rapid tooling Metal parts Other Total 11 3 6 20 (Some respondents gave more than one answer)

Are there any barriers to the wider use of RP within your company? Yes No Total 9 16 25

If yes, what barriers? Cost of models Lack of solid modelling Other

Total

7 3 8
18 (Some respondents wave more than one answer)

for designer? be RP the How could the use of made easier Reduce cost 5

Desktop RP machine 3 11 Other Total 19 (Not all respondentshave an answer)

B-14

How do you intend to keep abreast of developments in RP technology? Literature Conferences

RP companies Company expert

17 6
3 3

RP association
Total

3
32 (Some respondents have more than one answer)

B-15

APPENDIX

DATABASE

INPUT FORMS

Pert Data Entry

Form

C-1

Volume

Feature Data Entry Form

C-2

Surface Feature Data Entry Form

C-.

APPENDIX

PROGRAM

LISTING

FOR "LOAD_STL"

MODULE

Access Basic Program

Listing for "load

stl" Module

Option Compare Database 'Use database for order stringy comparisons 'Declarations
Dim Dim Dirn Dim Dim Dim Dim Dim Dim Dim Dim Dim Dim Dim Dim MyDB As Database Mytable As Recordset no of facets As Long normal x As Single normaly As Single normal z As Single first x As Single second x As Single third x As Single first As Single _y second_y As Single tllird_y As Single first_z As Single second_z As Single third_z As Single

'Main function
Function load_stl () 'Work with current database Set MyDB = DBEngine. Workspaces(0). Databases(0)

'Work with facets table Set Mytable = MyDB. 0penRecordset("Facets", DB_OPEN_TABLE) 'Open STL file
Open Forms! [convert_stl] ! [STL_file_name] For Binary As 1 'Open output file [Output_file_naine] ! [convert_stl] Forms! Open For Output As 2

facets 'Read and write number of facets # 1,81, Get no_of Write #2, no-of facets Function Exit Then facets If no of =0

D-1

'Read in normal and vertices values for each facet

For I=0

To (no_of_facets 1) Step I Get # 1, (85 + (I * 50)), normal x

Get Get Get Get Get Get Get # 1, (85 #1, (85 # 1, (85 #1, (85 # 1, (85 # 1, (85 #1, (85 + + + + + + + (I (I (I (I (I (I (I * * * * * * * 50) + 4), norrnal_y 50) + 8), normal z _ 50) + 12), first _x 50) + 16), first_y 50) + 20), first _z 50)+24), second _x 50)+28), second_y

Get #1, (85 + (I * 50) + 32), second_z Get #1, (85 + (I * 50) + 36), third _x Get #1, (85 + (I * 50) + 40), third_y
Get #1, (85 + (I * 50) + 44), third -z

Routine to calculate area of facet

first_y) ^ 2) + length_a = Sgr(((third_x first-x) ((third_y ^ 2) + ((third_z - first z) ^ 2))

length_b = Sgr(((first_x - second_x) A 2) + ((first_y - second_y) ^ 2) + ((first_z - second Z) /, 22 )) length_c = Sgr(((second_x - third_x) A 2) + ((secondy - thirdy) ^ 2) + ((second_z - third_z) A 2))
values = .5* (length_a + length_b + length_c)

facet area = Sgr(value_s * (value_s - len(-Yth_a)* (value_s - length b) (value_s - length_c))

Write values into Facets table in database Mytable. AddNew ' Create new record.

Mytable("Vertex = third _y Mytable(" Vertex = third_z _3_z") Mytable("Area") = facet area Write #2, M.ytable("Facet_number")

[STL_file_name] [convert_stl]! Forms! Mytable("STL_file_name") = Mytable("Normal_x") = normal_x Mytable("Normal_y") = normal_y Mytable("Normal_z") = normal_z first_x Mytable("Vertex = _x") _1 first_y Mytable("Vei tex_ 1 = _y") first_z Mytable("Vertex = _1_z") Mytable("Vert ex_2_x") = second_x Mytable("Vertex = second_y _2_y") Mytable(" Vertex = second_z _2_z") Mytable(" Vertex = third _x _3_x") 3_y")

D-2

Mytable. Update

Save changes-.

Write values multipled by 1000 to output file (to avoid problem with small numbers) Write Write Write Write Write Write Write Write Write Next I #2, Int(first_x * 1000) #2, Int(firsty * 1000) #2, Int(first_z * 1000) #2, 1nt(secondx * 1000) #2, Int(secondy * 1000) #2, Int(second_z * 1000) #2, Int(third_x * 1000) #2, Int(third__y * 1000) #2, Int(third_z * 1000)

Mytable. Close Close 2 Close 1

Tell user that processing is finished

Forms! [convert
End Function

_stl]

! [Processing] = "processing completed"

D-3

APPENDIX

PROGRAM

LISTING

FOR "FACETSIN"

MODULE

AutoLlSP
*k . *`

Program

Listing for "facetsin"

Module

i******************************************************** **

FACETSIN. LSP ** **
**

**

** Reads facet data from ACCESS databaseand creates graphical entities

**

**

Written by Ian Campbell, last updated 13/5/97 **

**

******************************'K*****************************

*****x**T*******

MAIN

PROGRAM

***************

(defiln C: FACETSIN () (init_facetsin)

(open_file) (process_file) (close_file)

Initialisation ;
file Open ; Create graphical entities ; file Close ;

xx*Xxx*******INITIALISATION

*****************

(defun init_facetsin () (setvar "cmdecho" 0) ;no echoing of commands

X X,

XX*X*"**

OPEN FILE FOR INPUT ***"`^`******`*`****

(defun open_file () input file input of name... ;user for input: ")) "\nFilenarne (getstring (setq file_narne "r")) filename (open in_file (setq

E-1

**********************

PROCESS-FILE

*****************

(defun process file (/ txth(-; t count facet pt t pt2 pt3 ptav firstx firsty firstz secondx secondy secondz thirdx thirdy thirdz nilfrstx milfirsty nilfirstz milsecondx milsecondy milsecondz milthirdx milthirdy rnilthirdz averagex averagey averagez diffx diffy diffz hit)

(setq count (read (read-linein_tile)))

(while (> count 0)
(setq facet (read (read-line in_file)))

in facets ;read number of

facet ;read number

;read vertices co-ordinates multiplied by 1000 (setq (setq (setq (setq (setq (setq (seta (setq (setq in_file))) (read (read-line milfirstx milfirsty (read (read-line in_file))) (read in_file))) (read-line rnilfirstz (read in_file))) (read-line rnilsecondx in_file))) (read-line (read milsecondy in_file))) (read-line (read rnilsecondz in file))) (read (read-line rnilthirdx in_file))) (read (read-line milthirdy in_file))) (read (read-line milthirdz

divide by 1000 all ; values

(setq firstx (setq firsty (seta firstz milfirstx 1000)) milfirsty 1000)) 1000)) milfirstz

(setq secondx milsecondx 1000)) (setq secondy milsecondy 1000)) (setq secondz milsecondz 1000)) (setq thirdx milthirdx 1000))
1000)) (setq thirdy milthirdy (setq thirdz (/ milthirdz 1000)) height text ;calculate required (seta (setq (setq (setq (setq (setq (setq firstx (/ (+ 3)) thirdx) secondx averagex (/ (+ firsty )) 3 thirdy) secondy averagey firstz (/ (+ 3)) thirdz) secondz averagez diffx (expt (- firstx averagex) 2 )) diffy (expt (- firsty averagey) 2 )) diffz (expt (- firstz averagez) 2 )) diffz)) dify difx 20)) (sqrt (+ txthgt

(if (< txthgt 0.1) (setq txtligt 0.1))

E-2

input ; a 3D polyline through three vertices and text at centre of facet ; (setq pt 1 (list firstx firsty firstz)) (setq pt2 (list secondx secondy secondz)) (seta pt-3'(list thirdx thirdy thirdz)) (setq ptav (list averagex averaey averagez)) (command "3dpoly" pt l pt2 pt-3)"C") (command "text" ptav txthgt "0" facet) (setq count (- count 1))

**x***** (defn close file ()

(close in_file)

CLOSE**********

XXx*Xx""
xx*xcic;

"*"*
cxxccx!

END OF FACETSIN. LSP ***"`*`**`**`*******

cx; ccc; cxcjcxxxcjccxhcjc; c(ccc; c;cxxxcxccccccxciccc

cx:; cxccxcx!

E-3

APPENDIX

PROGRAM

LISTING

FOR "SELFACET"

MODULE

AutoLISP

Program

Listing for "selfacet"

Module

**

'**

SELFACET. LSP

**

**
*x

** Selects facet numbers from AutoCAD drawing dumps list into file and
**

**

Written by Ian Campbell, last 18/7/97 updated **

**********x*********

MAIN

PROGRAM

******************

(defun C: SELFACET (/ facetno )

(init_selfacet) (open_files) Initialisation Open files

(dump_file) (cony file) (close-files)

Create facets list ; of Converts ; strings to numbers ;Close files

xx*

*****X

INITIALISATION

******

******x***

***

(defun init_selfacet () (setvar "cmdecho" 1) ;no echoing of commands

X;xXX*xXX*X"**
(defun open_files ()

OPEN FILE FOR INPUT

file input name... of output ,user (setq filename (getstring "\nFilenarne for surface feature: ")) (setq temp_file (open "temp" "w")) (setq out_file (open filename "w"))

F-i

********T***************

DUMP

FILE **************************

(defun dump file (/ entcount entname entlist enttext enttype index )

Ask user to select required facets from screen (print "Please select facet for feature") numbers surface
(seta entlist (ssget)) (seta entcount (sslength entlist)) (seta index 0) allows user to select entities counts number of entites selected

(seta facetno 0)
(while (< index entcount) (setq (setq (setq (if (= (if (= ; while number of entites not reached

entname (ssname entlist index)) , take next entity find its data entdata (entet entnarne)) ; find its (cdr (assoc 0 type enttype ; entdata))) enttype "TEXT" ) (setq enttext (cdr (assoc I entdata)))) enttype "TEXT" ) (print enttext ternp_file)) , if text then file to temp text ;print value (if (= enttype "TEXT" ) (setq facetno (+ l facetno))) ; count facets (seta index (+ I index))

(close ternp_file)
cxeccjeijcc rxc c. icx FxxccCcK k

CONVERT

k'XXk

k>k

xkkkFkkkkk

(defiun conv_file (/ string length column number char ) (setq temp file (open "temp" "r")) (read-line temp file) (while (> facetno 0) file temp ; open

facet to in number equivalent convert and string number a read (setq (setq (seta (seta (read-line temp_file)) string length (- (strien string) 2)) 1) column 0) number

(while (> length 1) (setq char (substr string length 1))

(if (if (if (if

(= (= (= (=

"0") char "1 ") char char "2") "3 char ")

(setq number (+ (setq number (+ (setq number (+ (seta number (+

(* number (* number (* number number (*

0 column)))) I column)))) 2 column)))) 3 column))))

F-2

(if (if (if (if (if (if

(= (= (= (= (= (=

char char char char char char

"4") "5") "6") "7") "S") "9")

(setq (setq (setq (setq (seta (setq

number number number number number number

(+ (+ (+ (+ (+ (+

number number number number number number

(* (* (* (* (* (*

4 5 6 7 8 9

column)))) column)))) column)))) column)))) column)))) column))))

(setq length (- length 1)) (setq column (* column 10))

write converted number to output file (print number out file) (seta facetno (- facetno 1))

(close temp file)

xxxxxxxrxX*xxxxX

CLOSE

xx*x***x**********x**********

(defun close-files ()
(close out file)

, **i; .

** `**********
c*x{cr,, ***x*: ***;

END OF SELFACET. LSP *********************

c;ce! c!cc*c; c>;cc**x,; c********c$ccccc**; cc*cxccc*>; c{ccccc

F-3

APPENDIX

PROGRAM

LISTING

FOR "ASSIGN

FACETS"

MODULE

Access Basic Program Listing for "assign_facets" Module

Option Compare Database 'Use database for order string comparisons

'Declarations
Dim MyDB As Database Dim Mytable As Recordset Dim facet no As Integer

'Main function Function assign facets ()

'Work with current database Set MyDB = DBEngine. Workspaces(0). Databases(0)

'Work with table F list Set Mytable = MyDB. OpenRecordset("F_list", DB_OPEN_TABLE)

'SQL command to clear records form F_list

DoCmd RunSQL "DELETE DISTINCTROW F_list. * FROM F_list; " 'Open facet list file file
Open Forms! [assi an_facets]! [Input_file_name] For Input As 1

'Read first line which is empty due to the wat AutoLisp writes data Input #1, newline

'Read in each facet number and write into F_list table Do While Not EOF(1)
Input # 1, facet no ' Create new record. Mytable. AddNew Mytable("facet_no") = facet-no ' Save changes. Mytable. Update

Loop Mytable. Close

G-1

'SQL command to look up in table Facets all the facets listed in F_list and to 'write the name of the facet list file in the surface feature name field

DoCrnd RunSQL "UPDATE DISTINCTROW F_list INNER JOIN Facets ON F_ list. facet_no = Facets.Facet_number SET Facets.Surface_feature =
[Forms]! [assign-facets]! [surface_feature_narne]; "

Close 1
End Function

G-2

APPENDIX

PROGRAM

LISTING

FOR "CALL

SURF FIN" MODULE

Access Basic Program Listing for "calc surf fin" Module

Option Compare Database 'Use database order for string comparisons

'Declarations Dim MyDB As Database Dim MyTable As Recordset Dim DataTable As Recordset Dim Row As Long Dirn Value, Facet, Worstfacet, Count, Average, Area, Total_area As Single Dim Vx, Vy, Vz, Xang, Yang, Ax, Ay, Vxnew, Vynew, Vznew, Vxy, Angle As Single Dirn Actual, Surfin, Ratio, Ratiomax, Bestax, Bestay, Bestsol As Single Dim Problem As String

'Main function
Function calc_surf fin ()

'Work with current database Set MyDB = DBEngine. Work spaces(0). Databases(0)
'SQL command to create a list of facets with surface finish values for current 'part in the form of a table called Surf fin_list DoCmd RunSQL "INSERT INTO Surf fin_list ( Part name, Surface feature, Surface finish, Facet_number, Normal x, Normal_y, Normal z, Area ) SELECT DISTINCTROW Parts. Part_name, Surf features. Surface feature, Surface_finish, Facets.Facet number, Facets. Normal_x, Surf _features. Facets. Normal_y, Facets.Normal_z, Facets. Area FROM (Parts INNER JOIN Surf features ON Parts. Part_name = Surf features. Part_name) INNER JOIN Surface_feature Facets. WHERE Surface_feature features. Surf ON Facets = ((Parts. Part_narne=[Forrns] ! [cal c_surf fin] ! [part_name])), " 'Work with Surf fin list table Set MyTable = MyDB. OpenRecordset("Surf fin-list", DB-OPEN-TABLE)

finish for particular process values 'Work with table containing surface fin] [Process_name], ! [calc_surf OpenRecordset(Forrns! Set DataTable =MyDB. DB OPEN TABLE) file 'Open an output

Output For As I [Output_file_narne] fin]! [calc_surf Open Forms!

H-1

'Set best solution to high very value Bestsol = 100 'Set initial X angle to zero Xang=0 'Loop to increment through rotation angles about x axis Do While Xang <= 1$0

'Set initial Y angle to zero Yang =0

'Loop to increment through rotation angles about y axis Do While Yang <= 180

'Move to first facet in Surf fin list table

MyTable. MoveFirst

'Set intitial values

Ratiornax =0 Count =0 Total area =0 Average =0 Problem = "ok"

'Convert degrees to radians

Ax = Xang * 01745-3292 . Ay = Yang * 017453292 .

fin_list in facet Surf increment table through 'Loop to every Do While Not MyTable. EOF ' Move to first record in data table DataTable.MoveFirst 'Read facet number and its surface finish value Facet = MyTable("Facet_number") Surfin = MyTable("Surface_finish")

H-2

'Read facet normal vectors and area

Vx = MyTable("Normalx") Vy = MyTable("Normalte") Vz = MyTable("Normal_z") Area = MyTable("Area") 'Calculate new normal vectors after rotation Sin(Ax))) Vxnew = (Vx * Cos(Ay)) + (((Vz * Cos(Ax)) + (Vy

* Sin(Ay))

Vynew = (Vy * Cos(Ax)) (Vz * Sin(Ax)) Vznew = (((Vz * Cos(Ax)) + (Vy * Sin(Ax))) Cos(Ay)) (Vx * Sin(Ay)) Vxy = Sqr((Vxnew * Vxnew) + (Vynew * Vynew)) 'Calculate angle of normal to vertical and set to 'value between 0 and 180 degrees Angle = ((Atn(Vxy / (Vznew +. 0000001))) * 57.3) If (Angle <= 0) Then Angle = Angle + 180

'Look-up table for actual surface finish If (Ankle >0 And Angle <= 5) Then Row =0 If (Angle >5 And Angle <= 10) Then Row =1 If (Angle > 10 And Angle <= 15) Then Row = 2
If If If If (Angle (Angle (Angle (Angle > 15 And >20 And > 25 And > 30 And Angle Angle Angle Angle <= <= <= <= 20) 25) 30) 35) Then Then Then Then Row Row Row Row = = = = 3 4 5 6

If (Angle > 35 And Angle <= 40) Then Row = 7 If (Angle > 40 And Angle <= 45) Then Row = 8 If (Angle > 45 And Angle <= 50) Then Row = 9
If (Angle > 50 And Angle <= 55) Then Row = 10

If (Anmale > 55 And Angle <= 60) Then Row = 11 If (Angle > 60 And Angle <= 65) Then Row = 12
If (Angle > 65 And Angle <= 70) Then Row = 13

If (Ankle > 70 And Angle <= 75) Then Row = 14

If If If If (Angle (Angle (Angle (Angle > > > > 75 80 85 90 And And And And Angle Angle Angle Angle <= <= <= <= 80) 85) 90) 95) Then Then Then Then Row Row Row Row = = = = 15 16 17 18

If (Angle > 95 And Angle <= 100) Then Row = 19 If (Angle > 100 And Angle <= 105) Then Row = 20 If (Angle > 105 And Angle <= 110) Then Row = 21

If (Angle >l 10 And Angle <= 115) Then Row = 22

If (Ankle > 115 And Angle <= 120) Then Row = 23 If (Angle > 120 And Angle <= 125) Then Row = 24

H-3

If If If If If If If If If If If

(Ankle (Angle (Angle (Angle (Angle (Angle (Angle (Angle (Angle (Angle (Angle

> > > > > > > > > > >

125 And 130 And 135 And 140 And 145 And 150 And 155 And 160 And 165 And 170 And 175 And

Angle Angle Angle Angle Angle Angle Angle Angle Angle Angle Angle

<_ <= <_ <= <= <_ <_ <= <_ <_ <=

130) Then Row = 25

135) Then Row = 26 140) Then Row = 27 145) Then Row = 28

150) Then Row = 29

155) Then Row = 30

160) 165) 170) 175) 180) Then Then Then Then Then Row Row Row Row Row = = = = =35 31 32 33 34

'Move to selected row in data table and read value DataTable. Move Row Actual = DataTable(" Value")

finish

'Calculate ratio between actual and required surface

Ratio = Actual / Surfin

'Keep note of worst ratio and facet where this occurs If (Ratio > Ratiornax) Then Worstfacet = Facet If (Ratio > Ratiornax) Then Ratiornax = Ratio
'Calculate weighted average value of ratio Count = Count +1 Average = ((Average * Total-area) + (Ratio * Area)) / (Total-area + Area)

Total area = Total area + Area

'Move to next record in Surf fin_list table and loop back MyTable. MoveNext Loop

'If actual surf fin is greater than required for any facet then
indicate a problem If (Ratiornax > 1) Then Problem = "problem"

H-4

'Write average surface finish ratio, rotation angles, worst 'surface finish ratio, worst facet and "problem" to file output Write # 1, Average, Xan, Yang, Worstfacet, Ratiornax, Problem 'Keep note of rotation ankles with lowest average ratio If (Average < Bestsol) Then Bestax = Xang If (Average < Bestsol) Then Bestay = Yank

If (Average < Bestsol) Then Bestsol = Average

'Increment Y angle and loop back Yang = Yang +5

Loop
'Increment X angle and loop back Xang = Xang +5

Loop 'Write best angles to output file

Write #1, "Best X angle and Y angle are", Bestax, Bestay Forms! [calc_surf fin] ! [Xang] = Bestax

Forms! [calc_surf fin] ! [Yang] = Bestay Close 1 MyTable. Close DataTable. Close
is for fin_list Surf time called module next ready 'SQL commnad to empty

*FROM fin_Iist. Surf DISTFNCTROW DoCrnd RunSQL "DELETE Surf fin list; " End Function

H-5

APPENDIX EXAMPLE OF NEUTRAL

I FACET DATA

FILE CONTAINING

Example of Neutral Facet File (shortened) 12 663 10000 10000 10000 10000 0 0 10000 10000 0 664 10000 10000 10000 10000 0 10000 10000 0 0 665 10000 10000 10000 10000 10000 0 0 10000 0 (number of facets) (facet nuber) (first x value) (first y value) (first z value) (second x value) (second y value) (second z value) (third x value) (third y value) (third z value) (next facet)

666 0
10000 0 0 10000 10000 10000 10000 10000

I-1

APPENDIX

EXAMPLE

OF OUTPUT FILE WITH ORIENTATION

VALUES

Example of Output 702417727582504, . 679849-3318571048, . 672354642869662 . , 707-33 16492599027, . 674661717522 346, 676200453530514, . 676754233768428, . 639440416186295, . 641491365951386, .

File (shortened) 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 5, 10, 15, 20, 25, 30, 35, 40, 791, 791, 83 7, 837, 839, 836, 836, 837, 834, 1.68, 1.68, 1.6, 1.67, 1.6, 1.6, 1.67, 1.6, 1.6, "problem" "problem" "problem" "problem" "problem" "problem" "problem" "problem" "problem"

630682449373435, . 618196206455709, . 613240364737681, .

640896525569451, .

0, 0, 0,
0,

45 , 50 , 55,
60,

836 , 836 , 831,

834,

1 67 . , 16 . , 1.6,
1.67,

" problem" "problem" "problem"

"problem"

416989341781684, 399652871726105, 380939638981562, 410949547810325, 414027215660877,

45, 45, 45, 45, 45,

135, 140, 145, 150, 155,

845, 845, 843, 843, 843,

1.17, 1.17, 1.07, 1.07, 1.07,

"problem" "problem" "problem" "problem" "problem"

407810204712362, 397922060079624, 404506738914734, 408323928748035, 411397151860743, 434965038007989,

55, 55, 55, 55, 55, 55,

135, 140, 145, 150, 155, 160,

845, 845, 845, 804, 804, 785,

1.07, 1.07, 1.07, 96, . 1.12, 1.28,

"problem" "problem" "problem" "ok" "problem" "problem"

448695505586906,

55,

165,

785,

1.6,

"problem"

1.6, 160, 839, 674661717522346, 180, . 1.67, 165, 837, 707316492599027, 180, . 1.6, 672'-)'54642869662, 170, 83 7, 180, . 1.68, 679849318571048, 175, 791, 180, . 1.68, 702417727582504, 180, 180, 791, . 45, 145 "Best X angle and Y angle are",

676200453530514, .

180,

155,

836,

1.6,

"problem"
"problem" "problem" "problem" "problem" "problem"

J-1

APPENDIX

RP USAGE ADVISOR QUESTIONS FROM LATEST VERSION OF DSS FOR RP

RP Usage Advisor component: 1.

suitability on questions -

of using RP for a particular

Will you require a physical model of this component at sometime during the design process?

?.
3.

Is it important that this model is made as quickly as possible?

Does the complexity of this component make it difficult to produce a model using CNC machining?

4.
5.

Can you produce a CAD solid model for the design of this component?
Do you have ready acces to rapid prototyping facilities?

K-1

APPENDIX L PUBLICATIONS

Publications 1.

Resulting

form Research: -

Campbell R. I. & Bernie M. R. N. Creating a Database of Rapid Journal of Materials Processing Pr-ofotyping System Capabilities. Technology, Vol 61, No 1-2,1996. pp 163-167. Campbell R. I. Using Feature-based Design to Optimize Rapid Prototypiiig. Journal of Engineering Design, Vol 7, No 1,1996. pp95103.

3.

Campbell R. I., Chrisp A. G. & Geldart M. Using Features to Integrate Desigui mid Rapid Pr-ototypiiig. Proc. 13th International Conference on Production Research, Freund Publ. House, London, 1995. Addendum ppa-6.

4.

Developing Campbell R. I. Design for- Rapid Prototypiiig; a , Methodology. Proc. of 10th National Conference on Manufacturing Research, Loughborough Univ. of Technology, 1994. pp 521-525.

L-1