Ghazy 12 PDF

Development of an Additive
Manufacturing Decision Support

System (AMDSS)
A thesis submitted to the Faculty of Science,

Agriculture and Engineering for the Degree of
Doctor of Philosophy
By
Mootaz M. Ghazy
M.Sc.
October 2012
School of Mechanical and Systems Engineering

Newcastle University
Abstract
Additive manufacturing (AM) technology describes a set of processes capable of

producing 3D physical products from CAD data directly. The rapid development
of AM technologies and their wide applications makes the selection of the suitable
process chains and materials a difficult task. Some researchers have tackled this
problem by developing selectors that should assist users in their selections. The
existing selector systems have some drawbacks: () often being outdated even
before they were completely developed because new processes and materials are
evolving continuously, () representing only the point of view of their developers
because users were not involved in the development process and (iii) not being
holistic and able to help in all AM aspects for example process chains, materials,
finishing methods and machines. This work has developed an updatable decision
support system that assists users in their selections regarding AM process chains,
materials, finishing methods, and machines. First, the study started by analyzing
the available additive manufacturing selector systems and identifying their
shortcomings. Secondly, the researcher identified target specifications for the new
system, investigated different possible architectures for the system, selected
knowledge based system (KBS) and database (DB) architecture to work together
as a versatile tool that achieves the required target specifications. Next, the first
version of the system was developed. Furthermore, verification and validation
processes were made to test the developed system. Three case studies were used
for the validation purpose: a typical consumer razor blade and two automotive
components. These case studies were manufactured using AM technologies and
then a comparison between real life decisions and the developed decision support
system decisions were made. In addition, a number of interviews were performed
in order to obtain users feedback about the first developed version. As a result of
the feedback and evaluation a second version of the system was developed and
evaluated. The results obtained from the second evaluation suggest that the second
version is more effective than the first version during the selection process. To
conclude, this study has shown that using KBS and DB together is effective to
develop an updatable additive manufacturing decision support system. In addition,
the user involvement in the development stage of the system enhances the system
performance.
i
Acknowledgments
First of all, I would like to underline that this research would not have been
possible without the support of my supervisor Prof. Kenneth Dalgarno. He has
been very helpful and offered me invaluable assistance, support and guidance
through the research journey. I would like also to express my gratitude to Dr.
Javier Munguia for his support and help in this work. His inputs were really
valuable and useful.
I want to express my gratitude to the Egyptian Factories that offered me case

studies: Lord Precisions Industries and Amreya Metal Co (AMC). Special thanks
to Eng. Ossama Saleh who facilitate the experimental work in Amreya metal CO.
I would like also to thank the organizations and persons who offered me case
studies in UK: RCID in Newcastle University, Keyworth Rapid Manufacturing
Limited in Leeds University, Paragon Rapid technologies Limited and finally Mr.
Shah Khan in my research group.
Many thanks to the people who tested the developed system, responded to
questionnaires and gave me a valuable feedback that helped me to improve my
work.
I would like also to thank my friend Dr.Shady Gadoue the lecturer at the school of
electrical and electronic engineering in Newcastle University for helping me to
learn the basics of Matlab which facilitate to me the programming part in my
thesis.
Furthermore, I would like to thank my University in Egypt: The Arab Academy

for Science & Technology & Maritime Transport for sponsoring me and giving
me the opportunity for doing research in Newcastle University in UK.
Last but not least, I would like to express my endless love to my family for their
understanding and support through the duration of my study.
ii
Table of Contents
Abstract.. .................................................................................................i
Acknowledgments ......................................................................................... ii
Table of Contents ......................................................................................... iii
List of Tables .............................................................................................vi
List of Figures ............................................................................................. vii
List of Abbreviations ..................................................................................... x
Chapter 1. Introduction ............................................................................... 1
1.1 Introduction ........................................................................................................... 1
1.2 Research Problem .................................................................................................. 2
1.3 Research Aim and Objectives ................................................................................ 3
1.3.1 Aim ......................................................................................................................... 3
1.3.2 Objectives .............................................................................................................. 4
1.4 Research Road Map ............................................................................................... 4
1.5 Thesis Outline ........................................................................................................ 5
Chapter 2. Literature Survey ...................................................................... 6

2.1 Additive Manufacturing Overview ......................................................................... 6
2.2 Additive Manufacturing Application Areas............................................................ 7
2.2.1 Rapid Prototyping .................................................................................................. 7
2.2.2 Rapid Manufacturing ............................................................................................. 7
2.2.3 Rapid Tooling ......................................................................................................... 7
2.3 Additive Manufacturing Processes ........................................................................ 8
2.3.1 Stereolithography (SLA) ......................................................................................... 9
2.3.2 Fused Deposition Modeling (FDM) ...................................................................... 10
2.3.3 Selective Laser Sintering (SLS) / Selective Laser Melting (SLM)........................... 11
2.3.4 3D Printing (3DP) ................................................................................................. 13
2.3.5 Ink-Jet / Multi-jet Modeling (MJM) ..................................................................... 14
2.3.6 Laser Engineering Net Shape (LENS) .................................................................... 15
2.3.7 Electron Beam Melting (EBM) ............................................................................. 16
2.4 Additive Manufacturing Process Chains .............................................................. 17
2.4.1 Rapid Tooling Process Chains .............................................................................. 17
iii
2.4.2 General AM Process chains.................................................................................. 20
2.5 Additive Manufacturing Selection Systems ......................................................... 22
2.5.1 Rapid Prototyping Selectors ................................................................................ 22
2.5.2 Rapid Manufacturing Selectors............................................................................ 25
2.5.3 Rapid Tooling Selectors........................................................................................ 28
2.6 Limitations of the Developed Selectors ............................................................... 33
2.7 Research Opportunities ....................................................................................... 33
Chapter 3. Additive Manufacturing Decision Support System (AMDSS)

Initial Development ................................................................. 35
3.1 Identifying System Target Specifications ............................................................. 35
3.2 Developing the System ........................................................................................ 36
3.3 AMDSS General Overview.................................................................................... 39
3.3.1 Graphical User Interface ...................................................................................... 43
3.3.2 Knowledge Based System .................................................................................... 44
3.3.3 Database .............................................................................................................. 78
3.4 Discussion ............................................................................................................ 81
Chapter 4. Verification and Validation.................................................... 82

4.1 Verification........................................................................................................... 82
4.2 Validation ............................................................................................................. 85
4.2.1 Razor Blade Case Study........................................................................................ 86
4.2.2 Automotive Case Study (1) .................................................................................. 88
4.2.3 Automotive Case Study (2) .................................................................................. 95
4.3 Discussion .......................................................................................................... 100
Chapter 5. User Feedback and Evaluation ............................................ 101

5.1 User Feedback.................................................................................................... 101
5.1.1 Questionnaire Design ........................................................................................ 102
5.1.2 Questionnaire Observations and Analysis ......................................................... 109
5.2 User Testing and Evaluation .............................................................................. 115
5.2.1 Cover Part .......................................................................................................... 115
5.2.2 Drain Fitting Part ................................................................................................ 118
5.2.3 Filter Cover Part ................................................................................................. 119
5.2.4 Dental Implant Part Assembly ........................................................................... 123
Chapter 6. Additive Manufacturing Decision Support System Further

Development and Assessment .............................................. 126
6.1 AMDSS Second Version...................................................................................... 126
iv
6.1.1 List of Modifications .......................................................................................... 126
6.1.2 Graphical User Interface Modifications ............................................................. 127
6.1.3 Knowledge Based System Modifications ........................................................... 128
6.2 Second Feedback ............................................................................................... 156
6.2.1 Second Questionnaire Design ............................................................................ 156
6.2.2 Second Questionnaire Observations and Analysis ............................................ 160
6.3 Second Version User Testing and Evaluation .................................................... 162
Chapter 7. Discussion ............................................................................... 169

7.1 Summary of the Work........................................................................................ 169
7.1.1 Selector System Shortcomings .......................................................................... 169
7.1.2 Target Specifications .......................................................................................... 170
7.1.3 AMDSS First Version Development.................................................................... 171
7.1.4 System Verification and Validation.................................................................... 171
7.1.5 First Feedback .................................................................................................... 172
7.1.6 AMDSS Second Version Development ............................................................... 173
7.1.7 Second Feedback ............................................................................................... 173
7.1.8 AMDSS Limitations............................................................................................. 174
7.2 Research Contributions ..................................................................................... 175
7.3 Possible Future Forms of AMDSS....................................................................... 176
7.3.1 Standalone Application ...................................................................................... 176
7.3.2 CAD Plug-in ........................................................................................................ 177
7.3.3 Online Application ............................................................................................. 178
Chapter 8. Conclusions and Future Work............................................. 179

8.1 Review of Objectives.......................................................................................... 179
8.2 Conclusions ........................................................................................................ 180
8.3 Future Work ....................................................................................................... 181
Appendix A. First Questionnaire Responses ........................................... 183

Appendix B. Second Questionnaire Responses ....................................... 213
References............................................................................................... 226
v
List of Tables
Table 2-1: Different Attempts for Additive Manufacturing Selection. .......................................... 32

Table 3-1: IDEF0 Node Tree Index.................................................................................................. 47
Table 4-1: An Example of Results Retrieved from a Verification Scenario. ................................... 84
Table 4-2: Part and Material Requirements for the Top Cap. ....................................................... 87
Table 4-3: Part and Material Requirements for the Two Parts of Automotive Case Study 1. ....... 91
Table 4-4: Part and Material Requirements for the Two Parts of Automotive Case Study 2. ....... 97
Table 5-1: AMDSS First Version Questionnaire (Page 1 of 6). ..................................................... 103
Table 5-7: Part and Material Requirements for the Cover Part................................................... 116
Table 5-8: Part and Material Requirements for the Drain Fitting Part. ....................................... 118
Table 5-9: Part and Material Requirements for the Filter Cover Part. ........................................ 120
Table 5-10: Part and Material Requirements for the Dental Implant Assembly Part.................. 124
Table 6-1: IDEF0 Node Index (version2)....................................................................................... 130
Table 6-2: AMDSS Second Version Questionnaire (Page 1 of 4).................................................. 157
vi
List of Figures
Figure 2.1: Additive Manufacturing Different Applications (Wohlers,2010) ................................... 8

Figure 2.2: Number of AM Machines Sold Each Year (Wohlers,2010). ........................................... 8
Figure 2.3: SLA Process (VG,2012). .................................................................................................. 9
Figure 2.4: FDM Process (Custompart.net,2012)........................................................................... 10
Figure 2.5: SLS Process (VG,2012).................................................................................................. 12
Figure 2.6: 3DP Process (Custompart.net,2012). ........................................................................... 13
Figure 2.7: Ink-jet Process (Hanemann et al.,2006)....................................................................... 14
Figure 2.8: MJM Process (Hanemann et al.,2006) ........................................................................ 15
Figure 2.9: LENS Process (Zhao et al.,2009) ................................................................................... 16
Figure 2.10: EBM Process (Lu et al.,2009). .................................................................................... 17
Figure 2.11: RIM Process................................................................................................................ 18
Figure 2.12: Investment Casting Process (adapted from Jones and Yuan,2003)........................... 20
Figure 2.13: Possible AM Process Chains ....................................................................................... 21
Figure 3.1: Schematic Diagram of Neural Network (Lu et al.,2012) .............................................. 36
Figure 3.2: Comparison between KBS and KBS + Database Architectures. ................................... 39
Figure 3.3: AMDSS Main Structure. ............................................................................................... 40
Figure 3.4: General Framework of the Selection Part of the KBS. ................................................. 41
Figure 3.5: Start Screen of the Developed AMDSS. ....................................................................... 44
Figure 3.6: IDEF0 Node Tree Diagram ............................................................................................ 45
Figure 3.7: A-0, Develop an Additive Manufacturing Decision Support System. .......................... 48
Figure 3.8: A0, Main Steps of the Selection Part of the KBS. ......................................................... 49
Figure 3.9: A1, Check Additive Manufacturing Possible Processes. .............................................. 50
Figure 3.10: A2, Check Possible Materials for the Possible Processes. ......................................... 51
Figure 3.11: A3, Rank Using SMART Method. ................................................................................ 52
Figure 3.12: Process Selection Screen. .......................................................................................... 53
Figure 3.13: List of the Possible Processes that Could Manufacture the Part. .............................. 54
Figure 3.14: Screen Shows the Process Reasons that Prevent AMDSS from Giving a Solution. ... 55
Figure 3.15: Table that Shows the Process Parameters that are not Feasible Together............... 55
Figure 3.16: A Window that Gives a Concern to the User about Part Quality.............................. 56
Figure 3.17: Capability Analysis Screen.......................................................................................... 56
Figure 3.18: Checking AM Processes Flowchart. .......................................................................... 58
Figure 3.19: Material Properties Screen. ....................................................................................... 59
Figure 3.20: List of Possible Processes and Material that Could Manufacture the Part. .............. 60
Figure 3.21: Screen Shows the Material Reasons that Prevent AMDSS from Giving a Solution. .. 60
Figure 3.22: Table that shows the material Parameters that are not Feasible Together. ............. 61
Figure 3.23: Material Capability Analysis. ...................................................................................... 61
Figure 3.24: Checking AM Materials Flowchart. ............................................................................ 63
Figure 3.25: Criteria Screen............................................................................................................ 65
Figure 3.26: Final Ranking Screen. ................................................................................................. 65
Figure 3.27: SMART Ranking Flowchart ......................................................................................... 67
Figure 3.28: SMART Ranking Flowchart (continued). .................................................................... 68
Figure 3.29: List of the Processes that Need Intermediate Materials. .......................................... 69
Figure 3.30: Selection of Intermediate Material Screen. ............................................................... 69
Figure 3.31: Selection of the Process that the User Wants to See its Finishing Methods. ............ 70
vii
Figure 3.32: Selecting from the Finishing Options Screen. ............................................................ 70
Figure 3.33: Selection of the Processes that the User Needs to See its Machines Screen............ 71
Figure 3.34: Selection of the Machines Screen.............................................................................. 71
Figure 3.35: Technical Summary Screen. ....................................................................................... 72
Figure 3.36: Browsing Processes by Material. ............................................................................... 73
Figure 3.37: Browsing Materials by Process. ................................................................................. 74
Figure 3.38: Adding a New Process................................................................................................ 75
Figure 3.39: Adding a New Material. ............................................................................................. 76
Figure 3.40: Adding a New Intermediate Material. ....................................................................... 76
Figure 3.41: Adding a New Machine. ............................................................................................. 77
Figure 3.42: Adding a Finishing Process. ........................................................................................ 78
Figure 3.43: Database Screen Shot. ............................................................................................... 79
Figure 3.44: Entity Relationship Diagram (ERD)............................................................................. 80
Figure 4.1: Guard Bar Part. ............................................................................................................ 86
Figure 4.2: Razor Case Study Manufactured Parts. ....................................................................... 87
Figure 4.3: AMDSS Output for the Guard Bar Part. ....................................................................... 88
Figure 4.4: The First Part of the Automotive Case Study 2. ........................................................... 89
Figure 4.5: The Second Part of the Automotive Case Study 2. ...................................................... 89
Figure 4.6: Rapid Tooling Patterns. ................................................................................................ 90
Figure 4.7: The Final Products after Casting. ................................................................................. 90
Figure 4.8: List of the Possible Processes and Materials for Automotive Case Study 1. ............... 92
Figure 4.9: Criteria Weighting for Automotive Case Study 1. ........................................................ 93
Figure 4.10: Final Ranking Screen for Automotive Case Study 1. .................................................. 93
Figure 4.11: Choosing Finishing Methods for Automotive Case Study 1. ...................................... 94
Figure 4.12: The Technical Summary Screen for Automotive Case Study 1. ................................. 94
Figure 4.13: The First Part of the Automotive Case Study 2. ......................................................... 95
Figure 4.14: The Second Part of the Automotive Case Study 2. .................................................... 95
Figure 4.15: SLS Fabricated Patterns. ............................................................................................ 96
Figure 4.16: The Castings. .............................................................................................................. 96
Figure 4.17: List of the Possible Processes and Materials for Automotive Case Study 2. ............. 98
Figure 4.18: Criteria Weighting for Automotive Case Study 2. ...................................................... 98
Figure 4.19: Final Ranking Screen for Automotive Case Study 2. .................................................. 99
Figure 4.20: Choosing Finishing Methods for Automotive Case Study 2. ...................................... 99
Figure 4.21: The Technical Summary Screen for Automotive Case Study 2. ............................... 100
Figure 5.1: Cover Part .................................................................................................................. 116
Figure 5.2: Final Ranking Screen of the Cover Part...................................................................... 117
Figure 5.3: Drain Fitting Part. ....................................................................................................... 118
Figure 5.4: Final Ranking Screen for The Drain Fitting Case Study .............................................. 119
Figure 5.5: Filter Cover Part. ........................................................................................................ 120
Figure 5.6: Screen that Shows the Material Parameters in Conflict Together. ........................... 121
Figure 5.7: Possible Process and Material that Can Manufacture the Filter Cover Part after
Removing the Black Color Requirement ...................................................................................... 122
Relaxing Hardness Requirement. ................................................................................................. 122
Figure 5.9: Dental Implant Assembly Part. .................................................................................. 123
Figure 5.10: Final Ranking Screen of the Dental Implant Part. .................................................... 125
Figure 6.1: Start Screen of the Second Version of the AMDSS .................................................... 128
viii
Figure 6.2: IDEF0 Node Tree Diagram for the Second Version of the AMDSS ............................. 129
Figure 6.3: Selection Path of the First Version of the AMDSS ..................................................... 131
Figure 6.4: Selection Paths of the Second Version of the AMDSS ............................................... 131
Figure 6.5: A-0, Develop an Additive Manufacturing Decision Support System (Version2) ........ 133
Figure 6.6: A0, Main Steps of the Selection Part of the KBS (Version 2). .................................... 134
Figure 6.7: A1, Check AM Processes and Materials (Version 2). ................................................. 135
Figure 6.8: A11, Check AM Processes (Version 2). ...................................................................... 136
Figure 6.9: A12, Check AM Materials by Properties (Version 2).................................................. 137
Figure 6.10: A2, Rank Using SMART Method (Version 2). ........................................................... 138
Figure 6.11: Process Selection Screen of AMDSS (Version 2). ..................................................... 139
Figure 6.12: Checking AM Processes Flowchart of the AMDSS (Version 2)................................. 141
Figure 6.13: Choosing Material Screening Method of the AMDSS (Version2). ........................... 142
Figure 6.14: Material Properties Screen of the AMDSS (Version 2). ........................................... 142
Figure 6.15: Material Class Screen of the AMDSS (Version 2). .................................................... 143
Figure 6.16: Checking AM Materials Flowchart (Version 2). ....................................................... 146
Figure 6.17: Checking AM Materials Flowchart (Version 2 Continued). ...................................... 147
Figure 6.18: Criteria Screen (Version 2). ...................................................................................... 148
Figure 6.19: SMART Ranking Flowchart (Version 2). ................................................................... 150
Figure 6.20: SMART Ranking Flowchart (Version 2 Continued). .................................................. 151
Figure 6.21: Final Ranking Screen of the AMDSS (Version 2). ..................................................... 152
Figure 6.22: SMART Calculations (Version 2). .............................................................................. 152
Figure 6.23: Finishing Options Screen in the AMDSS (Version 2). ............................................... 153
Figure 6.24: Add a Process in the Database (Version2). .............................................................. 154
Figure 6.25: Add a New Material (Version 2)............................................................................... 154
Figure 6.26: Add a New Intermediate Material (Version2). ........................................................ 155
Figure 6.27: Add a New Finishing Method (Version2). ................................................................ 155
Figure 6.28: Process Selection Screen Showing Volume Calculation (Version 2). ....................... 162
Figure 6.29: Finishing Options Screen for the Cover Part (Version 2). ........................................ 164
Figure 6.30: Finishing Options Screen for the Drain Fitting Part (Version 2). .............................. 165
Figure 6.31: Finishing Options Screen for the Filter Cover Part (Version 2). ............................... 165
Figure 6.32: Finishing Options Screen for the Dental Implant Part (Version 2). ......................... 166
Figure 6.33: Material Properties Screen for the Filter Cover Part (Version 2). ........................... 166
Figure 7.1: Adding ODBC Data Source in Windows ..................................................................... 177
ix
List of Abbreviations
ABS Acrylonitrile Butadiene Styrene

AHP Analytical Hierarchical Process
ALM Additive Layer Manufacturing
AM Additive Manufacturing
AMDSS Additive Manufacturing Decision Support System
ANP Analytical Network Process
ASTM American Society for Testing and Materials
CAD Computer Aided Design
DB Database
DLS Direct Laser Sintering
DMLS Direct Metal Laser Sintering
EBM Electron Beam Melting
EDM Electrical Discharge Machining
ES Expert System
FDM Fused Deposition Modeling
FL Fuzzy Logic
GUI Graphical User Interface
HDT Heat Deflection Temperature
HIP Hot Isostatic Pressing
HSM High Speed Machining
IC Investment Casting
IDEF0 Integration Definition Language 0
KBS Knowledge Based System
LENS Laser Engineering Net Shape
LM Layer Manufacturing
LOM Laminated Object Manufacturing
MJM Multi-Jet Modeling
NN Neural Network
QFD Quality Function Deployment
RIM Reaction Injection Molding
RM Rapid Manufacturing
x
RP Rapid Prototyping
RT Rapid Tooling
SFFF Solid Free Form Fabrication
SGC Solid Ground Curing
SLA Stereolithography
SLM Selective Laser Melting
SLS Selective Laser Sintering
SMART Simple Multi Attribute Decision Making
ST Silicone Tooling
TOPSIS Technique for Order Preference by Similarity to Ideal Solution
UV Ultra Violet
VC Vacuum Casting
3DP 3D Printing
xi
Chapter 1. Introduction
1.1 Introduction
Additive Manufacturing (AM) is defined by ASTM as the process of joining

materials to make objects from 3D model data usually layer upon layer, as
opposed to subtractive manufacturing technologies such as traditional
machining(ASTM,2012).
At the beginning the aim of the new technology developed was to build prototypes
quickly. The first generic name was rapid prototyping and the first process
developed in the late eighties was Sterolithography (SLA). When the technology
was later developed to produce tools and dies it was called rapid tooling. Finally,
when it is used to produce final products it was called rapid manufacturing (Levy
et al.,2003).
The technology advanced quickly and applications have widened into medical,
sculpture, architecture, industrial and many others domains. Different names were
given to this technology, for example free form fabrication, digital manufacturing,
layer manufacturing, 3D printing, additive fabrication, additive processes, additive
techniques, additive layer manufacturing and layer manufacturing (Pratt et
al.,2002, Dahotre and Harimkar,2008, Choi et al.,2011, ASTM,2012). The reason
for the confusion in the name is that the usage of this technology has passed over
its initial purpose.
After Sterolithography was developed many technologies have evolved like: fused
deposition modeling, 3D printing, selective laser sintering, selective laser melting,
electron beam melting, solid ground curing and laminated object modeling. Some
of these technologies are not used anymore and some are still present in the
market and are progressing. AM technologies are not only used as single
processes but also as part of a process chain; for example Sterolithography models
and investment casting: the first mould of the investment casting process is made
with SLA technology. Furthermore, most of the AM metal based processes such
as laser engineering net shape (LENS), selective laser melting (SLM) and electron
beam melting (EBM) are used for tooling processes to produce moulds and dies
for injection moulding and die casting.
1
The advances in the AM technology are not only limited to new processes, but
also new materials, new machines and new finishing methods. This expansion in
techniques and materials has created a strong need for a system that could help
and assist users in the selection of the possible process chains, materials, finishing
options and machines. This need was the main motivation for this research.
1.2 Research Problem
According to Gibson (2002) there are three main problems regarding rapid
prototyping that benefit from decision support system:
Selection support. What technology and material a user should use to build
a part?
Capital investment support. What technology should a user buy to fulfill

his requirements?
Process planning support. What are the parameters that need to be

controlled and what are the optimum process values that give the optimum
results.
The scope of this research covers the first question. When a user has a part with
specific requirements, what is the most appropriate technology and material that
could fulfill the requirements of the given part.
The target users of the system are inexperienced and average users. The
inexperienced users are those who start to learn and use AM technologies such as
students in universities while average users are more advanced and have basic
knowledge about AM technologies but they are not experts that use AM
technologies on regular basis. They have a little information about some AM
processes and materials. For example, design engineers that need to manufacture a
part and are used to conventional processes.
The growing numbers of AM technologies and the quick advance in their

processes, materials and machines make the selection of the suitable process a
hard task. There are 38214 AM systems sold around the world from 1988 until
2009 (Wohlers,2010). Experts and service bureaus are the main sources of
information for users that needs to select AM processes and materials.
2
Most users do not use constructive and formal evaluation procedures for the
suitability of the AM processes and materials for a given application (Ghazy and
Dalgarno,2009). In addition, the selection problem involves many alternatives and
many criteria that conflict with priorities which make the selection harder (Khrais
et al.,2011).
The right selection of a process is not only limited to a single process but to a
complete process chain from the material to the different finishing methods that
could be used. Some designers also do not include AM processes or materials
because of the lack of knowledge about these technologies and their capabilities.
Many designers stick to the conventional processes and materials that they know
which prevents them from discovering new opportunities in new developed
technologies. The development of AM technologies happened in less than twenty
five years which makes them relatively new. Selectors give the opportunity to
users to discover the new possibilities and new methods to produce product
quickly and effectively.
When a user or a designer has a part with specific requirements such as a specific
tensile strength or hardness or working temperature he needs to get an advice from
an unbiased person or system. The ad-hoc technique or the previously known
solutions could give users a selection that is not necessary the right one that
satisfies the criteria of users. The right choice for process chains, materials and
machines could be a critical element to the success of the developed product
whether it is a prototype visual or functional, a tool or die, and an end-use product.
1.3 Research Aim and Objectives
1.3.1 Aim
The aim of this research is to develop an additive manufacturing decision support

system (AMDSS) for inexperienced and average users that is capable of
evaluating and selecting the most appropriate additive layer manufacturing
processes to meet the requirements of a desired component.
3
1.3.2 Objectives
1. Exploring different additive manufacturing processes and examining the

most up to date information about new technologies in this field, and the
most widely used by industry.
2. Exploring different additive manufacturing materials, machines and

finishing options.
3. Using decision analysis tools to make selection among different layer

manufacturing technologies and processes.
4. Developing a system that is capable of assisting decision makers in:
I. Determining the feasibility of additive manufacturing processes for

a particular application.
II. Selecting the most suitable process.
5. Testing the developed system by applying verification and validation

methods.
1.4 Research Road Map
In order to develop an additive manufacturing decision support system, nine main

design steps were outlined. These steps were considered a research road map. The
researcher followed these steps to achieve the aim and the objectives of this
research. The nine steps are:
1- Analyzing available additive manufacturing (AM ) selector systems.
2- Identifying shortcomings of the available selector systems.
3- Identifying system target specifications.
4- Developing the system.
5- Verifying and validating the system.
6- Obtaining feedback from users.
7- Improving the system based on feedback.
4
8- Obtaining feedback for the improved system from users.
9- Launching the final version.
1.5 Thesis Outline
This thesis consists of eight chapters. The first chapter is this introduction. The
second chapter is the literature survey which covers different additive
manufacturing technologies, processes and materials. In addition, in this chapter
different AM selector systems are analyzed to indentify the strengths and
weaknesses of each system. Chapter two discusses the first and the second design
steps. The third chapter shows the development of the AMDSS. This chapter
covers the third and the fourth design steps. The fourth chapter discusses the
verification and validation of the developed AMDSS which is the fifth of the
design steps. Chapter five describes the sixth step which was obtaining feedback
from users. A questionnaire was designed to collect information and to analyze
the system. Chapter six shows the development of an improved system based on
feedback, which is the seventh step of the design steps. In addition, a second
questionnaire was designed to collect a further feedback and obtain users
opinions about the modified and updated AMDSS. This was the eighth step.
Finally, by launching the final version of the AMDSS the ninth and last step was
achieved. Furthermore, chapter seven discusses the results obtained and also
discusses the possible future forms of the developed AMDSS. Finally chapter
eight shows conclusions and discusses future work.
5
Chapter 2. Literature Survey
This chapter gives an overview about additive manufacturing technologies (AM).

It starts by discussing AM application areas then describes some popular AM
processes, process chains and materials. The previously developed selectors are
reviewed and their positives and negatives are discussed. Finally limitations of the
selectors are listed and the research gap is shown. The chapter discusses the first
and second of the design steps which are: 1- Analyzing available additive
manufacturing (AM) selector systems and 2- Identifying shortcomings of the
available selector systems.
2.1 Additive Manufacturing Overview
Additive Manufacturing (AM) is the process of producing 3D physical parts from

CAD directly by adding layer upon layer of material. The main concept is to add
material rather than to subtract material as in conventional machining processes
(Lan,2009).
The following steps describes a basic AM process (Cheah et al.,2005, Moeskopf

and Feenstra,2008) :
1- Creating a CAD model that represents the part that need to be built.
2- Converting the CAD model into STL format (standard triangulation

language) which is a language that describes the part using triangles.
3- Slicing the STL file into cross sectional 2D layers with a specified
thickness.
4- Generating support structures within the built part. This step is only for
the processes that need support structures such as stereolithography
(SLA) and fused deposition modeling (FDM).
5- Producing the part by adding the 2D cross section layer upon layer and
repeating the layer manufacturing until the full part is fabricated.
6- Post processing which includes cleaning, finishing the model and

removing support structures if any.
6
2.2 Additive Manufacturing Application Areas
AM applications can be described in three main types based on the end use for the
produced parts: rapid prototyping, rapid manufacturing and rapid tooling.
2.2.1 Rapid Prototyping
Rapid Prototyping (RP) is using AM technology to produce prototypes. The

prototypes can be used for design verification, functional testing, and marketing. It
is an important part of the product development cycle that shorten product
development times and lower cost (Armillotta,2006, Zhang and Liu,2009).
2.2.2 Rapid Manufacturing
Rapid Manufacturing (RM) is using AM technology to produce end product parts

that could be used by customers directly (Hague et al.,2003, Levy et al.,2003).
2.2.3 Rapid Tooling
Rapid tooling (RT) is using AM technology to produce a tool or a die. RT can be

divided into two main processes: direct rapid tooling and indirect rapid tooling. In
direct tooling the AM technology is used directly to produce the mould or die
using one of the AM machines. In indirect tooling AM technology is used to
produce a pattern then the pattern is converted to a mould.
Examples of rapid tooling are patterns for sand casting and patterns for investment
casting (Levy et al.,2003, Pal and Ravi,2007, Nagahanumaiah et al.,2008).
Furthermore, Dippenaar and Schreve (2012) reported the use of rapid tooling in
injection moulding, vacuum casting and electrical discharge machining.
Figure 2.1 shows different AM applications. Rapid prototyping (visual aids,

presentation models, functional models, fit and assembly) represents 26.2% of the
AM applications; rapid tooling (patterns for prototype tooling, patterns for metal
castings, and tooling components) represents 56.5% of the AM applications and
rapid manufacturing (direct part production) represents 15% of AM applications
(Wohlers,2010).
7
others, 2.3%
Direct part Visual aids,

Tooling production, 15.2%
components, 15.0% Presentation
3.7% models, 9.3%
Patterns for
metal
castings, 9.3% Functional
Patterns for models, 18.9%
prototype
tooling,
Fit and
13.2%
Assembly,
13.1%
Figure 2.1: Additive Manufacturing Different Applications (Wohlers,2010)
2.3 Additive Manufacturing Processes
In 2009, there were 35 AM system manufacturers that sold 6002 machines. The
number of AM systems sold increased each year. Figure 2.2 shows the number of
machines sold from 2005 to 2009 (Wohlers,2010).
Number of Machines Sold Each Year
7000 6002
6000 4996 5270
5000 4151
3526
4000
3000
2000
1000
0
2005 2006 2007 2008 2009
Figure 2.2: Number of AM Machines Sold Each Year (Wohlers,2010).
8
The system manufacturers produce different processes. There are too many
developed AM processes to cover them all here. Some of the most popular AM
technologies are therefore presented below.
2.3.1 Stereolithography (SLA)
Sterolithography was the first AM process introduced to the market by 3D

Systems in 1987 (Hopkinson et al.,2006). SLA is based on a platform that
contains vat containing a photopolymer liquid resin. An ultraviolet (UV) helium-
cadmium or argon ion laser is used to solidify the resin. The 3D CAD file is sliced
into a series of 2D cross section layers. Each sliced section is drawn individually
onto the surface of the resin. The laser solidifies the first layer, the platform is
lowered and the new layer of resin is scanned by the laser. The process is repeated
many times until the model is fully grown layer by layer in the vat. Some parts
need support structures, which is made by the same process, and at the same time
as the main parts. When the part is built, the base plate is then raised to lift the
part. The part is then drained, washed and the support structure broken away
(Pham and Dimov,2003). Figure 2.3 shows a schematic of the SLA process.
Figure 2.3: SLA Process (VG,2012).
The SLA process uses photopolymer liquid material and can also produce ceramic
shapes indirectly (green part) by the photopolymerization of a light sensitive
suspension such as dispersion of ceramic particles in a sensitive
monomer/oligomer resin (Chartier et al.,2012, Tomeckova and Halloran,2012).
9
After the fabrication of the green part, a thermal treatment is applied to the part at
low temperature for de-binding purpose. Next, the green part sintered with high
temperature to ensure the final properties of the ceramic piece (Badev et al.,2011).
The main advantage of the SLA process is that it is the superior process of all the
AM processes regarding accuracy (Melchels et al.,2010).
2.3.2 Fused Deposition Modeling (FDM)
Fused Deposition Modeling is an AM process commercialized by Stratasys in

1991 (Hopkinson et al.,2006). It extrudes and deposits thermoplastic filament
materials in layers on a platform. The head has two movable nozzles. The first
nozzle deposits molten material and the second nozzle deposits the support
material. The material is heated above its melting point so that it solidifies
immediately after the extrusion (Pham and Dimov,2003). The part that needs to be
built is sliced into layers, the two nozzles move together to form the first sliced
layer, then the platform is lowered and then the process is repeated until the
complete part is built by depositing a layer upon layer. Figure 2.4 shows a
schematic of the FDM process.
Figure 2.4: FDM Process (Custompart.net,2012).
10
The range of materials includes thermoplastics such as ABS and Polycarbonate,

biocompatible materials and heterogeneous materials (Choia et al.,2011). The
FDM process (sometimes called FDC fused deposition of ceramics when
producing ceramics) has been used to produce structural ceramics and
piezoelectric ceramics for incorporation into ceramic polymer directly and
indirectly (Grida and Evans,2003).
The advantages of the FDM process are: it is reliable, safe and simple fabrication
process, the range of materials includes many thermoplastic materials that have
wide applications, and many materials used are considered low cost materials
(Masood and Song,2004).
The major disadvantages of the FDM process are: parts fabricated with the
process lack the isotropy of mechanical properties, are very sensitive to process
parameters and the uneven heating and cooling cycles of the FDM process causes
residual stresses which lowers the strength of the parts fabricated (Sood et
al.,2010).
2.3.3 Selective Laser Sintering (SLS) / Selective Laser Melting (SLM)
Selective Laser Sintering
Selective laser sintering (SLS) and selective laser melting (SLM) are two similar
powder based processes. SLS produce polymers and can produce metals indirectly
while SLM produces metals directly.
Selective Laser Sintering (SLS) is an AM powder based process that was

originally developed by University of Texas at Austin in USA and
commercialized by a company called DTM (later acquired by 3D systems Inc)
(Hanemann et al.,2006). A CO2 laser beam fuses the material powder to form a
cross section or a slice of the desired product. The entire bed is lowered and after
that cylinders spread a new layer of the powder over the previously fused layer.
The laser beam repeats the fusing process again and forms another new layer.
Layer upon layer is deposited until a complete product is formed (Hon and
Gill,2003). The sintered material forms the product while the un-sintered material
remains in place for product support (Pham and Dimov,2003). Figure 2.5 shows a
schematic of the SLS process.
11
SLS can produce polymers directly such as polyamide, glass filled polyamide,
aluminum filled polyamide and carbon filled polyamide, while metals can be
produced indirectly. Indirect SLS consists of two stages by sintering a polymer
coated metal powder to produce a green part in the first stage. The second stage
involves heating the green part in an oven and infiltration with bronze or copper to
have a fully dense metal part (Levy et al.,2003, Ilyas et al.,2010). The SLS process
produces ceramic parts indirectly by sintering polymer-coated ceramic powders
but the parts are not fully dense (Tang et al.,2011).
Figure 2.5: SLS Process (VG,2012).
The advantages of the SLS process are: there is no need to have support structures
when building parts, so parts can be built freely in the building chamber which
increases productivity and lowers cost (Soe,2012) and also the parts produced are
characterized by having good mechanical properties.
Most of the materials used are highly crystalline or semi- crystalline polymers.
Amorphous materials do not have a specific melting point so they are heated until
12
reaching the glass transition temperature but the produced parts are weak and do
not have the good mechanical properties (Hopkinson et al.,2006).
Selective Laser Melting
Selective laser melting (SLM) is an AM process that is first commercialized by

MTT Technologies Group of Stone since 2003. SLM have the same building
technique as the SLS process except that in SLM the laser melts the powder and
does not sinter it. The laser melts metal powder on a powder bed and produce
parts layer upon layer. The range of materials include stainless steel, aluminum
alloys, titanium and tool steel (Wohlers,2010). Fully dense parts can be produced
directly by SLM (selective laser melting process) (Yves-Christian et al.,2010).The
almost full density gives an advantage of yielding bulk material properties (Yasa
et al.,2011).
2.3.4 3D Printing (3DP)
3D Printing (3DPTM) is a powder based AM process developed at MIT. The

process produces physical parts by spreading powder over the surface of a powder
bed. A binder material is used to join layers together. There is a piston that lowers
the powder bed after a new layer is formed so a new layer of powder can be
spread as shown in Figure 2.6Figure 2.6: 3DP Process (Custompart.net,2012).
Repeating this process produces the 3D physical part (MIT,2012).
Figure 2.6: 3DP Process (Custompart.net,2012).
13
The 3D printing can produce porous ceramics shapes. The head eject an organic
binder on ceramic powder (Parasad et al.,2006, Derby,2011). The SLA, FDM and
SLS are characterized by definition around 150 m while 3D printing can reach
50 m (Su et al.,2008b, Noguera et al.,2005). There are some disadvantages of the
ceramics parts produced by AM processes such as: 1- anisotropic shrinkage due to
residual stresses arisen from polymer binder drying in printing methods or high
shear field in extrusion based methods, 2- poor surface quality, 3- poor surface
finish, 4- poor dimensional accuracy and 5- the stair-steps effect decreases the
ceramics strength and toughness (Su et al.,2008a).
2.3.5 Ink-Jet / Multi-jet Modeling (MJM)
The ink-jet method patented by Solidscape and the multi-jet modelling (MJM)
patented by 3D systems, are very similar AM processes with little differences.
Ink-Jet
In the ink-jet process the jetting head has three nozzles. The first nozzle jets the
thermoplastic material while the second and the third nozzles jet the wax support
material. Each layer formed is milled to a specific layer thickness. The System
uses drop on demand (DOD) technology and high precision milling for each layer
(Hanemann et al.,2006). Figure 2.7 shows the ink-jet process.
Figure 2.7: Ink-jet Process (Hanemann et al.,2006).
The advantages of the ink-jet process are: it is able to deposit materials with
different chemical and physical properties, it has low operational cost (Khalate et
al.,2011), position accuracy is high and production reproducibility is also high
(Khalate et al.,2011).
14
The disadvantages of the ink-jet process is that it is low speed process, surface
finish of the parts are not very good , and jetting of high temperature materials is
difficult to achieve (Fathi et al.,2012).
Multi-Jet-Modeling (MJM)
In the multi-jet modeling (MJM) there is a head with multiple spray nozzles and
an ultra violet (UV) lamp. The nozzles deposit tiny droplets of hot photopolymer
liquid (thermopolymer) and the UV polymerizes the liquid. The process of
depositing the liquid and the polymerization forms the first layer, and is repeated
to form layer upon layer until the final part is made (Zemnick et al.,2007). Figure
2.8 shows the multi-jet modeling process.
Figure 2.8: MJM Process (Hanemann et al.,2006).
2.3.6 Laser Engineering Net Shape (LENS)
Laser Engineering Net Shaping is an AM process that deposits powder and fuses
it simultaneously (Pham and Dimov,2003). The LENS process was developed by
Sandia National Laboratories and commercialized by Optomec Inc since 1997
(Hanemann et al.,2006). Figure 2.9 shows a schematic of the LENS process.
The head of the LENS process consists of a laser beam and two nozzles that feed
coaxial powder. The process takes place in an atmospherically controlled
environment usually of argon gas. First the high power laser (750-1000 w) focuses
on a metal part to create a molten pool. Next, the powdered metal is supplied from
the nozzles to add more volume. The table that the part is on it moves in the X-Y
direction to create the first layer, while the head moves up and down in the Z
direction to build the part height. The process is repeated to build a layer upon
layer until the complete part is built (Amano and Rohatgi,2011).
15
Figure 2.9: LENS Process (Zhao et al.,2009).
The LENS can process various metals such as nickel based alloys, stainless steels,
tool steel, cobalt-chrome and titanium alloys. One of the advantages of the LENS
process is that it can produce fully dense materials. (Wohlers,2010, Zhao et
al.,2009).
2.3.7 Electron Beam Melting (EBM)
Electron Beam Melting is an AM process available from Arcam AB (Sweden)

since about 2005 and its main concept is that electrons are generated in a gun,
accelerated and focused using electromagnetic lens and electro magnetically
scanned by an embedded CAD program (Murr et al.,2012).
First, a powder metal is spread over a platform. The electron beam melts the metal
powder and builds the first layer of the part. Next, the platform is lowered by one
layer thickness and a new layer of metal powder is spread. This process is
repeated until the complete part is built by adding a layer upon layer over the
platform (Lu et al.,2009). Figure 2.10 shows a schematic of the EBM process.
16
Figure 2.10: EBM Process (Lu et al.,2009).
The range of materials used by EBM process include: titanium, cobalt-chrome,

titanium aluminide, inconel (625 & 718), stainless steel (e.g. 17-4), tool steel (e.g.
H13), aluminum (e.g. 6061), hard metals (e.g. NiWC), copper (e.g. GRCop-84),
beryllium (e.g. AlBeMet), amorphous metals, niobium and invar (Arcam,2012b).
The advantages of the EBM process is that the parts built have stable chemical
composition because of the vacuum environment, excellent mechanical and
physical properties because of the even temperature distribution within the part. In
addition, EBM has relatively high productivity because of the high deposition rate
(Arcam,2012a).
2.4 Additive Manufacturing Process Chains
2.4.1 Rapid Tooling Process Chains
In the literature, the AM process chain terminology is used mostly to express

direct or indirect rapid tooling such as using SLA or any other AM process to
create patterns for reaction injection molding, vacuum casting or investment
casting. The following are some examples of rapid tooling process chains used in
the industry:
17
i. Stereolithography + Reaction Injection Molding (SLA + RIM)
Reaction Injection Molding (RIM) is a process in which two or more low

viscosity, highly reactive chemicals are mixed and injected together into a mould
to produce polyurethane objects (Dias et al.,2012). The SLA process produces the
mould used in the RIM process. Figure 2.11 shows a schematic of the RIM
process.
Chemical Tank 1 Chemical Tank 2
Mixer
Mould Fabricated
using AM
Technology
Figure 2.11: RIM Process.
The two liquid chemicals are stored in tanks. They are fed to a mix head where
they are mixed together. Next, the two liquids flow from the mix head to the
mould under atmospheric pressure and the chemical reaction begins. The
following steps (Dias et al.,2012) describe the RIM process: 1- mixing the two
low viscous liquids into a mix head, 2- filling the mould, 3- Curing and
solidification of the mixture, 4- Part extraction, and 5- Post-curing to complete the
solidification and improve mechanical properties of the produced part.
ii. Stereolithography + Silicone Tooling + Vacuum Casting

(SLA + ST + VC)
The SLA + ST + VC process chain consists of three main steps: 1- the SLA
process creates a pattern of resin material, 2- silicone is cast under vacuum around
the SLA pattern to create a silicone mould and 3- the silicone mould is cured in an
oven at 40C for four to six hours. After curing, the mould is cut to create parting
lines and the pattern is removed.
18
The silicone mould is ready for the vacuum casting process. The silicone tool is
cheap and can produce complex shapes, fine patterns and deep groove and without
drafting angle (Zuo,2012). Vacuum casting is used to produce parts made of
metals (non-ferrous), wax and plastics (Thian et al.,2008). The process chain is
used for small series from 10 to 20 parts (Chua et al.,1998, Materialise,2012).
Any AM process such as SLS or FDM could replace the first step but it is more
common to use SLA process in industry.
iii. Stereolithography + Investment Casting (SLA + IC)
The investment casting process is an industrial process considered as an economic

approach for mass production of metal parts with complex shapes (Rahmati et
al.,2009). The investment casting process is a very old industrial and the term
investment refers to the slurry or investments used to produce the mould (Jones
and Yuan,2003).
The traditional steps in investment casting are: 1- producing moulds, 2- producing

wax patterns, 3- producing ceramic shell, 4- pattern removal 5- pre-heating and
firing, 6- casting and knockout.
The SLA process can be used to produce patterns made of epoxy resin instead of
the wax patterns in step number 2. There is a build style on SLA machines called
QuickCast which fabricates the SLA patterns with a quasi-hollow structure. The
main concept is that the hollow structures would soften at lower temperatures and
collapses inward upon it-self before critical stress levels are developed. The
QuickCast build style solves the problem of the difference of the coefficients of
thermal expansions between the SLA patterns and the ceramic shell. The benefits
is to reduce tooling cost and tooling time (Cheah et al.,2005).
19
Option 1: SLA produce

resin pattern
Option 2: 3DP produce

wax patterns
Traditional Investment Casting Investment Casting with AM technologies
Figure 2.12: Investment Casting Process (adapted from Jones and Yuan,2003).
There are also other AM processes that could fabricate either the wax patterns or
non-wax patterns for investment casting processes such as FDM, SLS and 3DP
but SLA is the most successful because of its good surface finish and good
dimensional accuracy (Cheah et al.,2005). Figure 2.12 shows a schematic of the
traditional investment casting process and the use of AM technologies with the
investment casting.
2.4.2 General AM Process chains
Although in the literature AM process chain terminology is mostly used for rapid
tooling this study looks at the AM process chain terminology from a broader
perspective and considers that AM process chains should have a more generalized
definition: an AM process chain is any manufacturing process route that involves
at least one additive manufacturing process in it. Any sequence of processes that
produce a part or a tool or a prototype using an AM process and other
conventional or non conventional processes, is considered an AM process chain.
Figure 2.13 shows different possible AM process chains.
The first reason to justify the definition of the general AM process chain is that
most of the time AM processes are not used alone to produce parts especially
when the purpose is to have a functional prototype or an end use product.
Secondary processes are needed for finishing like for example sanding,
20
machining, grinding, electroplating, polishing, sandblasting, texturing and/or

combinations of them. Leong et al. (1998) have discussed the secondary process
for finishing of jewellery models built using SLA such as: tumbling, filling,
sanding, buffing, tapping, sawing, trimming, and grinding. Lamikiz (2007) have
used laser polishing as a secondary process to improve the surface finish of a part
built by an SLS process. Galantucci et al. (2009) have made chemical post
treatment to improve the surface finish of ABS parts made on an FDM system.
AM Process Finishing Methods
Conventional
AM Process Finishing Methods
Processes
Non
AM Process Conventional Finishing Methods
Processes
AM Process Hard Tool Finishing Methods
AM Process Soft Tool Hard Tool Finishing Methods
Figure 2.13: Possible AM Process Chains
The second reason is that for some applications it is better in terms of cost or time
to use additive manufacturing and conventional manufacturing to manufacture one
part. For example, Das et al. (1999) have fabricated a titanium sidewinder missile
guidance section housing (an end-use product) using SLS/HIP (selective laser
sintering and hot isostatic pressing). Mognol et al. (2007) have proposed an
approach based on topological analysis of the tool and the manufacturability
possibilities of the involved processes. A decision is made to manufacture
different components (in the same tool) using high speed machining (HSM) or
electrical discharge machining (EDM) or direct metal laser sintering (DMLS).
Although, this method is proposed for a tool there is no reason that prevents using
the same method with any end-user part or a prototype. Ilyas et al. (2010) have
used an AM process chain to produce an injection mould tool. The chain consists
of using 1- indirect SLS process to produce the mould, 2- high speed machining
21
(HSM), 3- electron discharge machining (EDM) for primary finishing and

secondary finishing is made by grinding and polishing.
2.5 Additive Manufacturing Selection Systems
After the development of the first rapid prototyping system SLA many other
systems appeared. The advances were not limited to new technologies but also
new materials. The selection of the most appropriate systems became a problem
that many researchers have tackled by developing models and decision support
systems.
At the beginning researchers considered this from the perspective of rapid

prototyping and this later extended to rapid tooling and rapid manufacturing.
Looking into literature the work done can be divided into three main groups: 1-
Rapid Prototyping (RP) Selectors 2- Rapid Manufacturing (RM) Selectors and 3-
Rapid Tooling (RT) Selectors.
2.5.1 Rapid Prototyping Selectors
The first attempt to make rapid prototyping selection was a program developed
by Hornberger (1993) at Santa Clara University. The program provided
educational information about rapid prototyping and guided users to select RP
processes.
Campbell and Bernie (1996) developed a relational database that represented the
different capabilities of various RP systems. They developed a graphical user
interface that users could use to search databases on build envelope, material
properties and feature tolerances. After the search is done users can query the
database for actual capabilities of the RP system. This system is a search tool
rather than a selection system.
Muller et al. (1996) developed a rapid prototyping system selector that helped
users to find the best RP system to manufacture physical prototypes. The system
was based on relational databases of available machines and materials. The
developed system chose the best combinations of machines and materials to make
a prototype rather than selecting the most suitable RP process based on selection
criteria. They used the benefit value analysis method to evaluate machine-
material combination.
22
Phillipson (1997) developed a decision support system called RP advisor. He used

quality function deployment to identify the most important features of the RP
advisor. Three selection criteria were used: time, cost and quality. The system
considered six RP machines and the multi criteria optimization theory was used to
find the optimal selection. The system did not consider materials. The developer
made two interviews with expert users to check the system effectiveness.
Muller (1999) developed a method based on a check list to evaluate an RP process

chain within a European Project called RAPTEC. He has used qualitative criteria
and the RP processes were compared in pairs. This was just a method for
comparison and not a complete decision system.
XU et al. (2000) compared four RP systems: SLA, SLS, FDM and LOM, through
a benchmark part. They developed three generic models for surface roughness,
building time and building cost. The work was only limited to the mentioned four
technologies and was a benchmark study rather than a selection model. These
experiments became obsolete because of the advance in the technologies and the
materials.
Masood and A.Soo (2002) developed a rule-based expert system called IRIS for
the selection of RP machines. The system incorporated 39 RP systems available
from 21 manufacturers. When developing the system two types of questionnaire
were developed. The first questionnaire was for the vendors to collect information
about system features, prices and applications. From 21 vendors the responses was
about 70% around 15 vendors. The second questionnaire was for systems users
asking about selection criteria, applications and the performance of the machines.
From 136 users the responses was about 13% around 18 users. An updated
version of the IRIS was developed by Masood and Al-Alawi (2002). The new
version included 57 machines from 22 vendors. In the second version, users were
able to update the databases using MS Access. The developers commented that
the developed system needs to be always updated. The developed system was a
system that aid users to select the best machines to buy rather than to select a
process to suit specific application. It did not consider material selection.
Mahesh et al. (2003) proposed applying the six sigma approach towards
benchmarking of RP processes by assessing process capability against its
potential. They mention three types of benchmarks: geometrical, mechanical and
23
process benchmarks. The authors suggested that benchmarks can be stored in a

database that a decision support system could use it. They developed a case study
on DMLS. Finally, the paper outlined a basic web-based decision support system
that included the RP benchmarking and standardization database. This work was
only for DMLS process.
Byun and Lee (2005) developed a model that help users in the selection of rapid
prototyping process using a modified TOPSIS ranking method (Technique for
Order Preference by Similarity to Ideal Solution). The problem was considered as
multiple attribute decision making. The model considered six criteria which were
dimensional accuracy, surface finish, tensile strength, elongation, part cost and
build time. A questionnaire was designed to ask different users about the most
important selection criteria. Fuzzy numbers were used for part cost and build time.
A case study reported a comparison of six RP systems. The developed model is a
model that selects machines only and not processes.
Lan et al. (2005) developed new method to select the most appropriate rapid
prototyping system by integrating expert system and fuzzy synthetic evaluation.
An expert system was used for selection and fuzzy method was used for ranking.
The developed system consists of four modules which are: 1- knowledge based
expert system, 2- fuzzy synthetic evaluation, 3- databases, 4- user interface and
expert interface. This paper focused on the fuzzy syntactic evaluation and did not
explain the four modules of the whole system. The ranking of the alternatives (six
alternatives were considered: SLA, LOM, FDM, SLS, 3DP, and SGC) was made
using fuzzy AHP (analytical hierarchal process). The criteria considered were
dimension accuracy, surface roughness, maximum dimension, part complexity,
mechanical strength, heat resistance, running cost, post-processing cost, material
cost, equipment cost, scan speed, overhead time and post processing time. The
system selects only rapid prototyping processes. One of the limitations of this
model is that it did not include material selection. In addition, AHP method is
difficult to apply if the alternatives become very large (more than seven) or the
alternatives are heterogeneous (Giner-Santonja et al.,2012) .
Rao and Padmanabhan (2007) developed an RP process selection methodology

using graph theory and a matrix approach. The method gave each RP process an
index obtained from an RP process function which is calculated based on a RP
24
process attributes digraph. A digraph is a set of nodes and a set of directed arrows
where the nodes represent the RP attributes and the directed arrows represent the
importance of these attributes. The arrows are directed form the more important
attribute to the less important attribute. The ranking was done according to the
calculated index. The considered attributes were: dimensional accuracy, surface
finish, type of material, material properties, part cost, build envelope, range of
layer thickness, part size, feature type, feature size, ease of use, and environmental
affinity.
The authors used the case study reported by Byun and Lee (2005) who developed
the modified TOPSIS method. A comparison between the two models showed that
the graph theory and matrix approach is better than the modified TOPSIS because
it can enable more critical analysis since any number of attributes quantitative or
qualitative can be taken into consideration. The limitations of this method are that
the selection is for machines and not processes. In addition it is only a decision
model, cannot be updatable and is complex for novice users.
2.5.2 Rapid Manufacturing Selectors
Bernard (1999) and Bernard et al. (2003) discussed the development of an expert
system using CAPP expert system called ACPIR (aided choice for rapid
industrialization processes). The developed expert system was not only for layer
manufacturing machine selection but it was designed to be more general looking
at the product development integrating: CAD, reverse engineering and indirect
methods for metallic and plastic parts manufacturing. The developed system
contained two types of reasoning: 1- case based reasoning which uses old case
studies stored in the system to predict similar solution to similar problems, 2-
bottom-up generation of a process where the system suggested solution for the
user based on the knowledge using the expert system rules. The case based
reasoning is used first and if a solution is not found the bottom-up approach is
used. The limitations of this system are that users cannot update the system by
themselves because the logic of the expert system would have to be changed. In
addition, the case studies need to be updated so the case based reasoning would
remain useful. Furthermore, the system included only processes and topology of
materials. The material selection was not covered in a useful manner.
25
Bernard et al. (2003) also thought that managing the knowledge inside the
database and developing a semi-automatic system for the selection is the main
challenge for researchers developing decision support system for the selection of
new technologies and materials in AM field.
Swedish Researchers at Industrial Research and Development Corporation, IVF

(2005) also developed an online selector called RP selector. The selector asked
three main questions about material/function, quantity and end-user requirements.
The program was intended to aid designers and novel additive manufacturing
users in:
Giving fast, direct access to a great amount of information about practical

AM based methods.
An easy-to-learn, easy-to-use guide for novel users.
A base for user-oriented training.
A support for common understanding at internal and external discussions
A communication link for information about methods, materials and

suppliers of products and services.
The advantages of this system are: it is an online system, it contains good

information about some of the AM processes, it considers AM process chains and
not only AM machines, and it considers rapid prototyping, rapid tooling and rapid
manufacturing. On the other hand, the limitations of the system are: it has a
limited material database, users cannot update the system by any means and it
does not rank any process or material but it gives the user different choices that
the user has to trade off between.
An RM selector was also developed at Georgia Institute of Technology in 2005.

The main concepts around the selector can be found in Gibson et al. (2010). It is a
preliminary expert system that has been developed using Matlab environment tool
that selects feasible processes and machines, it includes qualitative assessments
based on several questions asking about shape similarity across production
volume, part geometric complexity, part consolidation and delivery time, it
26
separates feasible and unfeasible machines based on material, surface finish and
accuracy requirements. In addition, build time and cost can be estimated.
The advantages of the software are that: it enables users to explore AM

technologies, identify candidate material, explore build times, explore the life
cycles of AM products, select appropriate AM technology, explore case studies,
and update machine database. In addition, it supports quotation and capital
investment decisions. The main limitations of the RM selector are: it focuses on
machine selection rather than process chain selection and no appropriate material
selection is provided, only material exploration.
Mahesh et al. (2005) developed a system which is called IRPDMS (integrated

rapid prototyping decision making system) based on fuzzy decision making and
benchmarking for selecting the most appropriate rapid prototyping and
manufacturing systems and they implemented it as a web page using fuzzy if-then
rules. The developed web page is not available for use, only the design of the
webpage is provided. The developed system interacts with a previously developed
benchmark database in Wong et al. (2002) and its methodology consists of three
stages: 1- representation of the decision problem, 2- fuzzy set evaluation of goals
and constraints, 3- selection of the best solution. The limitations of the system are:
it considered only five RP process: SLA, SLS, FDM, LOM and DLS, the system
is based on benchmarking of a single part and it cannot be used as a generic
approach, no material selection is mentioned and updatability is not considered.
Smith and Rennie (2008) developed an additive layer manufacturing selector tool
for direct manufacturing called RM selector. The system consists of relational
databases that contain information about AM machines, materials, technologies
and the characteristics of parts created with a combination of machines and
materials. The RM selector is a web based program that searches the databases.
The limitation of the RM selector is that it is mainly a search tool rather than a
process or material selection. In addition, no ranking is available.
Munguia et al. (2010) has developed an AI based system called RMADS (Rapid
manufacturing advice system) which is a rapid prototyping and rapid
manufacturing selection system that integrates three modules: expert system,
fuzzy interface and databases.
27
There are three databases: process parameters, machines and materials. Users have
the ability to use the material database by itself separately. Finally the fuzzy
interface is used for aggregation and ranking. In his PhD thesis Munguia (2009)
presented a parametric model for cost estimation and a neural network model was
developed for cost estimation based on building time. One of the advantages of
the system is that it can select processes and materials in a user friendly manner.
Despite the system versatility it does not come without limitations: it selects
process only and not process chains, updating cannot be done without changing
the programming logic and users cannot make updates by themselves, and the
parametric cost model is an estimate at best.
In July 2010 a new project funded within the FP7 framework called KARMA was
launched (KARMA,2010).
According to Petrovic et al. (2011) the project aimed to establish an online KBE
system capable of :
Guiding users to choose the most appropriate technology, material and

build scenarios.
Offering a full mechanical and thermal characterization with emphasis on

part anisotropy.
Doing efficient process planning.
The project is ongoing and the project team does its own mechanical property
experiments instead of depending on vendors information. For 4 technologies and
8 materials 1216 tests have been made until 2011. The project incorporates only
four types of technology which are: SLM, SLS, SLA and EBM. The KBE
architecture is able to include more technologies.
2.5.3 Rapid Tooling Selectors
Durr and Kaschka (1998) have developed a method for examining and selecting
conventional technologies, RP technologies and hybrid technologies. The
selection criteria used are: type of part, material, purpose of the product,
geometrical complexity, size of the part, dimensional accuracy, surface quality,
and number of pieces. They have used a two step methodology:
28
1- recommendation of the most suitable process. 2- ranking according to the

weighted values of quality, time, cost and ecology. One advantage of this
methodology is that it considers the additive manufacturing process chains and not
the technologies only. The limitations to this method are: there is no material
selection and there is no possibility of updating.
Hermann and Allen (1999) have developed a rapid tooling test bed to select he
most feasible process and material combination for the development of injection
molding. Pairs of alternatives from each selection problem (process selection and
material selection) are evaluated simultaneously. The method used is called a
coupled selection-selection decision support problem. The study focused on the
creation of injection molds with SLA.
The key criteria used for process selection is cost, detail capability, accuracy and
build time. The key criteria used for material selection are: availability, density,
elastic modulus, tensile strength, elongation and heat deflection temperature. One
advantage of this test bed is the selection and the ranking of the process and
material together.
Bibb (1999) developed a rapid prototyping selection system using an expert

system that is composed of knowledge base and inference engine. The system
selects rapid prototyping and rapid tooling technologies. It contained two types of
rules: decision rules and calculation rules. The criteria used were required
accuracy and minimum wall thickness. The decision rules selects the most suitable
RP system and the secondary tooling while the decision rules estimated build time
and build cost. The system has been verified after it has been developed by trials
of seven novice users and one expert user giving their opinions regarding ease of
use, usefulness of the results, and validity of the results. The system does not help
in material selection. The knowledge base can be updated but not by the user
because the update will need to update the logic of the expert system.
Kascha and Auerbach (2000) developed a software system called Protool based on
the methodology developed by Durr and Kaschka (1998). An expert system
containing 180 rules was developed to select process chains. The expert system
takes the data from technological databases that contain process chains, process
steps, materials and machines.
29
The databases could be updated but user cannot modify it without changing the
logic of the expert system.
Hanumaiah et al. (2006) developed a QFD-AHP methodology to select the rapid

hard tooling process. The methodology consists of three phases: 1- prioritizing the
tool requirements using AHP, 2- the results of the AHP are used as weights in
QFD for selecting the most suitable tooling process, and 3- QFD is used again to
identify critical process parameters. The methodology was applied to two
industrial examples for validation. The limitation of this method is that it did not
include material selection. In addition, no software was developed for selection.
Pal et al. (2007) developed an integrated methodology combining quality function

deployment method (QFD) and analytical network process (ANP) method to
determine the engineering requirements of a cast part based on the customer needs
for selecting the rapid tooling processes. A case study was represented using five
RP processes which are: thermo jet printing, FDM, LOM, SLA and SLA
QuickCast method. The limitations to this methodology are that it did not consider
material selection and depending on the pair wise comparison which will be
complicated and frustrating to users if the number of the processes increases.
Pal and B.Ravi (2007) developed a database containing RT process capabilities

covering 20 different rapid tooling routes, both direct and indirect. The selection
of the RT route was made based on the methods developed by Pal et al. (2007). A
compatibility index had been calculated and the ranking was done according to
this index. The RT method with highest index is the most suitable RT process. In
addition, process planning was carried using case based reasoning to find a similar
process plan. There is no indication that users can update the databases. In
addition the selector is limited to sand casting and investment casting only.
Armillota (2007) developed an AHP decision model (analytical hierarchical

process) within a computer based tool. The AHP method is a method used for
ranking by a score resulting from the composition of priorities at different levels.
The model selects rapid prototyping and rapid tooling techniques. The model
considered sixteen alternatives and eleven selection attributes which are:
compliance to office environment, high build speed, low setup operations,
availability of functional or high strength material, good dimensional and
geometrical accuracy, good surface finish, economical processing, low cost
30
materials, low cost system usage and low setup cost. One of the limitations of this
model is that updating the model by adding more processes requires that users
perform pair wise comparison between all the alternatives again. Furthermore, no
material selection is reported.
Hanumaiah et al. (2007) developed a fuzzy-AHP methodology to evaluate the

manufacturability of the rapid tooling processes. It is based on three aspects: 1-
mould feature manufacturability, 2- secondary elements compatibility and 3- cost
effectiveness. A feature based database has been developed containing geometry
compatibility, dimension accuracy and surface finish. Selection between six
tooling processes was used for methodology validation. These are conventional
tooling, SLA, SLS, DMLS, SLA+ Investment casting and spray metal tool. The
database contained only information about these six tooling processes. No
software development is reported in this paper.
Hanumaiah et al. (2008) developed a system for selection of rapid tooling

processes and manufacturability evaluation for injection moulding. The QFD
method was used for process capability mapping against asset of tooling
requirements. The tooling requirements were prioritized using AHP. The
manufacturability evaluation was carried using fuzzy-AHP found in Hanumaiah et
al. (2007). Finally, RT cost was calculated and compared to conventional moulds.
The system is only limited to injection moulds parts and there is no material
selection.
Khrais et al. (2011) have used fuzzy if-then rules to select rapid prototyping
techniques. They have named four methods: 1- Prototyping, 2- Investment
casting, 3- Sand Casting, 4- Plastic moulding. The selection method did not deal
with rapid manufacturing or rapid tooling. The selection criteria were in two
groups: 1- static (does not depend on a particular application) and 2- dynamic
(varies from an application to another). The advantage of this model is that fuzzy
rules can deal with qualitative and quantitative variables. The disadvantage is that
it cannot be updated without changing the fuzzy rules. Table 2-1 shows a
summary of selector systems that is capable of selecting RP, RM and RT.
31
Selection
Selection
Selection
Machine
Material
Process
RM
RT
RP
Year System Developers
1993 RP program Hornberger

Database of RP Campbell &
1996
Systems Bernie
1996 RP system selector Muller et al.
1997 RP Advisor Phillipson
1999 ACPIR Bernard et al.
Rapid Tooling
1999 Herman & Allen
Testbed
Rapid Prototyping
1999 Design Advice Bibb et al.
System
Kascha &
2000 Protool
Auerbach
2002 IRIS Masood & Soo
Decision Support
2005 System for RP Lan et al.
Selection
2005 RP Selector IVF (Sweden)
2005 RM Selector Georgia Institute

of Technology
2005 IRPDMS Mahesh et al.
2007 RT databases Pal & Ravi
2007 Computer Aided
Hanumaiah et al.
RT Selection
2008 RM Selector Smith & Rennie
2010 RMADS Munguia et al.
2010 KARMA FP7 Project
Table 2-1: Different Attempts for Additive Manufacturing Selection.
32
2.6 Limitations of the Developed Selectors
The following is the summary of the limitations of the developed selectors:
In the literature as seen in Table 2-1 most of the developed selectors are
for a specific type like RP selection or RT selection or RM selection. Some
researchers developed selectors that are capable of selecting two types like
RP and RT selectors or RP and RM Selectors. There are only two selectors
that could be classified as general selectors that are capable of selecting the
three types together: 1- ACPIR expert system developed by Bernard
(1999) and Bernard et al.(2003) and 2- The RP selector developed by IVF
in Sweden (2005). The ACPIR was a general product development not
only for AM manufacturing but it lacks material selection and updatability
and the RP selector is like a general guide more than a selection tool,
which does not rate the alternatives.
Although many developed systems have used relational databases that

could be updated and modified but updating the databases would require
changes in the programming logic of the system.
There is no integrated selector system that includes of the process chains,

materials, finishing methods and machines all together.
No developed systems give reasons or explanation when the system cannot

give users an advice about their products.
None of the selector systems had reported the involvement of the users in
the development stages of the system. Some researchers took the opinion
of expert users after the development for validation purpose only.
Despite the selection of a process affects the selection of material and vice
versa, many developed systems rank the processes or the materials alone.
2.7 Research Opportunities
The need for a study that that assesses AM processes in their current state and
their potential to fabricate end user products is presented by Laoui (2010). He
presented an approach for developing knowledge based environment for rapid
manufacturing technology and emphasized that the knowledge base should
33
contain process capabilities, machines, heuristic knowledge and a database that

contains all the AM materials.
The limitations of the previously developed selectors open the door for research
opportunities based on non fulfilled needs of users. The following is a summary of
the key points that should be covered by new selectors to fulfill the needs of users:
There is a need for a general additive manufacturing system that could

help users to select rapid prototyping, rapid tooling and rapid
manufacturing. The focus of the AM selector should be on the product to
be produced not the technology and not the process. This product could be
one part or several parts or a tool. The AM selector should take the
requirements of a user for a given part and the required quantities of this
part and give the user a guided selection for his application.
There is a need to have system that could be updated by users without

changing the programming logic. Most of the systems are obsolete after a
while because they can only be updated by their developers.
When a user has a part that he needs to manufacture he needs to look for
materials, manufacturing processes, finishing methods and machines.
There is a need to have an integrated system that helps users in the
selection of the materials, process chains, finishing methods and machines.
In addition, many developed systems select technologies or processes.
There is a need to select AM process chains.
Giving reasons to users if the system fails to give them an appropriate

advice to manufacture their parts, consequently users can compromise and
trade off between their objectives by relaxing some of the constraints to
get the part made.
Involving users by obtaining their opinions and feedback will help to

create a robust system that represents the point view of users.
There is a need for a system that is capable of ranking the processes and
materials together so users can compare the rank of different materials for
the same process or vice versa.
34
Chapter 3. Additive Manufacturing Decision Support System
(AMDSS) Initial Development
This chapter presents the initial development of the Additive Manufacturing

Decision Support System (AMDSS). First, it starts with the system design
describing the basic steps used to develop the system and discussing the different
proposed system architectures and explaining which architecture is selected.
Secondly, it shows a general overview that outlines the framework of the selected
architecture. This chapter explains the third and fourth steps of the AMDSS
development.
3.1 Identifying System Target Specifications
The target specifications are:
1. Selecting the possible AM processes chains, materials, finishing

methods and machines.
2. Giving reasons if the system could not give a solution and available
AM processes or materials in the database cannot satisfy the users
requirements.
3. Ranking processes and materials according to the users criteria
4. Ability of being updated. Users can update processes, materials,

finishing methods and machines without the need to make changes in
the AMDSS logic or programming.
5. Ability of being customized. The user can change and edit in the
database the processes, materials, finishing methods and machines that
he needs to show in the final ranking.
6. Being simple, usable, customer focused and interactive helping the

user to understand the decision process step by step.
7. Being product focused, means that the selection is based on the final
product requirements that the customer need to use at the end.
35
Chapter 3. Additive Manufacturing Decision Support System (AMDSS) Initial Development
3.2 Developing the System
When developing the AMDSS different system architectures were considered.

These architectures are explained in detail below.
1- Neural Network (NN)

Neural network (NN) architecture works on the principle of developing certain
relationships between some inputs and their results as outputs. The input-output
relationships maybe defined in the form of rules or patterns. The user trains the
network and the NN learns the relationship between the inputs and the outputs.
Based on the learned patterns, the NN can predict an output for a new input. The
relationships between inputs and outputs remain hidden. For this reason neural
network is considered a black box model (Vouk et al.,2011).
The neural network consists of neurons organized in layers as seen in Figure 3.1.
The input Layer contains the input variables, the output layer contains the output
variable (s) and the hidden layer is the place where the calculations are done. NN
can have more than one hidden layer (Lu et al.,2012).
Figure 3.1: Schematic Diagram of Neural Network (Lu et al.,2012)
The first idea for the system development was to develop a neural network (NN)
by using different case studies with known inputs and outputs to train the network
until there was a small error percentage. The problem with this architecture is that
the neural network is a black box, so the end user of the system cannot know or
understand why a specific selection is made. In addition, to update the neural
network the developer re-trains it by using new inputs and outputs.
36
This method makes the end user incapable of updating it without specific
knowledge of the neural network logic.
2- Fuzzy Logic (FL)
The Fuzzy set theory was a concept developed by Lotfi Zadeh (Zadeh,1965) to
look at the data as a partial set of membership rather than crisp values or non
membership. Fuzzy logic (FL) is a problem solving methodology that uses vague
information. The main concept is that the human reasoning uses knowledge that
does not conform to precise boundaries. FL is determined by using linguistic
variables like good/bad, low/high. The membership function is a graphical
representation of the magnitude of participation of each input in fuzzy logic
(Olugu and Wong,2012).
The second idea for the system development was to develop a fuzzy logic (FL)
system that contains all the fuzzy rules so the user can have an output when
entering different fuzzy inputs for the different variables. The fuzzy rules use
fuzzy inputs and fuzzy outputs to describe knowledge or information. The first
problem is that some inputs are not fuzzy, so representing these inputs with fuzzy
variables is not realistic. The second problem is that the fuzzy rules cannot be
updated by the user; it has to be updated by the developer by defining new fuzzy
rules for new AM systems and materials. Consequently, the system will become
obsolete after a while.
3- Expert System (ES)
Expert system (ES) is an interactive program that contains knowledge acquired

from experts and helps users using the previously stored knowledge to solve
complex problems. An Expert system is considered as one type of knowledge
based systems (KBS). Information in the expert system is represented in the form
of If-Then rules called production rules. The expert system applies the stored rules
to a given data to reach a certain conclusion (Chen et al.,2012, Sahin et al.,2012).
There are two types of reasoning in expert systems forward chaining and
backward chaining. Forward chaining starts with data and ends with a goal or a
conclusion while backward chaining starts with a goal and moves backward until
finding suitable data for the desired goal.
37
The third idea for the system development was to develop an expert system that
contains expert data regarding the selection process. The main problem
encountered was that the expert system can only deal with the cases already it has
the solution for, but not any new data. The user could not update the system. If
there is new information, or new systems, or new materials the system cannot
handle it without developer update. The developer must update the rules to
include new data. An expert system would be an obsolete system in a short period
because of the quick development of the additive manufacturing technologies,
materials and applications around the world.
4- Knowledge based system (KBS) and database (DB)
Knowledge based systems (KBS) are computer systems that imitate human
problem solving methods using a combination of artificial intelligence and
knowledge base. The KBS consists of three main components: 1- knowledge base
(KB), 2- inference engine (IE) and 3- graphical user interface (GUI). The
knowledge base contains the knowledge used to solve problems, the inference
engine derives answers from the knowledge base and the graphical user interface
communicates with users taking inputs and showing outputs (Lai et al.,2011).
KBS can perform knowledge management, reasoning, explanation and decision
support (Li et al.,2011). Sometimes KBS and ES are used interchangeably to
represent the same term but this is not precise as mentioned in (Freiberg et
al.,2012). The KBS is the broader term that contains expert system as one type of
it. The types of KBS could include: expert systems, case based reasoning, genetic
algorithms, intelligent agents and data mining (Kendal and Creen,2007).
A database is a structured collection of data or records stored in tables with

relationship in a computer program. It could be queried to retrieve information
(Lai et al.,2011).
The fourth idea was to develop an integrated system that consists of a knowledge
based system (KBS) and database. The KBS contains flexible rules (If Then)
which use variables and not constant values. Each time the user use the system,
the variables are defined. The KBS retrieves the data from the database depending
on the value of the variables decided by the user. The KBS queries the database
and retrieves the information and gives it to the user.
38
The user can also update and customize the database, which is a key advantage.
The flexibility of the IF-Then rules (using variables) gives a further advantage that
the system could be updated easily by the user without changing the programming
of the developed system. Furthermore, it is a step by step system where the user
understands why selection is made and also the reasons why a selection cannot be
made. Figure 3.2 shows a comparison between KBS architecture and KBS +
database architecture. The KBS architecture is a fixed inflexible system in which
the user deals only with what the developer had previously fed into the system.
The database embedded in the KBS architecture is used to express the knowledge
base that contains the fixed rules and fixed expertise, while in the KBS + database
architecture, the database is used to express the information such as records of
process chains, materials and machines. The KBS gets the inputs from the user,
sends queries to the database, retrieves the information and gets back to the user
with the outputs. The user can update the database directly which gives this
architecture a good advantage. Consequently, the knowledge based system (KBS)
and database (DB) architecture was selected because it can achieve the system
target specifications explained in section 3.1.
Updating Database
User User
Updating
y
Outputs
Outputs
r
ue
Inputs
Inputs
n
io
at
rm
fo
In
KBS Developing Developer KBS Developing

Developer
KBS KBS + Database
Figure 3.2: Comparison between KBS and KBS + Database Architectures.
3.3 AMDSS General Overview
The AMDSS was developed in the Matlab environment (MathWorks, Matlab

version R2009b) by applying knowledge based system decision rules to select
from database (Microsoft Access Database) using the database tool box embedded
in Matlab. Figure 3.3 shows the AMDSS main structure. The AMDSS consists of
39
three main components: 1- graphical user interface (GUI), 2- knowledge based

system (KBS) and 3- database (DB).
The KBS consists of three main parts: 1- selection, 2- browse and 3- update. The
selection part assists users in the selection of the appropriate process according to
their criteria. The browse part helps users to explore the processes and or the
materials present in the database. The update part is where users can update the
database with the new materials, processes, finishing methods, intermediate
materials and machines.
The selection part of the KBS consists of six main elements which are: 1- process
selection, 2- material selection, 3- ranking, 4- intermediate material selection, 5-
finishing options selection and 6- machine selection.
The database component within the AMDSS contains five main tables: - process
table, 2- material table, 3-finshing table, 4- intermediate material table and 5-
machine table.
AMDSS
GUI KBS DB
Update Selection Browse
Process Selection Material Selection Process Table Material Table
Intermediate material Intermediate Material

Ranking Finishing Table Table
Selection
Finishing options Machine Table

Selection Machine Selection
Figure 3.3: AMDSS Main Structure.
40
The selection part in the KBS consists of two main phases: feasibility phase and
selection phase (Ghazy and Dalgarno,2011). Figure 3.4 shows the general
framework of the selection part of the KBS.
Interface Knowledge Based Interface

Database (Outputs)
(User Inputs) System
Process Process Possible Processes

Process Requirements
Selection Table
Material Requirements Material

Selection Material Possible
Table Processes+ Materials
Criteria Weights SMART Ranking

Method Ranked
Processes+ Materials
Machine
Selection Machine Possible Machines
Table
Finishing Finishing Possible Finishing

Methods methods Methods
Table
Figure 3.4: General Framework of the Selection Part of the KBS.
Phase 1: Feasibility Phase
The first phase is the additive manufacturing feasibility phase, and is done in two
steps:
The first step is a process filtering step, in which the user decides on the part
requirements like size, quantity, surface finish, minimum wall thickness and
accuracy level. The output of this step is the possible processes that could meet
the part requirements or if the part requirements are not feasible, the reasons why
the part cannot be made with additive manufacturing processes. The KBS
connects to database, queries them and gets the processes that satisfy these
requirements The AMDSS provides also capability analysis to clearly understand
the gap between part process requirements and capabilities.
The second step is a material filtering step, in which the user can select among
different material properties including general properties like density and colour,
mechanical properties like strength, hardness, and tensile modulus, electrical
properties like dielectric strength, thermal properties like heat deflection
41
temperature and heat resistance, and finally environment conditions like sanitary
conditions (medical consideration) and water resistance. The output of this step is
the possible materials that can do the job within the selected processes in step one
or if there is no materials that can meet all the requirements the reasons why the
part cannot be made using one of the available materials.
The AMDSS provides also capability analysis to clearly understand the gap
between part material requirements and capabilities.
Phase 2: Selection Phase
The second phase is the selection phase, in which the user with the aid of the
AMDSS selects the suitable processes and materials based on the user criteria.
Furthermore, user selects intermediate materials, finishing options and machines.
It starts by the user weighting some criteria by giving a scale from 1 to 10 with ten
representing very important and one represents not important. The nine criteria
used are:
Strength
Hardness
Heat deflection temperature
Density
Dielectric strength
Modulus
Wall thickness
Accuracy
Surface finish
The AMDSS uses a method called SMART (Simple Multi Attribute Rating
Technique) developed by Edwards in 1977 (Edwards and Barron,1994) to rank the
processes and materials according to the weights of the criteria selected. The
ranking value is obtained simply as the weighted algebraic mean of the utility
values associated with it (Edwards and Barron,1994).
42
SMART was selected as a ranking method because it is simple, flexible, can be

applied easily and most importantly because adding new alternatives to the
method does not affect the old alternatives (unlike pair-wise comparison methods
such as analytical hierarchy process (AHP) that needs that users make the pair-
wise comparison each time a new alternative is added). The method was used in
the AMDSS to rank different processes and materials together. Adding new
alternatives such as processes and materials could be done without the need to
change the programming logic which helps to make the developed system
updatable as required in the target specifications.
3.3.1 Graphical User Interface
The graphical user interface (GUI) was the first component of the developed
AMDSS. The user uses the GUI to send data and receives information to and from
the KBS. Figure 3.5 shows the first screen that appears when the AMDSS starts.
The screen has three panels.
The first panel is the selection panel and contains one button called
selection module. It is used to select the process, material, machines,
finishing options and intermediate material. When this button is pressed a
window screen of the part requirements opens. Figure 3.12 shows the
window screen.
The second panel is the browse panel and contains two buttons called
specific material module and specific process module. It is used to browse
the database for a specific process or a specific material.
The third panel is the update panel and contains five buttons called add a
process, add a material, add an intermediate material, add a machine, and
add a finishing process. It is used to update the process, material,
intermediate materials, machines and finishing options.
43
Figure 3.5: Start Screen of the Developed AMDSS.
To understand the relation between the KBS and the GUI, the selection GUI
figures are demonstrated in the next section.
3.3.2 Knowledge Based System
Knowledge Based System (KBS) is the second component of the developed

AMDSS and is composed of three parts: selection, browse and update. The
following sections describe them in detail.
1. Selection part
The KBS selection part consists of six elements which are: the process chain
selection element, the material selection element, the ranking element, the
finishing options element, the intermediate material selection element and the
machines options element. The selection part within the KBS is the core part of
the AMDSS. For this reason, an IDEF0 model and flowcharts are used to describe
how the selection process is made. Despite the versatility of the IDEF0 to express
and explain the development of systems it does not contain decision diagrams. To
44
overcome this, flowcharts are also used alongside the IDEF0 models in the
explaining of the system development.
IDEF0 Model Development for the Selection Part of AMDSS
IDEF is an Integration computer-aided manufacturing DEFinition language that

consists of a set of re-engineering techniques developed by the American Air
Force to facilitate manufacturing automation. Following a structured system
analysis methodology, the IDEF methods supply a powerful means of analysis and
development of systems (IDEF,2012). The selection part of the AMDSS was
analyzed using IDEF0 (Integration DEFinition language 0) method within the
IDEF family, which is used for functional or activity modeling of a wide variety
of automated and non-automated systems for existing and non-existing systems
(Kappes,1997).
IDEF0 is used here to describe every function performed within the selection part
of the KBS and to whatever detail needed for each function as illustrated by the
IDEF0 node diagram in Figure 3.6. Table 3-1 shows the IDEF0 node index.
A-0
A0
A1 A2 A3 A4 A5 A6
A11 A12 A13 A14 A15 A16 A21 A22 A23 A24 A25 A26 A31 A32 A33 A34 A34 A35
Figure 3.6: IDEF0 Node Tree Diagram
The A-0 diagram shown in Figure 3.7 is the top-level context diagram, on which
the subject of the model is represented by a single box with its bounding arrows.
This diagram, which is develop an additive manufacturing decision support
system, represents the basic and main function of the model. Figure 3.8
represents the A0 diagram which shows the main steps of the selection part of the
45
KBS. The intermediate materials are the materials used when there is an AM
process chain consisting of different stages. The possible material in step two
(check possible materials for the selected processes) is the final part material,
while the intermediate materials are the materials used in the different process
stages before the final stage. Figure 3.9 shows the A1 diagram which checks the
additive manufacturing possible process steps. Figure 3.10 represents A2 diagram
which checks possible materials for the possible processes. Figure 3.11 shows the
A3 diagram which is: rank using SMART methods.
46
A0 Develop an Additive Manufacturing Decision Support System (AMDSS)
A-0 Main Steps of the Selection Part of the KBS
A1 Check AM possible processes

A11 Connect to the database and get the processes that satisfy
minimum wall thickness.
minimum accuracy.
minimum quantity.
minimum surface finish.
A15 Connect to the database and get the processes that satisfy the
required dimensions.
A16 Intersect all process lists.
A2 Check possible materials for the possible processes

A21 Connect to the database and get all the material that satisfy the
general properties.
mechanical properties.
electrical properties.
thermal properties.
environmental properties.
A26 intersect all material lists.
A3 Rank using SMART method

A31 Enter the weights of each criterion from 1 to 10.
A32 Normalize the weights.
A32 Divide each alternative score value by the maximum alternative
score value for each criterion and multiply by 10.
A34 Multiply normalized score values by the normalized weights.
A35 Sum the weighted score values for each alternative and sort.
A4 Check possible machines
A5 Check possible finishing methods

Table 3-1: IDEF0 Node Tree Index
A6 Check possible intermediate materials
47
Specific
Machines Material Finishing
Capabilities Properties Methods
Processes
capabilities
Possible Processes
Part Infromation DEVELOP AN ADDITIVE Possible Materials

MANUFACTRUING
DECISION SUPPORT Possible Finishing Methods
SYSTEM
A0 Possible Machines
Possible Intermediate Materials
DSS Developer
NODE: A-0 TITLE: DEVELOP AN ADDITIVE MANUFACTRUING DECISION SUPPORT SYSTEM (AMDSS) NO.: 1
Figure 3.7: A-0, Develop an Additive Manufacturing Decision Support System.
48
Processes
capabilities
Part Check AM Report process problems
Infromation Possible
Processes
1 Material
A1 Properties
Check Report materials problems

possible
materials for
the possible
processes
Possible LM Processes 2
A2
Rank using Specific
SMART method Machines
Capabilites
3
Possible LM processes+ materials A3
Check possible Possible machines

machines
Ranked processes+ materials
Finishing
4 Methods
Check possible
finishing Possible finishing methods
methods
Intermediate materials
5
Possible
Check possible intermediate
intermediate materials
material
6
NODE: A0 TITLE: Main Steps of the Selection Part of the KBS NO.: 2
Figure 3.8: A0, Main Steps of the Selection Part of the KBS.
49
Min wall
thickness connect to the database and
get the processes that satisfy
Processes that satisfy wall thickness requirement
the min wall thickness
1
connect to the database and

Accuracy level get the processes that satisfy
the required accuracy level Processes that satisfy accuracy level requirement

Surface finish level get the processes that satisfy
the required surface finish Processes that satisfy surface finish level requirement
3

Required quantity get the processes that satisfy
Processes that satisfy quantity requirement
the required quantity
4

Required dimensions (X,Y,Z) get the processes that satisfy Processes that satisfy dimensions requirements
th erequired dimensions
Report process
5 problems
Intersect all variables Possible LM

processes
6
NODE: A1 TITLE: Check additive manufacturing possible processes NO.: 3
Figure 3.9: A1, Check Additive Manufacturing Possible Processes.
50
Density Connect to the database and get
the processes and materials that Processes + materials that
satisfy the general properties satisfy the selected (if any)
Color
1 general properties
Strength Connect to the database and get the Processes + materials that satisfy
processes and materials that satisfy the selected (if any) mechanical
Hardness
the mechanicali properties properties
Tensile
Modulus 2
Connect to the database and get the

Processes + materials that
processes and materials that satisfy the
Dielectric strength satisfy the selected (if any)
electrical properties
3
Heat deflection temperature Connect to the database and get the Processes + materials that
processes and materials that satisfy the satisfy the selected (if any)
Heat Resistance thermal properties thermal properties
4
Connect to the database and get the Processes + materials that

Sanitary conditions satisfy the selected (if any)
enviromental properties enviromental properties
Water resistance
5 Possible LM
processes+
Intersect the selected
materials
variables
NODE: A2 TITLE: Check possible materials for the possible processes NO.: 4
Figure 3.10: A2, Check Possible Materials for the Possible Processes.
51
Possible LM
Enter the
processes+
weights of
materials
each
criterion
from 1 to 10
Criteria
1
Normalized
Normalize
weights
the weights
Weights
2
Divide each alternative score value

by the maximum alternative score
value for each criterion and multiply
Score value for
by 10
each alternative
for each criterion 3
Normalized score Multiply

values normalized
score values by Weighted score
the normalized values
weights
4 Sum the weighted
score values for Sorted
each alternative and alternatives
sort
5
NODE: A3 TITLE: Rank using SMART method NO.: 5
Figure 3.11: A3, Rank Using SMART Method.
52
The Elements of the Selection Part of the KBS
I. Processes Chain Selection Element
The processes chain selection element is responsible for the selection of the
processes that could meet the part requirements regarding minimum wall
thickness, accuracy, surface finish, required quantity and dimensions X, Y and Z.
Figure 3.12 shows the process selection screen.
Figure 3.12: Process Selection Screen.
The minimum wall thickness can be selected by selecting an option from three
options in a drop down menu: very-thin < 0.5 mm, thin-average 0.5 - 2 mm and
average-wide > 2 mm. The Accuracy can be selected by selecting an option from
three options in a drop down menu: tight < 0.1 mm, Average 0.1 - 0.25 mm and
loose > 0.25 mm. The surface finish can be selected by selecting an option from
three options in a drop down menu: Excellent 0 - 0.32 m, good average 0.32
12.5 m and average rough > 12.5 m. The quantity can be selected by selecting
an option from four options in a drop down menu: one, low 2- 20 parts, medium
20-25 parts and high > 50 parts. The length, width and height are a free text boxes
that can be entered manually. The terminologies and range values of minimum
wall thickness, accuracy and surface finish chosen in this study are based on
53
terminologies and ranges values set by Munguia (2009). The output of this
element is either the possible processes that could manufacture the part or the
reasons why the part cannot be made.
Figure 3.13: List of the Possible Processes that Could Manufacture the Part.
When the user enters the requirements and presses the (Find suitable processes)
button another screen appears showing all the possible processes that could
manufacture the part as illustrated in Figure 3.13.
Manufacturability Evaluation and Capability Analysis
Manufacturability evaluation is one of the key features of the developed AMDSS.

First, the system tells the user if there is a problem with any of the part
requirements that could not be met for a specific reason. For example, as shown in
Figure 3.14, the available processes in the database cannot meet the required
surface finish and the required width (Y) which is the distance along Y axis.
Secondly, in some cases, the part requirements together cannot lead to the
selection of a possible AM process. KBS shows a table that clearly identifies the
variables which conflict with one another. The benefit is that the user could
change some of the part requirements so the part could be made with AM
processes.
54
Figure 3.14: Screen Shows the Process Reasons that Prevent AMDSS from Giving a
Solution.
If there is no conflict between the two variables the system shows (ok) but if there
is a problem the system shows (prb) indicating that there is a problem or a conflict
between the two variables. It means that there is no material in the database that
could satisfy both variables. For example, Figure 3.15 shows that there is a
conflict between X and the wall thickness, X and the accuracy, Y and the wall
thickness, Y and the accuracy, Z and the wall thickness, and finally Z and the
accuracy.
Figure 3.15: Table that Shows the Process Parameters that are not Feasible Together.
Furthermore, KBS shows a warning message which is a specific concern about

part quality when the accuracy is very tight and the wall thickness is very thin.
The AM technologies do not show satisfactory results in this case.
55
The KBS gives a warning to the user as shown in Figure 3.16, and lets him select
between three options. The first option is to accept the less quality and continue
with the system. The second option is to terminate the program. The third option
is to compromise between requirements of the part.
Figure 3.16: A Window that Gives a Concern to the User about Part Quality.
Figure 3.17: Capability Analysis Screen.
56
The capability analysis is another advantage of the AMDSS. As shown in Figure

3.17, the AMDSS could tell the possible X, Y, and Z and compare them to the
required X, Y and Z. For example the required width is 2000 mm while the
maximum possible is 1550 mm.
Process selection logic flowchart
Figure 3.18 shows a flowchart that explains how the KBS takes the user inputs
and makes the process selection. When the user enters the values of the required
fields in the process screen in Figure 3.12, the variables are saved to a workspace.
The first step in the process selection is that the KBS connects to the database and
creates an empty reasons list in which the KBS adds all the reasons (if any) that
prevent the processes in the database from manufacturing the required part. Next,
user obtains the values saved by the user in the workspace of process variables:
the wall thickness, accuracy, surface finish, quantity, length, width and height.
The KBS gets the lists of the processes that satisfy the required variables. For
example, the KBS connects to the database and gets all the processes that are
capable of doing the required minimum wall thickness. If the process list is empty
this means that no AM process in the database can meet the required wall
thickness. The same logic is used for all the process variables. The next step is the
intersection between all the process lists. The intersection between the process
lists gives a new list which is a list of the processes that satisfy all the
requirements.
The logic of the manufacturability evaluation is explained in the following part of

the flowchart: if the new list is empty, the reasons list is checked. If the reasons
list contains a least one reason, the KBS displays the reasons list but if the reasons
list is empty means that there are not common processes that satisfy all the process
variables. In this case, an intersection table is displayed. The KBS connects to the
database and checks each two process variables together. It tries to retrieve data
that satisfies the two variables. If there are common processes that satisfy the two
variables together it means that there is no problem between these two variables
and if not it means that the user has to modify one of the variables to find a
possible process.
57
Start
Connect to the database
Create an empty reasons list
For i =1: 8, get the value of the variable (i) from

the workspace (i=wall thickness, accuracy,
surface finish, quantity, X, Y, and Z)
Get list (i) of the processes that their variable (i)

is like or bigger than the required variable (i)
Yes Add to the reasons list : No

Is the list (i) of the
layer manufacturing can meet
processes empty?
your variable (i)
No
i=i+1
No
Is i= 9?
Yes
New process list = Intersect all the lists of the

processes
Is the Yes Display

Is new process Yes
reasons list Intersection
list empty?
empty? table
No No
Display new process list Display reasons list
Figure 3.18: Checking AM Processes Flowchart.
58
II. Material Selection Element
The material selection element is responsible for the selection of the materials of
the processes that have passed the processes selection and could meet the part
requirements regarding the material properties. The output of this element is either
the possible materials that could be potential alternatives for the user or the
reasons why the part cannot be made with any of the material available in the
database.
All the values of the properties are based on the minimum values. The user has to
select at least one property. Figure 3.19 shows the screen where the user enters the
values of the material properties and Figure 3.20 shows an example of the list of
feasible materials that could manufacture the part.
Figure 3.19: Material Properties Screen.
59
Figure 3.20: List of Possible Processes and Material that Could Manufacture the Part.
Material Properties Evaluation and Capability Analysis
Material properties evaluation is one of the key features of the developed

AMDSS. First, the system tells the user if there is a problem with any of the
material requirements that could not be met for a specific reason. For example, as
shown in Figure 3.21 the available materials in the database cannot meet the
required strength. Secondly, in some cases, the material requirements together
cannot lead to the selection of a possible AM material. The KBS shows a table
that clearly identifies the variables conflicting with each other. The benefit is that
the user could change some of the material requirements so the part could be made
with AM processes.
Figure 3.21: Screen Shows the Material Reasons that Prevent AMDSS from Giving a
Solution.
60
For example, Figure 3.22 shows that there is a conflict between colour and
strength. The user could change one of them to finally have a possible AM
solution.
Figure 3.22: Table that shows the material Parameters that are not Feasible Together.
Besides the material properties evaluation, the capability analysis is another

advantage of the AMDSS. As shown in Figure 3.23 the AMDSS could tell the
possible material capabilities in the database regarding density, strength, hardness,
dielectric strength, heat deflection temperature and tensile modulus and compare
them to the required material requirements. For example, the required strength is
1500 MPa while the maximum available in the database is 1430 MPa.
In addition to that, there is a button called (possible processes colour) which when
pressed allows the user to browse the available colours of the different AM
processes.
Figure 3.23: Material Capability Analysis.
61
Material Selection Logic Flowchart
Figure 3.24 shows a flowchart that explains in detail how the material selection
was done. When the user enters the values of one or more variables in the material
properties screen shown in Figure 3.19 the variables are saved to a workspace.
The material properties used are: density, colour, tensile strength, modulus, heat
deflection temperature (HDT), heat resistance, sanitary conditions (medical
consideration) and water resistance. These material properties were selected
because they are common properties used by design engineers and users when
they need to make selections. In addition, there is limited data availability for
other AM material properties.
The first step in the flowchart is that the KBS connects to the database and creates
an empty reasons list in which the KBS adds all the reasons that prevent AM
materials available in the database (within the feasible processes) from
manufacturing the required part. Next, the KBS retrieves the material variables
(material properties) entered by the user. The variables are divided into two
groups. The first group is the (minimum values group). For this group, the KBS
obtains from the database a material list that has values bigger than or equal the
required values set by users. This group includes six variables which are: density,
strength, tensile modulus, hardness, dielectric strength and heat deflection
temperature. The second group is the (like values group). For this group, the KBS
obtains from the database a material list that has values like exactly the values set
by users. This group includes four variables which are: colour, heat resistance,
sanitary conditions and water resistance.
Each time the KBS obtains a material list for each material variable, a checking is
done if the material list is empty which means that no AM materials in the
database within the possible processes can meet the required values set by users
for this specific material variable. The KBS adds to the reasons list this reason.
The next step is to check which material variables are selected by users because
they do not have to select all variables. An intersection between the selected
material lists is done. The intersection between the selected material lists gives a
new list which is a list of the materials that satisfy all the requirements.
62
Start B

No
C Is j= 5?
For i=1: 6 get the value of the variable (i) from the
Yes
workspace (i=density, strength, modulus,
hardness, dielectric strength, and HDT)
Check which variables (material properties) are
selected
Get a list (i) of the materials that their variable (i)

equals or bigger than the required variable (i)
within the feasible processes
New material list = Intersect all the selected
material lists
Yes Add to the reasons list:

No layer manufacturing
materials empty? No
can meet your variable (i) Display new Is new material
material list list empty?
Yes
No
i=i+1
No Is the Yes
reasons list
empty?
No
Is i= 7?
Display Intersection
Display reasons list
table
Yes
For j=1: 4 get the value of the variable (j) from the
C workspace (j= colour, heat resistance, sanitary
conditions and water resistance)
Get a list (j) of the materials that their variable (j)

is like the required variable (j)

Is the list (j) of the
materials empty?
can meet your variable (j)
No
j=j+1
Figure 3.24: Checking AM Materials Flowchart.
63
The logic of the material properties evaluation and capability analysis is explained
in the following part of the flowchart: if the new list is empty, the reasons list is
checked. If the reasons list contains a least one reason, the AMDSS displays the
reasons list but if the reasons list is empty means that there is no common material
that satisfies all the materials variables. In this case, an intersection table is
displayed. The KBS connects to the database and checks each two material
variables together. It tries to retrieve data that satisfy the two variables. If there is
at least a single common process that satisfies the two variables together it means
that there is no problem between these two variables and if not it means that the
user has to modify one of these two variables to find a possible material.
III. Ranking Element
The third element in the KBS selection part is ranking. Any ranking process
requires criteria or attributes to rank alternatives according to them. The selection
of the ranking criteria to be used in the developed system was made after
exploring literature surveys, checking available material properties data and
asking AM industry experts. In the literature, Hermann and Allen (1999) have
used two types of criteria: 1- process criteria such as: cost, minimum wall
thickness, accuracy and build time and 2- material criteria such as: availability,
density, elastic modulus, tensile strength, elongation and heat deflection
temperature. Byun and Lee (2005) have used accuracy, surface finish, strength ,
elongation, part cost and build time criteria. Hanumaiah et al. (2006) have used
cost, lead time, accuracy, surface finish, strength, and flexibility to changes
(selection was focused on rapid tooling) criteria. Munguia et al. (2010) have used
geometry, appearance, mechanical requirements and functional requirements
criteria.
The criteria selected for the AMDSS as shown in Figure 3.25 were: strength,
hardness, heat deflection temperature, density, dielectric strength, modulus, wall
thickness, accuracy and surface finish. The user enters values for the criteria. The
values entered are from one to ten. The user does not have to enter them all. If the
fields are left empty the system considers all the weights to be equal. Values could
be repeated, meaning that users could use any weight value several times.
64
When the user presses the next button, a final ranking screen appears as shown in
Figure 3.26. It includes three buttons which are: (Possible Intermediate Materials),
(Finishing Options), and (Machine selection).
Figure 3.25: Criteria Screen.
Figure 3.26: Final Ranking Screen.
65
Ranking Logic flowchart
Ranking is done using a method called SMART method. Figure 3.27 and Figure
3.28 show flowcharts that explain how the SMART method was used. The first
step is to get from the workspace all the processes and materials that satisfy the
user requirements. These are the successful candidates that passed all the required
inputs for processes and materials. The next step is to get the normalized score
value of each alternative regarding each criteria. In SMART method in general,
user can assign a value usually between 1 and 10 or 1 and 100 that represents the
performance of each alternative regarding certain criteria. Usually it is a
qualitative value set by the user. This value is called in the developed system
normalized score value and it has a range between 1 and 10. In the developed
system, there are nine criteria. Six of these criteria are quantitative: strength,
hardness, heat deflection temperature, density, dielectric strength, and modulus;
three are qualitative wall thickness, accuracy and surface finish. For the
quantitative criteria, the normalized value is calculated as follows: for each
criterion, the KBS gets the maximum value of this criterion of the successful
candidates list and the normalized score is calculated by dividing the value of each
alternative by the maximum value and finally multiplied by ten. For example the
strength criterion: the strength of each alternative is divided by the maximum
strength in the successful candidates list and then multiplied by ten. The same
approach was taken for all the quantitative criteria.
For the qualitative criteria, each of them has three levels that user can choose
between these levels like for example the wall thickness: the user can select 1-
Average-Wide, 2- Thin-Average, and 3- Very-Thin. The normalized score values
for each process regarding the minimum wall thickness criteria for average-wide
is four, for thin-average is seven and finally for very-thin is ten. The same
approach was taken for the other qualitative criteria: accuracy and surface finish.
Next the KBS gets all the weights of the user for each criterion. If a weight is not
set by a user the system gives it a value of one. Next a variable called weighted
score value is calculated: Weighted score value = normalized score value of each
alternative x weight of the user for each criterion. The sum of all the weighted
score values for each alternative gives the final score. Next the alternatives are
ranked according to the final score in descending order.
66
Start B
Get the processes and materials list that satisfy

the user requirements from the workspace Connect to the database
(successful candidates)
Get all the wall thickness(W.T), accuracy(Acc)

Get the maximum strength of the successful
and surface finish (S.F) of the successful
candidates
candidates
Strength of each material

Normalized Score Value 10 Is W.T= Yes Normalized score
Maximum strength Average Wide? value = 4
Get the maximum modulus of the successful No

candidates
Normalized
Is W.T= Thin Yes
score value
Average?
=7
Modulus of each material

Normalized Score Value 10
Maximum modulus No
Normalized score
value = 10
Get the maximum hardness of the successful

No
candidates
Yes Normalized score

Is Acc= loose?
value = 4
hardness of each material
Maximum hardness
No
Yes Normalized
Is Acc=
Get the maximum density of the successful score value
Average?
candidates =7
No
density of each material Normalized score

Normalized Score Value 10 value = 10
Maximum density
No
Get the maximum dielectric strength(D.E) of the Is S.F= Yes

Normalized score
successful candidates Average
value = 4
Rough?
D.E of each material No

Maximum D.E Yes
Normalized
Is S.F= Good
score value
Average?
=7
Get the maximum heat deflection temperature
(HDT) of the successful candidates
No
Normalized score
value = 10
HDT of each material
Maximum HDT
B C
Figure 3.27: SMART Ranking Flowchart
67
Get all the weights of the users for each criterion
criterion does not Yes

have a weight? Weight =1
No
Weight = Users weight
Weighted score value =

normalized score value for each criterion X users
weight for this criterion
Score = Sum all the weighted score values
Rank the scores in descending order
Display ranked scores
Figure 3.28: SMART Ranking Flowchart (continued).
IV. Intermediate Material Selection Element
When the (Possible Intermediate Material) button in Figure 3.26 is pressed, a

screen appears to ask the user about the AM process chain that he needs to select
its intermediate material as shown in Figure 3.29. This list shows only the process
chain and does not show any single process by itself because single processes do
not need intermediate materials. When the user selects the process (SLA +
Investment Casting in this case), the KBS connects to the database, retrieves from
the intermediate material table all the intermediate materials of the process
selected and shows them in a table as shown in Figure 3.30. The intermediate
material in this case is the material of the SLA process, while the final material is
the material for investment casting.
68
The user can select more than one intermediate material at the same time using the
control button. This could be the case when there are more than two processes in
the process chain. When the selection has been made the back button returns to
the final ranking screen. The selections of the user are saved for retrieval purpose.
Figure 3.29: List of the Processes that Need Intermediate Materials.
Figure 3.30: Selection of Intermediate Material Screen.
69
V. Finishing Options Selection Element
When the (Finishing Options) button shown in Figure 3.26 is pressed, a screen
appears to ask the user to select an appropriate finishing option as shown in Figure
3.31. By double clicking on the process that he needs to select its finishing option
(SLA in this case), the KBS connects to the database, retrieves from the finishing
options table all finishing options of the process selected and shows them in a list
as shown in the finishing options screen shown in Figure 3.32.
Figure 3.31: Selection of the Process that the User Wants to See its Finishing Methods.
Figure 3.32: Selecting from the Finishing Options Screen.
70
The user can select more than one finishing option by pressing the CTRL button.
When the selection has been made done the back button returns to the final
ranking screen. The selections of the user are saved for retrieval purpose.
VI. Machines Selection Element
When the (Machine Selection) button shown in Figure 3.26 is pressed, a screen
appears to ask the user to select a specific machine as shown in Figure 3.33. By
double clicking on the process needed, the KBS connects to the database, retrieves
from the machines table all machines of the process selected and shows them in a
table as shown in Figure 3.34. The user can press in any cell in the row of the
machine needed. When the selection has been made the back button returns to the
final ranking screen. The selections of the user are saved for retrieval purpose.
Figure 3.33: Selection of the Processes that the User Needs to See its Machines Screen.
Figure 3.34: Selection of the Machines Screen.

71
Technical Summary
After making all the selections of the intermediate materials, finishing options,
and machine selection there is a button in the final ranking screen (Figure 3.26)
called (Technical Summary). This button when pressed shows a screen that
contains a summary of all the selections that have been made. The KBS retrieves
all the saved users selections and displays them in one table.
Figure 3.35: Technical Summary Screen.
For illustration purpose, Figure 3.35 shows a technical summary. The process
selected is SLA, the material is ACCURA 10. The finishing method for the SLA
process is sanding (fine sand paper). There is no intermediate machine selected
because this is a single process and not a chain. The selected SLA machine is
IPRO 9000. Choosing finishing options, the intermediate material or the machines
is not mandatory. The user can select the process and the final material only. The
information presented in the technical summary screen is according to user
selection.
2. Browse Part
If the user needs a specific material and needs to explore the different available
processes that could manufacture a part using this specific type of material, he
presses the specific material module button and is presented a list of the current
materials in the database. The KBS connects to the database and retrieves all the
current materials available in the material table and shows them in a pop-up menu
72
shown in Figure 3.36. The pop-up menu will be always displaying the current
materials in the database, if a material is deleted it will disappear from the pop-up
menu and if a material is added it will appear in the pop-up menu. When a user
selects a material, he presses search the database and the KBS connects to the
database and shows a table that presents all the processes that could use this type
of material.
If the user knows the process and needs to explore the different available materials
for this process, he presses the specific process module button and is presented
with the current list of the processes in the database. The KBS connects to the
database and retrieves all the current processes available in the process table and
shows them in a pop-up menu shown in Figure 3.37. The pop-up menu will be
always displaying the current processes in the database, if a process is deleted it
will disappear from the pop-up menu if a process is added it will appear in the
pop-up menu. When a user selects a process, he presses search the database and
the KBS connects to the database and shows a table that presents all the materials
that could be used by the selected process.
Figure 3.36: Browsing Processes by Material.
73
Figure 3.37: Browsing Materials by Process.
3. Update Part
The update panel contains five buttons called add a process, add a material, add an
intermediate material, add a machine, and add a finishing process. To add a
process to the database user presses the (ADD A PROCESS) button and a window
appears as shown in Figure 3.38. The user has to enter all the fields so it can be
added in the database. The requested fields are: process name, minimum wall
thickness, accuracy, surface finish, quantity and dimensions (X, Y, and Z) of the
maximum build envelope of the process. When adding a new process, the process
name must be preceded by # as shown in Figure 3.37 and if it is a process chain
that has more than one stage it must have # between each stage. When the user
presses the (ADD) button the KBS connects to the database and inserts the new
process in it.
74
Figure 3.38: Adding a New Process.
To add a material to the database user presses the (ADD A MATERIAL) button
and a window appears as shown in Figure 3.39. The user has to add all the fields.
The fields are: process name, material strength, modulus, hardness, heat deflection
Temperature, dielectric strength, colour, water resistance, sanitary conditions,
density and heat resistance. The user could select between two options YES or
NO in pop-up menus for water resistance, sanitary conditions and heat resistance
variables. When the user presses the (ADD) button the KBS connects to the
database and inserts the new material in it.
To add an intermediate material to the database user presses the (ADD

INTERMEDIATE MATERIAL) button and a window appears as shown in Figure
3.40. The user has to enter all fields and they are the same as the material fields.
When the user presses the (ADD) button the KBS connects to the database and
inserts the new intermediate material in it.
75
Figure 3.39: Adding a New Material.
Figure 3.40: Adding a New Intermediate Material.
76
Figure 3.41: Adding a New Machine.
To add a machine to the database user presses the (ADD A MACHINE) button
and a window appears as shown in Figure 3.41. All the fields have to be filled.
The fields are: process name, machine name, build envelope and layer thickness.
When the user presses the (ADD) button the KBS connects to the database and
inserts the new machine in it.
To add a finishing process to the database user presses the (ADD A FINISHING
PROCESS) button and a window appears as shown in Figure 3.42. All the fields
have to be filled. The fields are: process name (AM process) and possible
finishing methods. When the user presses the (ADD) button the KBS connects to
the database and inserts the new finishing process in it.
77
Figure 3.42: Adding a Finishing Process.
3.3.3 Database
The database is the third element of the developed AMDSS. The database used
here consists of five tables: process table, material table, machine table, finishing
options table and intermediate material table.
The process table contains the following fields: ID, process name, wall thickness,
accuracy, surface finish, Max X, Max Y, Max Z and quantity. The process
selection, the browsing specific process module, the ranking by SMART method
and the process update use this table to get and send the required information from
and to it.
The Material table contains the following fields: ID, process name, material name,
tensile strength, tensile modulus, hardness, heat deflection temperature, sanitary
conditions, color, density, heat resistance, water resistance, and dielectric strength.
The material selection, the browsing specific material module, the ranking by
SMART method and the material update use this table to get and send the required
information from and to it.
The machine table contains the following fields: ID, process name, machine
name, build envelope, and layer thickness. The machine selection and the machine
update use this table to get and send the required information from and to it.
78
The finishing options table contains the following fields: ID, process name, and
the finishing method name. The finishing options selection and the finishing
methods update use this table to get and send the required information from and to
it.
Finally, the intermediate material table contains the same fields as the material
table. The intermediate material selection and the intermediate material update use
this table to get and send the required information from and to it.
The developed database contains 13 different processes, 133 different materials,

10 intermediate materials and 49 finishing methods. The information was
collected through vendors websites and service bureaus recommendations.
Figure 3.43: Database Screen Shot.
Entity Relationship Diagram (ERD)
One important advantage about database is relationships between records which

ensure data integrity. To understand these relationships a data modeling diagram
called entity relationship diagram (ERD) is shown in Figure 3.44. The ERD is a
graphical representation of how the records within the table deal with each other.
The software used to draw the ERD was graphical software called Microsoft
Visio. The fields in each table are also called entities. For example, the entities in
the process table are: ID, process name, wall thickness, accuracy, surface finish,
Max X, Max Y, Max Z and quantity. Each table has a primary key (PK) which is a
key that uniquely identify a single entity instance (it cannot be repeated). The
primary key in process table is the process name For example if a user enters a
new record for process field such as (#SLS) that is already present in the database
this new record will not be accepted in the database because the process entity is a
primary key. For all other tables (materials, intermediate materials, machines and
finishing) the primary key is an ID which is an automatic number that is
automatically generated for each new record. If a primary key of a certain table is
79
presented in another table it is called foreign key (FK). Each table could have
multiple foreign keys. The process name is a primary key in the process table and
a foreign key in all the other tables. Any changes that occur in one table will occur
in all the other tables. For example, if a process is added or deleted or altered in
one table the change will affect all the other tables. Another issue that ensures the
integrity between the tables is the type of relationships between them. The
possible relationships are: one-to-one, one-to-many and many-to-many. One-to-
one relationship means the record in table A could only be repeated once in table
B, the one-to-many relationship means that the record in table A could be repeated
several times in table B but not vice versa and the many-to-many relationship
means that the record in table A could be repeated in table B and vice versa. The
relationships between the process table and all other tables are one-to-many. The
process name can be repeated several times in any of the tables, which means that
one process could have many machines, many materials, many intermediate
materials and many finishing methods.
Material Table Intermediate Material Table

PK ID PK ID
Material Material
Strength Strength
Tensile Tensile
Hardness Hardness
HDT HDT
Sanitary Relationship 1 to many Sanitary
Colour Colour
Density Density
Heat Resistance 1..*
Heat Resistance
Wear Resistance Wear Resistance
Dielectric Strength Dielectric Strength
FK1 Process 1..* FK1 Process
Process Table
PK Process
ID
Wall Thikness Foreign Key
Accuracy
Surfae Finish
Max X
Max Y Primary Key
Max Z
Machine Table
Finishing Table PK ID
PK ID 1..*
Machine Name
Build Envelope
Finishing Method 1..* Layer Thikness
FK1 Process FK1 Process
Figure 3.44: Entity Relationship Diagram (ERD).
80
3.4 Discussion
This chapter has discussed the additive manufacturing decision support system
(AMDSS) design and development. To develop the system nine steps were used.
The third and fourth steps were discussed in this chapter. Step three focuses on
identifying system target specifications. Step four focuses on developing the
system. Different architectures of the system were considered and knowledge
based system (KBS) + database (DB) architecture was selected because it could
meet all the system target specifications.
One of the important points that helped to have an updatable customizable system
is the flexible IF-then rules used within the KBS system. The conditions of the IF-
then are based on variables and not constant values. Consequently, each time a
user uses the system and the values of the variables change, the rule gives a
different output. This flexibility gives the developed system an updatability
characteristic that prevents it from being obsolete after a while.
81
Chapter 4. Verification and Validation
This chapter explains how the verification and the validation of the developed
AMDSS were carried out. The verification was done through checking the logic
of the developed system. For validation, three industrial case studies were used.
The selection of the processes, materials, finishing methods and machines for
these three case studies was done in consultation with experts and service bureaus.
In addition, the same inputs for the three case studies are fed to the AMDSS. The
outputs of the AMDSS are compared with selections made with in consultation
with experts to assess how well the system work.
4.1 Verification
Before using the developed AMDSS, it has to be tested and verified. The aim of
the verification is to make sure that the developed system is working properly.
The verification was done through several ways.
First, while building the system, the system logic is checked at each stage before
moving to the next one. In addition, after the system had been completely
developed, the whole system was checked to make sure that the system was
working properly all together.
Some scenarios were applied to the system to test the input-output relationships
and the system performance when subjected to inputs change. To test the process
selection stage, some process inputs like size, quantity, surface finish, minimum
wall thickness and accuracy level were determined and the AMDSS retrieved the
AM processes that satisfy these inputs. The same inputs are retrieved manually
from the database. The process outputs from the ADMSS are compared to those
manually retrieved. As an example of one of the scenarios, the following inputs
were used:
Minimum wall thickness = thin-average
Accuracy = average
Surface finish = good-average
Quantity = 1
82
Length = 300 mm
Width = 300 mm
Height = 100 mm
The outputs determined by examination of the data (which were the processes that
could manufacture the part) were:
EBM
Laser Cusing
LENS
SLA
SLA+ Investment Casting
SLA+ Reaction Injection Molding
SLS
The results retrieved from the AMDSS and manually retrieved from database
were exactly similar. When changing these inputs, the outputs also change. This
checking process was carried out several times to assure that the AMDSS
retrieving logic of the process selection was correctly programmed. The same
testing method was done to the material selection stage and the outputs of
AMDSS were compared to those retrieved manually from the database. As an
example, using the same process inputs mentioned above, some material
requirements were used:
Strength = 100 MPa
Tensile = 5000 MPa
The outputs determined by examination of the data (which were the materials that
could be used to manufacture the part within the previously selected process) were
shown in Table 4-1.
83
Process Name Material Strength Modulus

#SLA#INV.CASTING A 201 (A12010 427 74600
#SLA#INV.CASTING 354 (A03540) 345 76700
#SLA#INV.CASTING C90710 390 82000
#SLA#INV.CASTING C86500 838 103000
#SLA#INV.CASTING G10500 705 216000
#SLA#INV.CASTING ASTM CF-20 580 199000
Table 4-1: An Example of Results Retrieved from a Verification Scenario.
It has to be mentioned that some providers of some metal fabrication processes

such as EBM, Laser Cusing and LENS do not provide complete datasheets for
their materials. For this reason, their materials were not listed because the strength
or the tensile information was not availabe in the database.
were exactly similar. When changing these inputs, the outputs also change. This
checking process was carried out several times to assure that the AMDSS
retrieving logic of the material selection was correctly programmed.
In addition, to test the manufacturability evaluation and the capability analysis,

some process inputs that are known to not be able to be manufactured using
additive manufacturing technology were fed to the system and then the system
checked for the correct reasons being provided. As an example of one of the
scenarios, the following inputs were used:
Minimum wall thickness = thin-average
Accuracy = average
Surface finish = excellent
Quantity = high
Length = 3000 mm
Width = 3000 mm
Height = 2000 mm
84
The output of the AMDSS was that no possible processes could meet your
required process inputs because of the following reasons:
No AM process within the database can meet your required length (X).
No AM process within the database can meet your required length (Y).
were exactly similar.
The same method was applied to the material properties evaluation and capability
analysis. For example, a required strength of 3000 MPa is fed to the AMDSS and
the output was that:
No AM process within the database can meet your required strength.

When checking this manually in the database it can be noticed that the maximum
strength available is 1430 MPa.
The comparison between the outputs of the manual retrieving from the database
and the ADMSS outputs for different stages was done throughout the
development so that errors were corrected immediately. These scenarios verify
that the model is working correctly and logically.
Furthermore, the calculations of the ranking of the SMART method have to be
tested. This is done though calculating using spreadsheets and comparing the
results with the AMDSS results. Gaining the same results proved that the logic of
the SMART method was used correctly within the developed system.
4.2 Validation
The main purpose of the validation process was to test that the developed system
gives realistic and technically feasible answers to the users and to assure that the
additive manufacturing selection knowledge and expertise were captured and built
correctly within the developed system. It means that the selections of the process
chains, materials, intermediate materials, finishing methods and machines made
by the AMDSS conform to the same selections made by a service bureau or an
expert user. To achieve this, three industrial case studies were chosen and the
selection of all the needed parameters done with the consultation of AM experts.
These three case studies were manufactured according to the selection so that the
tangible products can assure the compatibility to the user inputs. In addition,
capturing the knowledge needed for the additive manufacturing decision support
85
system was not only done by asking the experts and the service bureaus, but also
by learning from experimental work. The three case studies were fed to the
AMDSS and the outputs were compared to the real life decisions that were made
with consultation with the AM experts. Next sections explain the case studies in
detail.
4.2.1 Razor Blade Case Study
The first case study was a part consisting of smaller components assembled
together to form a disposable razor blade. The reason the factory needs to
manufacture this part is to have a functional prototyping for testing. Figure 4.1
shows STL file of one of the required parts called guard bar. The STL file was
sent to the additive manufacturing machine so the part can be made.
The experimental work
Different small parts are assembled together to form a disposable razor blade.
Figure 4.2 shows the different small components made using selective laser
sintering technology (SLS) in Nylon (trade name Duraform) using a 3D Systems
SLS machine at the Keyworth Institute, University of Leeds. Also these parts were
made using fused deposition modeling technology (FDM) in ABS material using
an FDM machine in Newcastle University. Both technologies did not show
satisfactory results according the accuracy and surface finish required by the user.
The first lesson taught from this case study is that the small wall thickness and the
very tight accuracy are difficult to achieve using the both used technologies.
Furthermore, experts confirm that this rule applies to most of additive
manufacturing technologies.
Figure 4.1: Guard Bar Part.
86
Figure 4.2: Razor Case Study Manufactured Parts.
Selecting using AMDSS
To validate the developed AMDSS, the case study information was fed to the
system and the output was compared to the real life application. Table 4-2 shows
the part information.
Part Requirements Material

Length ( mm) 39.6 Requirements
Strength (MPa) 50
Width (mm) 10.56
Height (mm) 5.9
Minimum Wall Very Thin (0.5 mm)
Thickness (mm)
Accuracy (mm) Tight <0.1 mm
Surface Finish Ra Good-Average (4 m)
(m)
Quantity 1
Table 4-2: Part and Material Requirements for the Top Cap.
When the part information was fed to the system, the output was the screen shown
in Figure 4.3 indicating that there is a concern about the quality of the product
because of the accuracy is tight and the wall thickness is very thin. The user has
three options: to accept this notice and continue with the system or to end the
program because additive manufacturing is not a solution for this case or to get
back and change some of the part information regarding accuracy and surface
finish. Because both accuracy and surface finish were important for this case
study, the user decided to terminate the system understanding that additive
87
manufacturing is not the right answer for his case. The AMDSS selections
conformed to selections made in consultations with experts. The user did not
select any AM process because there was not one that could satisfy their needs.
The same result was given by the AMDSS.
Figure 4.3: AMDSS Output for the Guard Bar Part.
4.2.2 Automotive Case Study (1)
The second case study was an automotive spare part. It is split into two parts to be
used as indirect patterns for sand casting to produce a pressure plate of a car
clutch. Figure 4.4 and Figure 4.5 show the first and second parts respectively.
88
Figure 4.4: The First Part of the Automotive Case Study 2.
Figure 4.5: The Second Part of the Automotive Case Study 2.
To manufacture the patterns a rapid tooling process that consists of three stages
was proposed by Ghazy et al. (2010). The first stage was to build the pattern
model using an additive manufacturing system. The second stage was to cast
epoxy resin into the pattern to create the mould. The final stage was to cast epoxy
resin into the mould to create the pattern.
The proposed rapid tooling process was applied to the automotive case study as
follows: in the first stage of the process the AM patterns were made in Nylon
89
(trade name Duraform) using a 3D Systems SLS machine at the Keyworth

Institute, University of Leeds. Figure 4.6 shows the produced patterns.
Figure 4.6: Rapid Tooling Patterns.
In the second stage an epoxy resin mould was created. Furthermore, sand paper
was applied to the produced mould to improve the surface finish. Finally, the third
stage was also applied; the epoxy resin was poured into the mould to produce the
required pattern. Figure 4.7 shows the final product after the sand casting process.
Figure 4.7: The Final Products after Casting.
90
When the part requirements shown in Table 4-3 were fed to the system, the
AMDSS retrieved the possible processes or process chains that could do the job
and listed them. The retrieved processes (single processes or process chains) were:
EBM
FDM
Laser Cusing
LENS
SLA
SLA + Investment casting
SLA + Silicone tooling + Vacuum Casting
SLA + Reaction Injection Molding
SLS
Part Requirements Material Requirements

Length ( mm) 230 Strength (MPa) 40
Width (mm) 245
Height (mm) H1 = 10 H2 = 25
Minimum Wall Average wide (2 mm)
Thickness (mm)
Accuracy (mm) Average= 0.20 mm
Surface Finish Ra Average Rough (18 m)
(m)
Quantity 1
Table 4-3: Part and Material Requirements for the Two Parts of Automotive Case Study 1.
The next step was the material requirements step. The user entered the material
requirements shown in Table 4-3 for the parts. The material requirement in this
case is the tensile strength equals 40 MPa. Figure 4.8 shows the possible processes
and materials that can manufacture this part. The user presses the next button, so
the criteria screen appears as shown in Figure 4.9. Because the strength was
important in this case, so the user gave weight of eight to it. Figure 4.10 shows the
91
final ranking screen. The first alternative was SLA process and DMX-SL100
material with the highest score of 42. This score was calculated with the SMART
method and depended on the user weights. If the weights were changed, the score
and consequently the order would have been changed. The user selected the SLS
process and the PA2200 material. Next, the user selected the finishing options
which was sanding as shown in Figure 4.11. The user can also select the machine.
Finally, the technical summary as shown in Figure 4.12 summarized the user
selections. The AMDSS selections conformed to selections made in consultations
with experts. It has to be underlined that the selections of process and material
made by the user were not the first ranked as shown in Figure 4.10 because there
are other factors that affect the selections such as price given by service bureau
and the availability of processes and materials.
Figure 4.8: List of the Possible Processes and Materials for Automotive Case Study 1.
92
Figure 4.9: Criteria Weighting for Automotive Case Study 1.
Figure 4.10: Final Ranking Screen for Automotive Case Study 1.
93
Figure 4.11: Choosing Finishing Methods for Automotive Case Study 1.
Figure 4.12: The Technical Summary Screen for Automotive Case Study 1.
94
4.2.3 Automotive Case Study (2)
The third case study was an automotive spare part too, and it was split into two
parts to be used as patterns for sand casting to produce rear axle brake housing
(Ghazy et al.,2011). Figure 4.13 and. Figure 4.14 show the first and second parts
respectively.
Figure 4.13: The First Part of the Automotive Case Study 2.
Figure 4.14: The Second Part of the Automotive Case Study 2.
The patterns were fabricated using SLS technology in polyamide material and
finished using filler and sanding process. The following steps were applied: 1-
manufacturing the parts using SLS technology from polyamide material, Figure
4.15 shows the SLS fabricated patterns, 2- applying the plastic filler to the parts,
95
3- smoothing the parts using sand papers. The Ra for the first pattern was 16.47
m and the Ra for the second pattern was 19.69 m. These values were acceptable
for the sand casting patterns and 4- starting the sand casting process Figure 4.16
shows the pictures of the casted products.
Figure 4.15: SLS Fabricated Patterns.
Figure 4.16: The Castings.
Because the patterns were going to be applied to heat and pressure of the sand
casting process, the user chose to have strength equals 50 MPa and heat deflection
temperature (HDT) equals 80 C. When the part requirements shown in Table 4-4
for the first and the second parts were fed to the system, the AMDSS retrieved the
possible processes that could do the job and listed them which were:
EBM
FDM
Laser Cusing
96
LENS
SLA
SLA+ Investment casting
SLA+ Silicone tooling +Vacuum Casting
SLS

Length ( mm) 205 Strength (Mpa) 50
Width (mm) 236 Heat Deflection Temp 80
Height (mm) H1 = 94.36 H2=54.10 (Degree Celsius)
Minimum Wall Average wide (2 mm)
Thickness (mm)
Accuracy (mm) Average= 0.20 mm
Surface Finish Ra Average Rough (18 m)
(m)
Quantity 1
Table 4-4: Part and Material Requirements for the Two Parts of Automotive Case Study 2.
The next step was the material requirements step. The user entered the material
requirements shown in Table 4-4. Figure 4.17 shows the possible processes and
materials that can manufacture this part. The user pressed the next button, so the
criteria screen appeared as shown in Figure 4.18. Because the strength and the
heat deflection temperature were important in this case, so the user gave weight
ten to both of them. Figure 4.19 shows the final ranking screen. The first
alternative was SLA process and Accura Bluestone material with the highest score
of 256. This score was calculated with the SMART method and depended on the
user weights. If the weights were changed, the score and consequently the order
would have been changed. The User selected the SLS process and the PA2200
material. Next, the user selected the finishing options which were vibro-finishing
and sanding as shown in Figure 4.20. The user selected the machine to be EOS
P730. Finally, the technical summary as shown in Figure 4.21 summarized the
user selections. The AMDSS selections conformed to selections made in
consultations with experts. It has to be underlined that the selections of process
and material made by the user were not the first ranked as shown in Figure 4.19
because there are other factors that affect the selections such as price given by
service bureau and the availability of processes and materials.
97
Figure 4.17: List of the Possible Processes and Materials for Automotive Case Study 2.
Figure 4.18: Criteria Weighting for Automotive Case Study 2.
98
Figure 4.19: Final Ranking Screen for Automotive Case Study 2.
Figure 4.20: Choosing Finishing Methods for Automotive Case Study 2.
99
Figure 4.21: The Technical Summary Screen for Automotive Case Study 2.
4.3 Discussion
To conclude, the verification process has proven that the developed system logic
is correct. The three case studies have helped in the validation process and also
have allowed the researcher to understand some concepts of additive
manufacturing technologies. The first lesson from case study one, showed that the
small features and the high level of quality together is difficult to be achieved
using additive manufacturing. The second lesson from case study two and three
showed that the selections made were not the best technically because cost factor
was taken into consideration. This showed a limitation in the developed ADMSS,
which is that the cost not included.
The case studies, by giving some lessons and underlining some AMDSS
limitations, showed a strong need that focusing on the users and testing the system
with them will give positive important feedback. The next chapter will test the
AMDSS with some users to obtain the feedback to understand the limitations of
the developed system, so further development can be done to finally have a better
system. Furthermore, most of the research done before in the additive
manufacturing selection area did not focus on the user or the customer.
Developing an additive manufacturing decision support system that includes the
user is one of this research objectives and contributions.
100
Chapter 5. User Feedback and Evaluation
A major drawback of many AM selectors is that they do not represent users and
they only show the point of view of their developers. To overcome this problem in
the developed AMDSS, one of the target specifications is to be a user focused
system. Obtaining feedback from users is the sixth step of the AMDSS
development steps discussed in chapter three and this chapter explains how this
was achieved. Furthermore, four users tested the developed AMDSS with four
products by comparing the selections of the system for these products to their
previously made selections.
5.1 User Feedback
There are three types of interviews (Myers and Newman,2007):
1- Structured interview in which the interviewer uses a set of rules and questions
that cannot be changed across all interviewees.
2- Semi-structured interview in which the interviewer could add questions or ask

for clarifications or make explanations.
3- Non-structured interview which is a general discussion between the

interviewer and the interviewee rather than a formal interview.
In this study, the feedback from users was obtained using a semi-structured
interview. This type was selected because it combines structural organization of
the questions with flexibility. It gives the interviewer the possibility of explaining
to interviewees different aspects of the system developed and asking about their
opinions.
The interview starts by giving an outline to the user at the developed AMDSS, and
explaining the different sections of the questionnaire. Next, the interviewee thinks
of a product that he would like to manufacture using AM technologies and starts
using the AMDSS and answers the questions. Some of the interviewees asked to
review some of the screens while answering the questions. Each interview took
approximately from forty five minutes to one hour.
101
5.1.1 Questionnaire Design
The questionnaire was designed to collect information from users in an organized

way. It consisted of three parts:
1- User profile. It gave the researcher an overview of users knowledge about

AM technologies and processes.
2- AMDSS screens. It asked users about the terminology used and the
information presented in the main screens
3- General questions. It gave information about the overall system performance

by asking users about the advantages and the limitations of the system.
Thirteen users responded to the questionnaires. They were classified into two
groups: advanced users and average users. The advanced users are technology
experts who work for service bureaus or deal with AM technologies on regular
basis in research centers. The average users are inexperienced users that have
from little to average experience in the AM field.
According to this classification four advanced users and nine average users
responded to the questionnaire. The questionnaire is shown in Table 5-1, Table
5-2, Table 5-3, Table 5-4, Table 5-5 and Table 5-6.
102
Additive Manufacturing Decision Support System (AMDSS) Feedback
1- User profile
Question 1:
How often do you need to select additive manufacturing process or
material?
a. Weekly
b. Monthly
c. Sporadically
d. Other:_______________________
Question 2:
Have you used any knowledge based system for selecting additive process
or material before? If yes please name it.
Question 3:
How do you normally select the following:
a. Additive manufacturing process
b. Material
c. Finishing methods
d. Machines
2- AMDSS Screens
Question 1: Process screen
i. Information Section
a. Do you think that the information required is appropriate?
b. For your application, do you have data for all the presented
fields?
c. Are there any fields that you would remove?
d. Are there any fields that you would add?
Table 5-1: AMDSS First Version Questionnaire (Page 1 of 6).
103
ii. Terminology Section

1- Do you think that the terminology used in the following drop down
menu is easy to understand? Use the scale give below each term.
[1= difficult to understand and 5 =easy to understand.]
Min wall thickness
(Very-Thin 0-0.5mm) / (Thin-Average 0.5-2mm) / (Average-Wide (0.2mm)
1 2 3 4 5
Accuracy
(Loose <0.1mm) / (Average 0.1-0.25mm) / (Tight>0.25mm)
1 2 3 4 5
Surface Finish
(Average-Rough>12.5 um) /(Good-Average 3.2-12.5 um) / (Excellent 0-0.32 um)
1 2 3 4 5
Quantity
(One) / (Low2-20) / (Medium 20-25) / (High>50)
1 2 3 4 5
If you think the terminology is difficult to understand please suggest

alternative phrasing.
Question 2: Material properties screen

fields?
104
ii. Screening Method Section

1- Do you prefer the material screening method to be :
a. A generalized screening method like by selecting material class

(Metals, Plastics, Ceramics, Composites, Wood. .etc)
b. A properties screening method by giving values to one or more
material properties.
c. I would like to choose every time I make a selection between
the two screening methods.
2- According to the material properties screening method do you

think that it should be based on:
a. Minimum values.
b. Maximum values
c. Both
Question 3: Criteria weighting screen

Information Section
fields?
Question 4: Final ranking screen

Scoring Results Section
a. Are the scoring results clear and understandable? If no please
comment.
b. Would you add or remove any information presented?
105
Question 5: Process update screen

Information Section
fields?
Question 6: Material update screen

Information Section
fields?
Question 7: Machine update screen

Information Section
fields?
Question 8: Finishing options update screen

Information Section
fields?
106
3- General Questions
Question 1:
The Decision Support System (DSS) has been designed to choose process
first than material. Do you think that this is appropriate? Please comment
on your answer.
Question 2:
Did you find the choice of the finishing methods that you want useful?
Please comment.
Question 3:
Did you find the DSS helpful when you have applied it to your case study
for selecting:
i. The possible processes chains.
ii. Materials
iii. Machines
iv. Finishing methods
Question 4:
When the DSS indicates that the part cannot be made using additive
manufacturing process, is the information provided clear?
Question 5:
manufacturing material, is the information provided clear?
Question 6:
Did you find that the DSS provides sufficient information on the ranking
and the choices?
Question 7:
How important is the ability of updating the process chain, material,
finishing and machines in an additive manufacturing DSS?

107
a. Not Important
b. Important
c. Very Important
d. Extremely Important
Question 8:
What elements do you like to add in the DSS?
Question 9:
What did you like most?
Question 10:
What did you dislike most?
Question 11
How do you rate the following elements in the DSS:
Elements Rating
Process selection Weak Fair Good Very Good Excellent

Material selection Weak Fair Good Very Good Excellent
Intermediate material Weak Fair Good Very Good Excellent
selection
Finishing selection Weak Fair Good Very Good Excellent
Machines selection Weak Fair Good Very Good Excellent
Process update Weak Fair Good Very Good Excellent
Material update Weak Fair Good Very Good Excellent
Finishing update Weak Fair Good Very Good Excellent
Machine update Weak Fair Good Very Good Excellent
DSS clarity Weak Fair Good Very Good Excellent
DSS length Weak Fair Good Very Good Excellent
Question 12:
Looking long term, do you think that the best form for this DSS is:
a. A stand alone application c. A CAD Plug-in.
b. Other (please state) d. Online application
108
5.1.2 Questionnaire Observations and Analysis
This section highlights the most important remarks of the users for the three parts
of the questionnaire and discusses it.
1. User Profile
Eleven out of thirteen users did not use any AM selectors which point that
AM selectors in general are not very popular and users do not rely on them
in practical.
Users select AM processes, materials and finishing methods based on own

experience, product needs, cost, lead time and service bureaus
recommendations.
Users select AM machines based on build size, build time, availability,

knowledge and service bureaus recommendations.
2. AMDSS Screens
Process screen
Users had the following suggestions for improving the process screen:
Sorting the fields of the screen in more logical order: length, width, height,
quantity, accuracy, surface finish and minimum wall thickness instead of
the current order which is: minimum wall thickness, accuracy, surface
finish, quantity, length, width and height.
Adding part volume.
Adding delivery time.
Adding platform shape.
Adding overhung feature.
Changing the wall thickness values to be thin < 2 mm, average from 2 mm
to 6 mm and thick > 6 mm.
Changing the surface finish terminology to be smooth, average and rough

instead of excellent, good average and average.
109
Comments
It is convenient to do the following changes to the process screen: sort the

fields in more logical order, add part volume and change the surface finish
terminology.
Adding platform shape is of no value because AM can in principal address

all shapes. The AM technologies can do any complex shapes which is the
biggest advantage over the conventional manufacturing (Valentan et
al.,2011).
Adding overhung features is not a main concern because after discussing

with experts and service bureaus all AM technologies can overcome this
difficulty by a way or another. The idea that geometrical issue is less
important is presented by Hague et al. (2004) where they stated that any
complex shapes or features produced by CAD can be translated directly to
the final part.
Changing the wall thickness values as proposed from one of the users to be
thin < 2 mm, average from 2 to 6 mm and thick > 6 mm is not helpful as
the range used in the system is the most discriminatory for process choice.
Material Properties Screen
Users had the following suggestions for improving the material properties screen:
Renaming sanitary conditions to medical consideration because sanitary

condition term is not clear.
Adding alternative units for material properties.
Adding flexural modulus, biocompatibility, environmental impact,

porosity and hygroscopic (ability of material to hold water) properties.
Users had the following preferences for the screening method:
Choosing every time users make a selection between the general class
screening and the properties screening.
Having both minimum and maximum values when making selections.
110
Making close matches materials appears on the possible material list so

compromising can be done if the values of properties are near.
Comments
Renaming sanitary conditions to medical consideration or bio-compatible

was considered a good proposal.
Having alternative units is not very useful because conversion can be done
easily outside the AMDSS.
Adding flexural modulus and environmental impact is not possible

because of data unavailability.
Adding porosity and hygroscopic was a good suggestion but not possible
for all materials because of data unavailability.
Both general class screening and the properties screening methods will be
used.
Using ranges (minimum and maximum) could overcome the problem of

close matches material.
Having both selections (selection based on minimum and selection based

on maximum) would give a flexibility aspect to the AMDSS.
Criteria Weighting Screen
Users had the following suggestions for improving the criteria weighting screen:
Adding build time.
Adding impact strength and elongation.
Adding cost or relative ranking of indicating cost (high, medium, low).
Comments
Many users have seen the criteria presented sufficient.
Adding elongation and or impact strength would not be possible because

of data unavailability.
111
Adding cost would be misleading to users because there is a difference

between cost and price, so if a user takes a recommendation from the
developed system and goes to a service bureau he will find different
values.
Final Ranking Screen
Users had the following suggestions for improving the final ranking screen:
Adding indicating cost.
Adding information on material type such as resin, nylon.
Some users did not want the score to be shown, while other wanted to see
the breakdown of the score.
Comments
Adding indicating cost was a good suggestion but could be misleading to

users and giving inaccurate information.
Adding information on material type or class was a good proposal.
Showing the breakdown of the score was considered a good proposal.
For the process update screen and material update screen almost all the
requirements of the users are the same like the process screen and material screen.
Machine Update Screen
Users had the following suggestions for improving the machine update screen:
Adding beam diameter for laser machines because this is an indicator to

accuracy.
Adding machine availability.
Comments
No need to add accuracy in machine because it was presented in the

process screen.
112
Machine availability differs from one place to another so it is difficult to

be included in the system.
Finishing Update Screen
Users had the following suggestions for improving the finishing update screen:
Adding the achieved finishing level from the finishing method.
Adding finishing time of the finishing method.
Adding information about the surface the finishing is applied to such as

raw material or paint or lacquer and specifying surface reflectivity.
Comments
Quantifying levels of finishing and knowing if the finishing method is

applied to a raw metal or lacquer or paint were considered good
suggestions.
Adding finishing time is a good point but could be difficult because of the
lack of information and also the variability from product to another.
3. AMDSS Screens
The general questions part of the questionnaire helped in understanding the

weaknesses and strengths of the AMDSS. The following are some remarks and
observation on the users answers:
Despite selecting process before material or material before process is the

same because at the end users have to choose both, many users prefer to
have the options to choose the order of the selection.
The finishing options part was not helpful for many users and needed
enhancement to give more information and assistance. Suggestions like:
quantification and characterization, costing, ranking of the finishing
methods, linking to applications and more process performance were all
considered good suggestions that cannot be all done because of the data
unavailability but still some improvement can occur.
113
Advanced users did not see that the AMDSS helped them in the AM
selection in general because they already know the answers. This is normal
because they were not the target users for the system. The target users
were the inexperienced and average users.
One advanced user commented in the material selection that he has known
a new material that he did not considered before. This shows that the
available database could help advanced users to be updated with new
materials instead of being stacked to the material they already know and
frequently use.
Majority of the users in the answer to the importance of system updating

chose very important and extremely important. An updatable system
without any programming required is one of the AMDSS strength points.
Adding graphical examples and showing more help was considered an

important point.
Adding economic consideration was also a good suggestion. Although, it

is difficult to develop a costing method that is realistic and could be
applied correctly. There is a difference between price and cost.
Users liked many things about the developed system such as: simplicity,
easiness, range of materials, possibility of being updated and being a good
source of information.
Users did not like incapability of getting the close matches properties
when selecting materials, lack of help information, different size of
windows and screens, not having a good graphical interface, lack of
guidance, having too many pop-up menus, changing the positions of back
and next buttons and the lack of showing the breakdown of the final
ranking score.
Users were asked to rate the following elements: process selection,

material selection, intermediate material selection, finishing selection,
machine selection, process update, material update, finishing update,
machine update, AMDSS clarity and AMDSS length with the following
ratings: weak, fair, good, very good and excellent.
114
Many users rated from weak to fair the finishing selection, finishing
update. The finishing selection is the weakest section in the AMDSS in
providing helpful information.
The choices of average and advanced users for the possible future of the
AMDSS were as follows: the online application was chosen eight times,
the CAD application was chosen six times, and the stand alone application
was chosen twice.
One user suggested that the update should not be the responsibility of the
users. This remark needs to be underlined. Most of the users liked the
update but they did not want to do it themselves. Some of them suggested
having the possibility to download the updates over the internet from a
trusted source.
5.2 User Testing and Evaluation
Four users have tangible products already manufactured and they have reached a
decision on how to manufacture them. The four users used the AMDSS to get a
recommendation and compared their selections to the system proposals. The aim
of this process was to test the AMDSS by users already made a selection for their
parts using their knowledge or by taking advice from experts.
5.2.1 Cover Part
Figure 5.1 shows the cover case study design. This case study was offered by
RCID which is a multi-disciplinary engineering consultancy operating based in
Newcastle University. Table 5-7 shows the part requirements and the material
requirements for the part. The cover needed to be water resistant.
115
Figure 5.1: Cover Part

Length ( mm) 140 Water resistance Yes
Width (mm) 87
Height (mm) 177
Minimum Wall Average- Wide (>2
Thickness (mm) mm)
Accuracy (mm) Average (0.1-0.25 mm)
Surface Finish Ra (m) Good-Average (0.32-
12.5m)
Quantity 1
Table 5-7: Part and Material Requirements for the Cover Part
When the part information was fed to the AMDSS, the processes that the system
offered were:
EBM
FDM
Laser Cusing
LENS
SLA
SLA + Investment Casting
SLA + Silicone Tooling + Vacuum Casting
116
SLS
When the material information was fed to the AMDSS, only the processes that
have water resistant material are shown in the list. Figure 5.2 shows the final
ranking screen for processes and the materials. The user selected the Nano tool
material and the SLA process. For the finishing process, sand blasting was
selected.
The outputs of the AMDSS conformed to the user selections for the process,
material, finishing methods and machine. It has to be underlined that the
selections of process and material made by the user were not the first ranked as
shown in Figure 5.2 because there are other factors that affect the selections such
as price given by service bureau and the availability of processes and materials.
The Nano Tool material was proposed by the service bureau to the RCID although
it is an expensive material. This ensures that recommendations of the service
bureaus sometimes are for their interest and not for the interest of users. This case
shows an example where the AM selectors would be very useful. User would have
been able to go with the final ranking screen to the service bureaus and asks them
about the cost of each alternative and then chose the optimum one instead of
having the most expensive one.
Figure 5.2: Final Ranking Screen of the Cover Part.
117
5.2.2 Drain Fitting Part
Figure 5.3 shows the drain fitting part design. Table 5-8 shows the part
requirements and the material requirements for the part. This case study was
offered by Keyworth Rapid Manufacturing Limited. The quantity required was 50
parts with a required strength of 50 MPa.
Figure 5.3: Drain Fitting Part.
Part Requirements Material

Length ( mm) 97.5 Requirements
Strength 50
Width (mm) 66 (MPa)
Height (mm) 43.5
Minimum Wall Thin-Average (0.5 -2
Thickness (mm) mm)
Surface Finish Ra (m) Good-Average (0.32-
12.5m)
Quantity 50
Table 5-8: Part and Material Requirements for the Drain Fitting Part.
When the part information was fed to the AMDSS, the processes that the system
offered were:
118
SLS
The user then input the material information. The output of the system was the
final ranking screen which showed the processes and the materials which could
manufacture the part as shown in Figure 5.4 The user selected the PA2200 PA12
material and the SLS process. For the finishing process, sand blasting was
selected. The outputs of the AMDSS conformed to the user selections for the
process, material, finishing method and machine. It has to be underlined that the
selections of process and material made by the user were not the first ranked as
shown in Figure 5.2 because there are other factors that affect the selections such
as price given by service bureau and the availability of processes and materials.
The price was the main factor that made the user chose the PA2200 PA12 and also
the availability of an SLS process in-house.
Figure 5.4: Final Ranking Screen for The Drain Fitting Case Study
5.2.3 Filter Cover Part
Figure 5.5 shows the filter cover case study design. This case study was offered by
Paragon Rapid Technologies Limited a rapid prototyping service bureau. Table
5-9 shows the part requirements and the material requirements of the part. The
material requirements were: black colour, a heat deflection temperature of 110 C,
a hardness (shore D) of 90, and a need to be water resistant.
119
Figure 5.5: Filter Cover Part.

Length ( mm) 148 HDT (C) 110
Width (mm) 30 Colour Black
Height (mm) 148 Hardness (shore D) 90
Minimum Wall Thin-Average (0.5 -2 Water Resistance Yes
Thickness (mm) mm)
Surface Finish Ra (m) Good-Average (4 m)
Quantity 1
Table 5-9: Part and Material Requirements for the Filter Cover Part.
When the part information was fed in the AMDSS, the processes selected were:
EBM
Laser Cusing
LENS
SLA
SLS
120
When the material information was fed to the system, Figure 5.6 appeared and
showed that there was conflict between colour and hardness. There was no black
material that also had hardness shore (D) equal to 90. Because many materials
could be pigmented at the end of the process, the user decided to relax this
constraint. The back button was pressed and the black colour was removed.
Figure 5.6: Screen that Shows the Material Parameters in Conflict Together.
Figure 5.7 shows the possible processes and materials found by the system after
removing the black colour requirement. The only process was SLA and the only
material was a material called Nano Tool. Although, the user of this case study
was an expert user he did not know about this material before. This underlines that
sometimes the AMDSS system could be beneficial to the expert users not from the
point view of the logic of selection but from the point of view of providing new
updated information.
The real selection of the process and the material for this part was SLS process
and nylon 12 (PA2200 PA 12) and then pigmented to black color. Looking at the
material data sheet of the PA2200 PA12 material, a remark was made that the
hardness is 75 shore D and not 90. The expert user when making the actual
selection had compromised on hardness. So the back button was pressed and the
hardness requirement was relaxed from 90 to 75.
121
Removing the Black Color Requirement
Relaxing Hardness Requirement.
When the hardness was changed to 75, the SLS process and the PA2200 PA12
appeared in the list. The user commented that adding a range of minimum and
maximum values would help. So in this case instead of having one value which
was the 90, it could be a range of values between 75 and 90.
122
Finally, the user selected sanding as the finishing process and chose the EOS P730
machine for part manufacturing.
After exploring the decision making process, the outputs of the AMDSS
conformed to the user selections for the process, material, finishing method and
machine. It has to be underlined that the selections of process and material made
by the user were not the first ranked as shown in Figure 5.8 because there are
other factors that affect the selections such as cost of the service bureau and the
availability of processes and materials. In addition, the color of material is not
very important in most of the cases because many materials could be pigmented
easily within a surface finish process.
5.2.4 Dental Implant Part Assembly
Figure 5.9 shows the dental implant part design. This part was one of several parts
assembled together. This case study was offered by one of the research candidate
at Newcastle University named Mr. Shah Fenner Khan. Table 5-10 shows the part
requirements and the material requirements for the part. The required heat
deflection temperature was 90 C.
Figure 5.9: Dental Implant Assembly Part.
123

Length ( mm) 36 HDT (C) 90
Width (mm) 47
Height (mm) 10
Minimum Wall Thin -Average(0.5-2.00
Thickness (mm) mm)
Accuracy (mm) Tight <0.1 mm
Surface Finish Ra Average Rough(>12.5 m)
(m)
Quantity 1
Table 5-10: Part and Material Requirements for the Dental Implant Assembly Part.
When the part information was fed to the AMDSS, the processes for this part were
as follows:
EBM
Laser Cusing
LENS
SLA
When the material information was fed to the AMDSS, only SLA was
recommended because only some of the SLA materials could achieve the heat
deflection temperature (HDT) of 90 C. The user of this case study gave the HDT
criterion weight of 10 and left the rest criteria empty. Figure 5.10 shows the final
ranking screen which showed the ranked processes and the materials that can
manufacture the part. The user selected the Protogen 18920 material and the SLA
process. For the finishing process, polishing was selected and finally for the
machine Viper SLA was selected. The outputs of the AMDSS conformed to the
user selections for the processes, materials, finishing methods and machine. It has
to be underlined that the selections of process and material made by the user were
not the first ranked as shown in Figure 5.10 because there are other factors that
affect the selections such as price given by service bureau and the availability of
processes and materials. The price was the main factor here.
124
Figure 5.10: Final Ranking Screen of the Dental Implant Part.
Comments
The responses of the users to the questionnaire and the AMDSS testing
with the four given parts showed that there is a need for further
development and more work is needed so the developed system could
present what customers and users need. Chapter 6 describes how the
AMDSS was further developed.
The testing of the AMDSS with the given four parts showed that the AM
selector could be useful for the inexperienced and average users because it
shows them all possible AM manufacturing routes for their parts regarding
process chains, materials and finishing options which saves them time and
cost and provides them with the information needed before going to
service bureaus to manufacture a part.
125
Further Development and Assessment
This chapter describes the seventh and eighth steps respectively in the AMDSS
design: further development made to the AMDSS and obtaining further feedback
from users for assessment purpose. The feedback was gathered through making a
second questionnaire which was a modified version of the first questionnaire. A
total of six persons responded to the questionnaire. Three of them were advanced
users and the other three were average users. Two of the six users have responded
previously to the first questionnaire and the other four were new users and have
not tested the system before.
6.1 AMDSS Second Version
6.1.1 List of Modifications
The following are the modifications that were made to the first version of the
AMDSS based on the inputs of the users:
1. Option of selecting process first or material first.
2. Option of selecting material by class (polymers, metals and ceramics).
3. Volume calculations are shown when users enter the length, width and height
of the part.
4. Rearranging the order of the inputs of the process selection screen to be

length, width, height, quantity, accuracy, surface finish and minimum wall
thickness.
5. Renaming the surface finish ranges to be: smooth, average and rough.
6. Adding the maximum values to some of the inputs of the materials properties
screen (strength, modulus, hardness shore D and heat deflection temperature).
7. Adding new inputs to the material properties screen like: hardness Vickers
and melting point. Both new inputs have minimum and maximum ranges.
8. Renaming sanitary conditions to bio-compatible in the material properties

screen.
126
9. Changing the sequence of the windows starting from the final ranking screen
to be in the following order: 1- selecting intermediate material, 2- selecting
finishing methods, 3- selecting machines and 4- previewing the technical
summary. There is a possibility that users can skip any step and go to the
technical summary directly.
10. Improving the finishing selection by:
i. Changing the finishing to be finishing methods path instead of single

finishing method.
ii. Adding the surface finish level (Ra) that can be achieved by the
selected path.
iii. Adding finishing advice that could be updated.
11. Changing the updating panels to match the modifications made to the new
improved version of the AMDSS.
12. Adding help files in some screens to guide users and give them appropriate
information.
13. In ranking the alternatives with SMART method there is a possibility to

revert the direction of sorting for some criteria such as density, hardness and
dielectric strength. These criteria could to be ranked considering that
minimum values are better or considering that maximum values are better
(which is the default) because some users would request that lower density
and/or lower hardness and/or lower dielectric strength would be better for
their designs.
6.1.2 Graphical User Interface Modifications
Some screens were changed and some screens were added based on the
modifications made. The first change made was to the first screen that appears
when users start the system. Figure 6.1 shows the revised start screen. There are
two options in the selection panel in the second version: 1- select process first and
2 select material first. In the first version, the start screen had only one button
called selection module and this was selecting process first.
127
The other screens that had been changed were described in the next section to
better understand the KBS modifications and the relationships between the KBS
and the GUI within the developed AMDSS.
Figure 6.1: Start Screen of the Second Version of the AMDSS
6.1.3 Knowledge Based System Modifications
The KBS consists of three parts which are: 1- selection, 2- browse and 3- update.
The modifications were made to the selection part. Consequently, the update part
needs to be changed to cope with the changes made in the selection part. No
changes have been made to the browse part. The following sections discuss: 1- the
selection part modifications and 2- the update part modifications.
1. Selection Part Modifications
IDEF0 model development for the selection part of the AMDSS second
version
An IDEF0 model was developed to describe the selection part of the second
version of the AMDSS, which was a modified version of the first version. Figure
128
6.2 shows the IDEF0 tree of the second version and Table 6-1 shows the IDEF0
node index. In the IDEF0 diagrams the modified and new activities have been
shaded to show the changes were made.
A-0
A0
A1 A2 A3 A4 A5
A11 A12 A13 A21 A22 A23 A24 A25 A26
A111 A112 A113 A114 A115 A116 A121 A122 A123 A124 A125
Figure 6.2: IDEF0 Node Tree Diagram for the Second Version of the AMDSS
One important piece of information about the IDEF0 is that the activities in the
rectangular boxes do not have to be in sequence. When a person develops an
IDEF0 model the two options are available: building the boxes in sequence or
building the boxes without sequence (IDEF,2012). The need for building boxes
without sequence appeared in the second version of the AMDSS model in A0
diagram.
The sequence of the selection of AM processes and materials in the second

version could be made via four paths and not in only one path selection as for the
first version. The use of the block diagram to illustrate the new different selection
paths was essential to avoid repeatability in the IDEF0 model. Figure 6.3 shows a
block diagram of one path selection of the first version. Users made the selections
in the following order: process, material by properties, ranking, intermediate
materials, finishing methods and machines.
129
A0 Develop an Additive Manufacturing Decision Support System (AMDSS)
A-0 Main Steps of the Selection Part of the KBS
A1 Check AM possible processes and Material

A11 Check AM possible process
A111 Connect to the database and get the processes that
satisfy minimum wall thickness.
satisfy minimum accuracy.
satisfy minimum quantity.
satisfy minimum surface finish.
satisfy the required volume.
A116 Intersect all process lists.
A12 Check AM possible materials by properties
A121 Connect to the database and get all the material that
satisfy the general properties.
satisfy the mechanical properties.
satisfy the electrical properties.
satisfy the thermal properties.
satisfy the environmental properties.
A126 intersect all material lists.
A13 Check AM possible materials by class
A2 Rank using SMART method
A21 Enter the weights of each criterion from 1 to 10.
A22 Normalize the weights.
A23 Divide each alternative score value by the maximum
alternative score value for each criterion and multiply by 10.
A24 Revert the direction of ranking for some criteria if needed.
A25 Multiply normalized score values by the normalized weights.
A26 Sum the weighted score values for each alternative and sort.
A3 Check possible machines
A4 Check possible finishing methods
A5 Check possible intermediate materials

Table 6-1: IDEF0 Node Index (version2).
130
Select Select Select

Select Rank using Select
material by intermediate finishing
process SMART machines
properties materials methods
Figure 6.3: Selection Path of the First Version of the AMDSS
Figure 6.4 shows the block diagram of the new four paths. Users can select 1-
process then material by properties, 2- select process then material by class, 3-
select material by properties then select process and 4- select material by class
then select process. For all the four paths the rest of the path order are the same:
ranking, select intermediate materials, select finishing methods and select
machines.
Select
Select
material by
process
properties
Select
Select
material by
process
class
Select Select
Rank using Select
intermediate finishing
SMART machines
materials methods
Select
Select
material by
process
properties
Select
Select
material by
process
class
Figure 6.4: Selection Paths of the Second Version of the AMDSS
Figure 6.5 shows the node A-0 of the second version which is the same as the first
version A-0 node. Figure 6.6 shows the A0 diagram which explains the revised
main steps of the selection part of the KBS. Figure 6.7 shows the A1 diagram
which is: check the AM possible processes and materials. Figure 6.8 shows the
A11 diagram which is: Check AM possible processes. The only change is in
activity five. The KBS checks the volume of the processes and not each dimension
by itself like in the first version. Figure 6.9 represents the A12 diagram which is:
Check AM material by properties.
131
The modifications to the A12 diagram were: 1- New inputs added to the
mechanical properties box such as: maximum hardness (shore D), minimum
hardness Vickers, maximum hardness Vickers, maximum modulus and maximum
strength. 2- New inputs added to the thermal properties such as: maximum heat
deflection temperature, minimum melting point and maximum melting point. 3-
Replacing the sanitary conditions with the bio-compatible condition.
Figure 6.10 shows the A2 diagram: Rank using SMART method. No changes
were made to this diagram.
132
Specific
Machines Material Finishing
Capabilities Properties Methods
Processes
capabilities
Possible Processes
Part Infromation DEVELOP AN ADDITIVE Possible Materials

MANUFACTRUING
DECISION SUPPORT Possible Finishing Methods
SYSTEM
A0 Possible Machines
Possible Intermediate Materials
DSS Developer
NODE: A-0 TITLE: DEVELOP AN ADDITIVE MANUFACTRUING DECISION SUPPORT SYSTEM (DSS) NO.: 1
Figure 6.5: A-0, Develop an Additive Manufacturing Decision Support System (Version2)
133
Processes Material
capabilities Properties
Report materials problems

Part Check AM
Information Possible
Processes and
Materials Report process problems
1
A1
Rank using Specific

SMART method Machines
Capabilites
2
A2
Possible LM processes+ materials
Check possible Possible machines

machines
Ranked processes+ materials
Finishing
3 Methods
Check possible
finishing Possible finishing methods
methods
Intermediate materials
4
Possible
Check possible intermediate
intermediate materials
material
5
NODE: A0 TITLE: Main Steps of the Selection Part of the KBS NO.: 2
Figure 6.6: A0, Main Steps of the Selection Part of the KBS (Version 2).
134
Processes
capabilities
Part Possible AM processes

Information Check AM
possible
processes
1 Report process problems
A11 Material
Properties
Possible AM materias
Check AM
Required material properties possible
Materials by
properties Report material problems
2
A12 Material
classes
Possible AM materias
Required material class Check possible
materials by
class
3
Report material problems
NODE: A1 TITLE: Check AM processes and Materials NO.: 3
Figure 6.7: A1, Check AM Processes and Materials (Version 2).
135
Min wall
thickness connect to the database and
get the processes that satisfy
Processes list that satisfy wall thickness requirement
the min wall thickness
1

Accuracy level get the processes that satisfy
the required accuracy level Processes list that satisfy accuracy level requirement

Surface finish level get the processes that satisfy
the required surface finish Processes list that satisfy surface finish level requirement
3

Required quantity get the processes that satisfy
Processes list that satisfy quantity requirement
the required quantity
4

Required dimensions (X,Y,Z) get the processes that satisfy Processes that satisfy volume requirement
the required volume
5 Report process problems
Intersect all process lists

Possible AM processes
6
NODE: A11 TITLE: Check additive manufacturing possible processes NO.: 4
Figure 6.8: A11, Check AM Processes (Version 2).
136
Min density Connect to the database and get
the processes and materials that Materials list that satisfy the
satisfy the general properties selected (if any) general
Color
1 properties
Min/Max hardness
(Shore D)
Connect to the database and get the Materials list that satisfy the
Min/Max strength processes and materials that satisfy selected (if any)
Min/Max modulus the mechanical properties mechanical properties
Min/Max hardness 2
(Vickers)
Connect to the database and get the

Materials list that satisfy the
Min Dielectric strength selected (if any) electrical
properties
3
Min/max heat deflection temperature Connect to the database and get the Materials list that satisfy the
processes and materials that satisfy the selected (if any) thermal
Min/ Max melting point
thermal properties properties
Heat resistance
4
Connect to the database and get the Materials list that satisfy the
Bio-compatible condition selected (if any)
enviromental properties environmental properties
Water resistance
5 Possible
AM
Intersect the selected materials
material lists
NODE: A12 TITLE: Check AM possible materials by properties NO.: 5
Figure 6.9: A12, Check AM Materials by Properties (Version 2).
137
Possible LM
Enter the
processes+
weights of
materials
each
criterion
from 1 to 10
Criteria
1
Normalized
Normalize
weights
the weights
Weights
2
Divide each alternative score value

by the maximum alternative score
value for each criterion and multiply
Score value for
by 10
each alternative
Normalized score
for each criterion 3
values
Multiply
normalized
Revert the score values by
direction of the normalized
ranking for some weights Weighted score
criteria if needed 5 values
Criteria that need 4
their ranking Sum the weighted
direction to be score values for Sorted
reverted each alternative and alternatives
sort
6
New Normalized Score Value
NODE: A2 TITLE: Rank using SMART method NO.: 6
Figure 6.10: A2, Rank Using SMART Method (Version 2).
138
Modifications to the elements of the selection part of the KBS
The selection part of the KBS consists of six elements as described before in
chapter 3 which are: the process chain selection element, the material selection
element, the ranking element, the finishing options element, the intermediate
material selection element and the machine options element. The modifications
were done to only four elements: the process chain selection element, the material
selection element, the ranking element and the finishing options element.
I. Process Chain Selection Element
Figure 6.11: Process Selection Screen of AMDSS (Version 2).
Figure 6.11 shows the process selection screen of the second version where users
enter the values of the process input variables. Based on feedback of the users the
order of the input variables have been changed to be length, width, height,
quantity, accuracy, surface finish and minimum wall thickness instead of
minimum wall thickness, accuracy, surface finish, quantity, length, width and
height in the first version. In addition, the terminology of the surface finish ranges
were changed to be smooth (0 - 0.32 m), average (0.32-12.5 m) and rough
139
(>12.5 m). The terminologies in the first version were excellent (0 - 0.32 m),
good-average (0.32-12.5 m) and average rough (>12.5 m).
The volume calculation is added to the screen. When users enter values of X, Y
and Z, the volume is calculated automatically and appears on the screen.
Furthermore, the selection of the possible processes in the first version was based
on comparing the X, Y and Z fed to the system by user and comparing them to the
maximum X, Y and Z available in the database (parameter by parameter). This
logic has been changed in the second version. The selection is based on a
comparison between the required volume and the available volume. The processes
that are not satisfying the volume requirement are excluded from the possible
processes list.
There are two issues regarding the volume calculations that have to be mentioned:
First, the orientation of the part was not taken into consideration. This means that
sometimes the part will fit in a specific volume but this is not the optimum build
scenario for it. The reason of this issue is that the process planning was not
included in the system and is beyond the scope of this research. Secondly, the
filtering of the possible processes using volume can fails in some circumstances.
For example, in an extreme conditions such as the a process that has very small
dimensions table X, Y and very high Z the volume calculation would tell the user
that the process could meet his requirements but in reality the X and Y of the part
cannot be laid down on the process table because the X and Y of the part are
bigger than X and Y of the table.
Process selection logic flowchart
Figure 6.12 shows a flowchart that describes the steps of the process selection. A
modification was done that the KBS deals with the calculated volume. The shaded
part in the flowchart shows the new steps: get from the database all the processes,
calculate the volume of each process and get a list of the processes to which their
volume is bigger than or equal to the required volume.
A decision then has to be made: if the processes list is empty then add to the
reasons list that there is no process in the database can meet the required volume
and continue to the next step but if the processes list is not empty then the KBS
140
continues the next step directly which is intersecting all the process lists to get a
process list that satisfies all process variables.
Start B

Is the list of the No layer manufacturing
processes empty? can meet your required
Create an empty reasons list volume
For i =1: 4, get the value of the variable (i) from No

the workspace (i=Wall thickness, Accuracy, New process list = Intersect all the lists of the
Surface finish and quantity) processes
Get list (i) of the processes that their variable (i)

is like or bigger than the required variable (i)
Is the Yes
Is new process Yes
reasons list
list empty?
empty?
Yes Add to the reasons list :
processes empty? No No
can meet your variable (i)
Display new process list

No
i=i+1
Display
reasons list
No
Is i=5? Display
Intersection
table
Yes
Get the value of the required volume from the

workspace
Get from the database all the processes
Calculate the volume for each process
Get a list of the processes that their volume

equals or bigger than the required user volume
Figure 6.12: Checking AM Processes Flowchart of the AMDSS (Version 2).
141
II. Material Selection Element:
The second version of the material selection element has two options for the
material selection as shown in Figure 6.13: 1- selecting material by properties,
when selected a new window is opened shown in Figure 6.14 and 2- selecting
material by class such as polymers, metals and ceramics, when selected a new
window is opened shown in Figure 6.15.
Figure 6.13: Choosing Material Screening Method of the AMDSS (Version2).
Figure 6.14: Material Properties Screen of the AMDSS (Version 2).
142
Figure 6.15: Material Class Screen of the AMDSS (Version 2).
Selecting material by properties
In the second version, new inputs were added to the material input screen. These
inputs were: 1- maximum strength, 2- maximum tensile modulus, 3- maximum
hardness (Shore D), 4- minimum hardness (Vickers), 5- maximum hardness
(Vickers), 6-maximum heat deflection temperature, 7- minimum melting point, 8-
maximum melting point. The idea of adding the maximum ranges of some
variables is that some users or designers would like to have a range instead of just
having a minimum value. In addition some users would use the maximum only.
For the hardness, users could select only one type of hardness either shore D or
hardness Vickers, they cannot choose both.
Selecting material logic flowcharts
Figure 6.16 and Figure 6.17 show flowcharts of the assessment of AM materials
by the second version of the AMDSS. The flowcharts are modified flowcharts of
the first version. The modifications are shaded to allow them to be distinguished.
The KBS obtains the material variables from the workspace. The material
variables are divided into four groups.
The first group is (minimum values group). It includes six variables which are:
minimum density, minimum strength, minimum modulus, minimum melting
point, minimum dielectric strength and minimum heat deflection temperature.
Compared to the first version, hardness was removed from this group and
minimum melting point was added because hardness will need some more
programming steps.
143
The second group is the (like values group). It includes four variables: colour, heat
resistance, bio-compatible and water resistance. Compared to the first version, the
bio-compatible variable has replaced sanitary conditions variable.
The third group is the (maximum values group). It includes four variables:
maximum strength, maximum modulus, maximum melting point and maximum
heat deflection temperature. For this group, the KBS obtains from the database a
material list that has values less than or equal the required values set by users.
The fourth group is the (hardness group). In the second version, the hardness is a
special case because in the first version the used hardness was the Shore D
hardness and only the minimum value but in the second version includes two
types of hardness which are: Shore D and Vickers. Furthermore, users can enter
minimum and maximum value for any type of hardness. There is a constraint that
user can select only one type when he enters the hardness values. if user enters a
value for minimum Shore D hardness, automatically the boxes of Vickers
hardness both minimum and maximum would turn inactive.
For the first group of variables the KBS connects to the database and gets a list of
the materials (within the feasible processes) that their variable values exceed the
required minimum variable. The same logic is used for the second group of
variables except that when the KBS connects to the database, it gets a list of the
materials (within the feasible processes) that their variable values are less than the
required maximum variable. For example, if a maximum melting point is set by a
user the KBS connects to the database and gets all the materials for which their
melting point is less than or equal to the required melting point.
For the (hardness group) some new boxes in the flowcharts in Figure 6.16 and
Figure 6.17 that need to be explained. These boxes tackle the hardness variable.
First the KBS gets from the database a list of all the materials within the
successful processes. Next, interpolation is done to transfer all the Vickers
hardness to shore D hardness. Next, the two columns of the interpolated Vickers
hardness and the shore D hardness are merged together to form one column in
shore D hardness units.
144
There is now a new material list called material list (a) that has one Shore D
hardness column. At this point a decision has to be made, the KBS checks if the
user entered a shore D value or not? If not it means that he entered Vickers value.
Consequently, interpolation is used to transform the required Vickers value(s) to

shore D value(s). Next, a list of the materials for which the hardness is greater
than or equal to the required minimum hardness is retrieved from the database. A
check is made if the material list of the hardness is empty, if yes the KBS adds to
the reasons list that no AM material within the database could meet the required
minimum hardness.
Next a list of the materials for which their hardness is less than or equal to the
required maximum hardness is retrieved from the database. A check is made if the
material list of the hardness is empty, if yes the KBS adds to the reasons list that
no AM material within the database could meet the required maximum hardness.
Selecting material by class
When users choose to select material by class the window shown in Figure 6.15
appears. Users can select polymers, metals, or ceramics. If a user selects polymers
the KBS connects to the database and all the materials that are classified as
polymers. If a user selects metals the KBS connects to the database and retrieves
all the materials that are classified as metals. Finally, if a user chooses ceramics
the KBS connects to the database and retrieves all the materials that are classified
as ceramics.
145
Start B

No
C Is j= 5?
For i=1: 6 get the value of the variable (i) from the
workspace (i=minimum value of density, strength, Yes
modulus, melting point, dielectric strength, and
HDT) For k=1: 4 get the value of the variable (k) from
the workspace (k=maximum value of strength,
modulus, melting point and HDT)
Get a list (i) of the materials that their variable (i)
equals or bigger than the required variable (i)
within the feasible processes
Get a list (k) of the materials that their variable (k)
equals or less than the required variable (k)within
the feasible processes

Is the list (i) of the Add to the
materials empty? reasons list: No
can meet your variable (i)
layer
Yes manufacturing
Is the list (k) of the
materials empty? can meet your
variable (k)
No
i=i+1
No
k=k+1
No
Is i= 7?
No
Is k= 5?
Yes
For j=1: 4 get the value of the variable (j) from the
C workspace (j= colour, heat resistance, bio- Yes
compatible and water resistance)
Get the values of min hardness Vickers, min
hardness Shore D, max hardness Vickers and
max hardness Shore D from the workspace
Get a list (j) of the materials that their variable (j)
is like the required variable (j) within the feasible
processes
Get a list of all the materials within the feasible

processes

Is the list (j) of the
materials empty? Use interpolation to transfer all the Vickers
can meet your variable (j)
hardness to Shore D hardness
No
Merge the interpolated Vickers hardness with the
j=j+1 Shore D hardness column to be in one single
column (shore D hardness)
B D
Figure 6.16: Checking AM Materials Flowchart (Version 2).
146
material list (a)=List all the materials within the

successful processes with one hardness column
Shore D (merged Colum)
Did the user No

Use interpolation to transform the Vickers
enter value(s) for
entered value(s) to shore D value (s)
the Shore D?
Yes
Get a list of the materials from the material list (a)

that their hardness equals or bigger than the
required minimum hardness
Add to the reasons list: No

Yes
Is the list of the layer manufacturing can meet
materials empty? your required minimum
hardness
No
Get a list of the processes that their hardness

equals or less than the required maximum
hardness
Add to the reasons list: No

Yes
Is the list of the layer manufacturing can meet
materials empty? your required maximum
hardness
No
Check what material properties are selected
New material list = Intersect all the selected

material lists
Is the Yes
Is new material Yes list of
Display Intersection table
list empty? reasons
empty?
No No
Display new material list Display reasons list
Figure 6.17: Checking AM Materials Flowchart (Version 2 Continued).
147
III. Ranking Element:
The third element in the KBS selection part is ranking. In the first version, the
ranking was done considering that the maximum values are better than the minim
values. An important modifications in the second version is that the possibility
that some criteria such as density, hardness and dielectric strength would be
ranked in a reverted direction. This means that the minimum values are better than
the maximum values. Some users would needs in their parts that lower density
(and/or) lower hardness (and/or) lower dielectric strength is better. A new criteria
screen was designed as shown in Figure 6.18. The new screen includes three radio
buttons, when any of them is pressed means that the user needs to sort this
criterion according to the minimum values are better than the maximum values.
Figure 6.18: Criteria Screen (Version 2).
Figure 6.19 and Figure 6.20 show flowcharts that explain how the SMART
method was used. The modifications are shaded. The melting point (usually for
metal or ceramics) of the material and the heat deflection temperature (usually for
plastics) are merged together into one column called temperature column. Next
the normalized value is calculated by dividing the temperature of each material to
the maximum temperature in this column. The result is multiplied by ten.
148
The merging is done only if the material selection is by properties but if the
material selection is by class the temperature column takes the heat deflection
temperature column in case of polymers class and the melting point column in
case of metals class and ceramics class without any merging.
Furthermore, to rank some criteria in another direction the logic is as follows: if a

user presses a radio button for criterion (i) a variable called x (i) is assigned to 1.
If the user does not press the buttons so x (i) is assigned to 0. If x (i) =1 then this
equation is used:
Norm.score value criterion (i) = [[Norm.score value criterion (i) -10] x -1]
In order to the reverting process to take place, the new normalized score values
take effect and replace the old normalized score values for the selected criteria and
the ranking process continues normally after that.
149
Start B
Get the processes and materials list that satisfy Temperature of each material
the user requirements from the workspace Normalized Score Value 10
Maximum temperature
(successful candidates)
Get the maximum strength of the successful

candidates
Get all the wall thickness(W.T), accuracy(Acc)

Strength of each material
Normalized Score Value 10 and surface finish (S.F) of the successful
Maximum strength candidates
Get the maximum modulus of the successful Yes

Is W.T= Normalized score
candidates
Average Wide? value = 4
No
Modulus of each material
Maximum modulus
Is W.T= Thin Yes Normalized score
Average? value = 7
Get the maximum hardness of the successful

candidates No
Normalized score
value = 10
hardness of each material

Normalized Score Value 10 No
Maximum hardness
Yes Normalized score

Is Acc= loose?
value = 4
Get the maximum density of the successful
candidates
No
Is Acc= Yes Normalized score

density of each material
Normalized Score Value 10 Average? value = 7
Maximum density
No
Get the maximum dielectric strength(D.E) of the Normalized score

successful candidates value = 10
No
D.E of each material Is S.F= Yes

Normalized Score Value 10 Normalized score
Maximum D.E Average
value = 4
Rough?
Merge the heat deflection temperature(HDT) and No

the melting point into one column: temperature
column. C D
Get the maximum temperature from the

temperature column of the successful candidates
Figure 6.19: SMART Ranking Flowchart (Version 2).
150
C D
Yes
Is S.F= Good Normalized score
Average? value = 7
No
Normalized score
value = 10
Get all the weights of the users for each criterion
criterion does not Yes

Weight =1
have a weight?
No
Weight = Users weight
For i=1:3, get user selections for the ranking

directions x(i) (binary variable) (i=hardness,
density and dielectric strength)
No
Normalized score value of criterion (i)=
Is x(i) =0?
[Normalized score value of criterion (i) -10] x -1
Yes
i=i+1
No
Is i=4?
Yes
Weighted score value =
normalized score value for each criterion X users
weight for this criterion
Score = Sum all the weighted score values
Rank the scores in descending order
Display ranked scores
Figure 6.20: SMART Ranking Flowchart (Version 2 Continued).
After the KBS finished ranking a final ranking screen appears as shown in Figure
6.21. The processes and the materials are ranked in descending order. Users have
four further steps appeared in the following order: 1- selects intermediate material
151
(in case the process needed an intermediate material), 2- selects finishing

methods, 3- select machines and 4- goes directly to technical summary. Any step
can be skipped. In addition a button was added for the users who which to see the
SMART calculations in details as shown in Figure 6.22. This is a modified screen
based on the feedback of the users. In the first version, users needed to make
selections for the intermediate materials for example and then pressed back to the
final ranking screen again and the same for the finishing and the machines. Many
users did not like to go back after going forward. The second version made the
AMDSS selections go into one direction and any step can be skipped based on
user preferences.
Figure 6.21: Final Ranking Screen of the AMDSS (Version 2).
Sum product of these 2 Score of the

= alternative
columns
Figure 6.22: SMART Calculations (Version 2).
152
IV. Finishing Element:
The finishing element was changed in the second version of the AMDSS. First of
all, the finishing table contains a finishing method path and not a single method.
In addition, the finishing level that could be obtained from the finishing path is
displayed. Furthermore, there is a general finishing advice that could help users in
their selections of the best finishing methods. When a user selects a process like
SLS for example, and then selects choose finishing methods the KBS connects to
the database and gets all the finishing methods and the surface finish level that can
be achieved. Figure 6.23 shows an example of the finishing options screen for the
SLS process.
Figure 6.23: Finishing Options Screen in the AMDSS (Version 2).
2. Update Part Modifications
Due to the changes made in the selection part of the AMDSS, there were changes
also made to the update part.
Add a process
The only thing that has been changed in the add a process screen is that the popup
down menu of the surface finish input has changed to smooth (0 - 0.32 m),
average (0.32-12.5 m) and rough (>12.5 m) instead of excellent (0 - 0.32 m),
good-average (0.32-12.5 m) and average rough (>12.5 m) respectively.
153
Figure 6.24: Add a Process in the Database (Version2).
Add a new material
Figure 6.25 and Figure 6.26 show the modified add a new material screen and add
a new intermediate material respectively. It can be shown that hardness Vickers
and melting point was added. In addition, sanitary condition is replaced by bio-
compatible. When user enters values for these variables the KBS connects to the
database and adds the new material to the database with the given specifications.
Figure 6.25: Add a New Material (Version 2).

154
Add a new intermediate material
Figure 6.26: Add a New Intermediate Material (Version2).
Add a new finishing method
The last change is in the finishing update screen shown in Figure 6.27. It include
process name which means the AM process name, finishing methods so the user
can add a path and not a single method and finally the finishing level expected
from this finishing path.
Figure 6.27: Add a New Finishing Method (Version2).
155
6.2 Second Feedback
The purpose of the second feedback was to make assessment of the second
version of the AMDSS. The first feedback was for system improvement while the
second feedback was more for the system evaluation.
6.2.1 Second Questionnaire Design
The questionnaire contains two parts: the first part is the user profile and the
second part is general questions. The first part gives the researcher an overview
about the user knowledge about the AM technologies and processes. The second
part is a general question that gives information about the system overall by
asking users about the advantages and limitations of the system. As previously
discussed in chapter five, users were classified into two groups: advanced users
and average users. According to this classification there were three advanced users
and three average users. Table 6-2,Table 6-3, Table 6-4 and Table 6-5 show the
second version questionnaire.
156
1- User Profile
Question1:
How often do you need to select additive manufacturing process or
material?
a. Weekly
b. Monthly
c. Sporadically
d. Other:_______________________
Question2:
Have you used any knowledge based system for selecting additive process
or material before? If yes please name it
Question3:
b. Material
d. Machines
Question1:
Would you like to add any information to any of the AMDSS screens?
Question 2:
Would you like to remove any information to any of the AMDSS screens?
Question3:
Did you find the AMDSS helpful when you have applied it to your case
study for selecting:
I. The possible process chains.
II. Materials
III. Machines
IV. Finishing methods
Table 6-2: AMDSS Second Version Questionnaire (Page 1 of 4).

157
Question 4:
How important is the ability of updating the process chain, material,
finishing and machines to you in an additive manufacturing DSS?
a. Not Important
b. Important
c. Very Important
Question 5:
What elements do you like to add to the AMDSS?
Question 6:
Question 7:
Question 8:
How do you rate the following elements in the AMDSS:
Elements Rating

Intermediate material Weak Fair Good Very Good Excellent
selection
158
Question 9:
Looking long term, do you think that the best form for a DSS of this type
is:
a. A stand alone application
b. Online application
c. A CAD Plug-in.
d. Other (please state)
Question 10:
If you have used the first version of this system, how do you compare the
1st version to the 2nd version with regard to:
I. The process selection
a- Same
b- 2nd version better
c- 1st version better
Comment:
II. The material selection:
a- Same
Comment:
III. Finishing selection
a- Same
Comment:
IV. Machine selection
a- Same
Comment:
159
Question 11:
Please comment on the following capabilities of the AMDSS:
I. Selecting material first or process first.
II. Selecting by material class or by material properties.
III. Determining minimum and maximum values of the material
properties.
IV. Helping on the finishing strategies selection.
Question 12:
Do you have any suggestions for improving the AMDSS?
6.2.2 Second Questionnaire Observations and Analysis
The responses of the users were satisfactory in general and agreed that
improvement happened. Some users have some remarks and suggestions. It has to
be mentioned that these remarks are based on a relatively small sample (six users).
The reason of the small sample was because the purpose of the second
questionnaire was assessment rather than a feedback. In addition, some average
and expert users who tested the system on the first time did not have the time to
re-test it again. The following are some points that need to be underlined:
There was a suggestion to add more machines. The system was designed
to be updatable so any user can add his required machines.
There was an opinion to remove some SLA materials from the database
because the present list is very exhaustive. The system was designed to be
editable so users can remove any things that they do not want. In addition,
this is a general system so perhaps some users would like to have more
options and choices.
Most of users of the second questionnaire found the AMDSS helpful when
they have applied it to their case studies and tried to make a selection
regarding process chains, materials, machines and finishing methods.
160
One user found the updating important, another user found it very
important and four users found it extremely important.
Some users suggested adding to the AMDSS: 1- matching engineering

materials with AM materials, 2- costing and 3- mechanical properties
variation versus build orientation.
1- For matching: selecting material by properties let users enter values

they need. If one user needs a specific engineering material he can feed
the system with the properties and the system will show the AM
materials that can manufacture the part.
2- For costing: it is difficult to make a model that can be used because

there is a big difference between cost and price.
3- For adding mechanical variations versus build orientation: there is

no enough data available for the materials. Many experiments have to
be done for each material to build such database.
Some users liked the improvements, the simplicity, the surface finish
selection and the user assistance.
No user disliked anything about the AMDSS second version.
All the average and advanced users rated the process selection, material
selection, intermediate material selection, process update, material
update, finishing update, machine update, AMDSS clarity and AMDSS
length from good to excellent.
All users rated finishing selection and machine selection from good to
excellent except one user who rated these two elements from weak to
fair.
For the best future form of the AMDSS stand alone application was
chosen twice, online application was chosen four times and CAD plug-
in was chosen four times.
The only two advanced users who used the first version of the AMDSS
had seen that the second version is better for process selection and
161
material selection. For the finishing selection and machine selection one
user thought that the second version was like the first version and the
second user thought that the second version is better.
Most of users of the second questionnaire thought that adding the option
of selecting process first or material first, selecting material by class and
by properties, determining minimum and maximum values of the
material properties, and helping on the finishing strategies was helpful
and considered it a good improvement.
One user suggested for improvement: adding material class such as

polymers, metals and ceramics in the material list that appears to the
user.
6.3 Second Version User Testing and Evaluation
In section 5.2 four users have tested the developed first version of the AMDSS
with four parts that their selections was already made. The data of the four parts
were fed to the system and a comparison was made to the AMDSS results with the
real results. The selections of the first version were compatible with those of the
real ones.
Volume Calculation
Figure 6.28: Process Selection Screen Showing Volume Calculation (Version 2).
162
The second version of the AMDSS was tested using the same inputs of the four
parts. This testing was done by the researcher and not by the users of these case
studies. The Selections of the second version for the four parts were the same
selections of the first version but there are some improvements that need to be
highlighted which are the adding of volume calculations, finishing paths and
minimum and maximum property ranges.
Volume Calculations
While the first version of the AMDSS did not use the volume or showed to users,
the second version showed the volume calculation of the part in the process
selection screen when the user enters the values of the length, width and height as
shown in Figure 6.28. The volume calculations for the four parts were as follows:
Cover part = 2155860 mm3.
Drain fitting part = 279923 mm3.

Filter cover part = 657120 mm3.
Dental part assembly =16920 mm3.
Finishing Paths
While the first version of the AMDSS showed only finishing processes with very
limited information, the second version showed finishing paths instead of just one
finishing method.
Figure 6.29 shows the finishing paths for the cover part. The finishing path
selected was sanding + sandblasting instead of sandblasting in the first version
and the system showed the expected Ra level of the surface finish which was
average level (Ra = 0.32 m - 12.5m). Furthermore, the finishing advice list was
also an addition that gave advices to the users when making finishing selections.
163
Figure 6.29: Finishing Options Screen for the Cover Part (Version 2).
The finishing options that included sandblasting method (selected in the first
version) for the drain fitting part were six paths as shown in Figure 6.30:
1- Sandblasting (surface finish = Rough. Ra >12.5m).
2- Sandblasting + Polysurfacer Paint (surface finish = Average. Ra = 0.32-

12.5m).
3- Sandblasting + Polysurfacer Paint + Sanding + Primer Paint (surface finish =

Average. Ra = 0.32-12.5m).
4- Sandblasting + Polysurfacer Paint+ Sanding + Mat Paint (surface finish =

Average. Ra = 0.32-12.5m).
5- Sandblasting + Polysurfacer Paint + Sanding + Satin Paint (surface finish =

Average. Ra = 0.32-12.5m).
6- Sandblasting + Polysurfacer Paint + Sanding + High Gloss Paint (surface

finish = Smooth Ra = 0-0.32m).
The second version gave the user six finishing paths so he could select between
them and to know what would be the expected surface finish that he could obtain
from each surface finish path.
164
Figure 6.30: Finishing Options Screen for the Drain Fitting Part (Version 2).
The finishing options that included sanding method (selected in the first version)
for the filter cover were four paths as shown in Figure 6.31:
1. Sanding + Vibro-finishing (surface finish = Average. Ra = 0.32-12.5m).
2. Sanding (fine sand paper) (surface finish = Average. Ra = 0.32-12.5m).
3. Sanding (Average sand paper) (surface finish = Average. Ra = 0.32-

12.5m).
4. Sanding (coarse sand paper) (surface finish = Rough. Ra >12.5m).
Figure 6.31: Finishing Options Screen for the Filter Cover Part (Version 2).
165
The finishing option for the dental part assembly as shown in Figure 6.32 had
changed from polishing in the first version to a finishing path sanding + polishing.
In addition, the surface finish expected was shown which was smooth (Ra = 0 -
0.32m). The surface finish advice list shows an advice that says that polishing
gives an excellent surface finish.
Figure 6.32: Finishing Options Screen for the Dental Implant Part (Version 2).
Figure 6.33: Material Properties Screen for the Filter Cover Part (Version 2).
166
Minimum and Maximum Material Property Ranges
While the Material and properties screen of the first version included only the
minimum value of the different properties, the second version showed the
minimum and maximum ranges of different properties.
In the first version, in the material selection for the filter cover part the user
selected the minimum hardness to be 90 Shore D. There was a conflict between
the black colour and the required hardness and because the part could be
pigmented to black colour the user relaxed the colour constraint. This user in the
first version asked for having the possibility of having ranges so the trade-off
between the variables and compromising can be made. The user could accept a
hardness range between a minimum of 75 Shore D and a maximum of 90 Shore D.
In the second version, the user had the possibility to enter the two ranges as shown
in Figure 6.33.
Comments:
There are some remarks that need to be underlined from testing the second version
of the AMDSS using the four parts provided by users:
The volume was shown in the process selection when users entered the
part dimensions. The AMDSS checks this volume against the different
process maximum volumes. Some users asked for the volume to know the
amount of material needed and to use it for costing and to check how many
parts could a machine do in one build.
Users had the option to select between selecting materials by properties

and selecting materials by classes. Users of the four parts have already
some specified value of one or more property so selecting material by
classes was not used.
Users had the option to select process first or material first. This study
tried both routes and the results were the same.
The filter cover part showed how users can benefit from the minimum and
maximum ranges.
167
The finishing options in the system was improved by giving the possibility
to have finishing paths instead of a single finishing method, providing Ra
level and providing finishing advice that could help in the selection.
168
Chapter 7. Discussion
7.1 Summary of the Work
The need for an automated procedure for the selection of AM processes and
materials to meet the requirement of a desired component was the main motive of
this research. A system called AMDSS (additive manufacturing decision support
system) was developed. Nine main steps were used to achieve this goal: 1-
analyzing the selector systems, 2-identifying their shortcomings, 3- indentifying
target specifications, 4- developing the system, 5- verifying and validating the
system, 6- obtaining feedback from users, 7- improving the system based on
feedback, 8- obtaining a second feedback for the improved system from users and
finally 9- launching the system.
7.1.1 Selector System Shortcomings
The selectors were classified into three types which are: 1- rapid prototyping
selectors, 2- rapid manufacturing selectors and 3- rapid tooling selectors. The
number of discussed systems and methodologies developed were eleven studies
for rapid prototyping, ten studies for rapid manufacturing and eleven studies for
rapid tooling. Out of all the above studies only eighteen produced developed
systems, the rest were proposed methodologies. A comparison was made between
the eighteen developed systems. Among these eighteenth systems, six are for RP
selection only, four for RT selection only, five for both RP and RM selections,
one for both RP and RT selections and two systems for RP, RT and RM
selections.
Through the investigation of the developed systems the following shortcomings

were identified:
1. While only two developed systems were able to perform general selections
(RP, RT and RM), one of them lacked material selection and could not be
updated by users and the second system was a general guide rather than a
selector.
169
2. In most of the developed systems users cannot update the system database
and if they can do so there is a need to change the programming logic
which makes the systems obsolete after a while.
3. Many systems lacked material selection.
4. Most systems focused on processes only or machines only but there was
no integration between process chains, materials, finishing methods and
machines for a specific product.
5. Systems did not give reasons if a solution was not found.
6. Involvement of the users in the development of the selectors was not

present.
7. Although selection of processes and materials affect each other, there was
no ranking for processes and material together.
7.1.2 Target Specifications
Using the shortcomings of the developed selectors seven target specifications

were set in chapter three:
1. Selecting the possible AM process chains, materials, finishing methods

and machines.
2. Giving reasons if the system could not give a solution or satisfy users
requirements.
3. Ranking processes and materials according to user criteria.
4. Ability of being updated.
5. Ability of being customized.
6. Being simple, usable and customer focused.
7. Being product focused.
170
7.1.3 AMDSS First Version Development
Different architectures were considered and finally knowledge based system

(KBS) and database (DB) architecture was selected because it could achieve the
system target specifications. KBS and DB together form a versatile tool that
contains flexible rules (If-Then) that uses variables and not constants. The KBS
retrieves the data from the database depending on the user entry. The flexibility of
the rules prevents the AMDSS system from becoming obsolete.
Three components form the structure of the AMDSS which are: 1- graphical user
interface (GUI), 2- knowledge based system (KBS) and 3-Database (DB). Users
use the GUI to input data and get the required information or advice. The KBS
contains the programming logic and the database contains information about AM
processes, materials, intermediate materials, finishing options and machines.
IDEF0 and flowcharts tools were used to structure the system.
The KBS within the AMDSS consists of three main parts: 1- selection part, 2-
browse part and 3- update part. The selection part helps users to select processes,
materials, intermediate materials, finishing options and machine. The browse part
helps users to explore the different processes and materials available in the
database. The update part helps users to update the database.
7.1.4 System Verification and Validation
The AMDSS was verified to be sure that it was working properly and that the
programming logic was correct. Different scenarios were proposed, the input-
output relationships were tested and the result was that the system worked
properly and logically.
Three industrial case studies were used for validation: 1- razor blade case study, 2-
automative case study (1) and automotive case study (2). The input parameters
were fed to the AMDSS and the results were compared to the selections made in
consultation with experts and both selections were compatible with each other.
171
7.1.5 First Feedback
A questionnaire was designed and the first version of the AMDSS was tested with
thirteen users (nine of them average users and four advanced users). In addition,
four users tested the AMDSS with four products already manufactured. The
outputs of the AMDSS confirmed to the selections of the users
The responses of the users to the questionnaires were analyzed. They have
outlined some weak points regarding the AMDSS system. The key points from the
first feedback of users were:
1. AM selectors are not popular among avergae users and experts

because they are obsolete. Users select AM processes in an ad-hoc
way, or using websites or by asking service bureau which give
them what they want to sell and what they know and not the
optimum solution to the problem.
2. Some users preferred to select material before selecting process

and this was not possible in the first version.
3. Some users liked to select materials by class and not by property

screening.
4. Most of the users found the first version of the AMDSS was
helpful in the selection of the process chains, materials and
machines but not the finishing methods selection.
5. Most of the users rated the updating possibility in the system from
important to extremely important.
6. Most of the users liked the simplicity and the small number of steps
of the developed AMDSS.
7. Many users disliked that there was no ranges for material

properties (minimum and maximum values).
8. Users prefer to have an AM system in the following forms in the

following order: 1- online, 2- CAD plug-in, 3- standalone.
172
7.1.6 AMDSS Second Version Development
Thirteen modifications were made to the first version: 1- possibility of selecting

process first or material first, 2-possibility of selecting material by class, 3-volume
calculation of the part, 4- re-arranging the order of some inputs in the process
screen, 5- renaming surface finish ranges, 6- adding the maximum values for
some inputs of material properties, 7- adding hardness Vickers and melting point
to the material properties screen, 8- replacing sanitary condition with bio-
compatible, 9- changing sequence of some windows, 10- improving the finishing
selection by adding finishing path, finishing level (Ra), and finishing advice that
could be updated, 11- changing the updating panel to match the modifications and
12- adding help screens and 13- possibility of reverting the direction of sorting for
some criteria such as hardness, density and dielectric strength.
The effect of these changes on the programming logic was expressed using IDEF0
and flowcharts.
7.1.7 Second Feedback
The purpose of this step was mainly assessment more than a feedback. A second
questionnaire was designed which was a modified version of the first
questionnaire. Six users responded to the second questionnaire, three of them were
average users and three of them were advanced users. Two out of the six users
have responded to the first version of the questionnaire and both were considered
advanced users.
The responses of the users were analyzed. In general, the analysis of the responses
of the users showed that the second version is an improved version by covering
the weak points in the first version. The key points of the second feedback of users
were:
1. Most of the users found that the second version of the AMDSS was
helpful in the selection of the process chains, materials, machines
and the finishing methods.
2. All users rated the updating possibility in the system from

important to extremely important.
173
3. Some users liked the AMDSS improvements, others liked the

simplicity and others liked the surface finish selection option.
4. Users preferred to have an AM system in the following forms in

the following order: 1- online, 2- CAD plug-in, 2- standalone. The
online and the CAD plug-in were chosen the same number of
times.
5. Most users found that adding the option of selecting material first
or process first is a good improvement.
6. Most users found that adding the option of selecting material by

class or by properties is a good improvement.
7. Most users found that adding material ranges (minimum and

maximum values) is a good improvement.
8. Most users found assistance with the finishing options is very

useful.
7.1.8 AMDSS Limitations
Despite the satisfaction of some users who tested the system regarding the
developed AMDSS, it does not come without its limitations. These limitations are:
The system did not include any costing because there is a big difference
between cost and price. If the system gives an advice based on cost and
then the user went to a service bureau and was quoted a figure based on
price so this would be a misleading piece of information.
The system did not include geometry consideration, believing that AM

processes can produce complex geometries.
The information presented in the database was based on information from

vendors websites, papers and service bureaus. The accuracy of the data
may not be as accurate as it should be but getting independent materials
data for every material can only be done by experiment which would be
very costly and time consuming.
174
Material isotropy was not considered because of material data

unavailability.
Some material properties were not considered because of data

unavailability (such as compressive strength, flexural modulus,
environmental impact, hygroscopic property and porosity).
The process selection calculates the volume of the required part and the
build volume of the process and compares them. If the build volume is
greater than the required part than the process is considered as one of the
possible alternatives. Optimum build orientation was not considered.
Building time was not considered because it differs from a part to another
and parametric calculations give inaccurate figures.
7.2 Research Contributions
The main research contributions are:
1- Developing and evaluating an AM selector system using knowledge based

system (KBS) and Database (DB) architecture with the following
characteristics:
i. Integrated system that could select process chains, materials, finishing

methods and machines covering RP, RT and RM areas.
ii. Updatable without the need to change programming logic to avoid

embedded obsolescence. Users can update the information in the database
such as processes, materials, intermediate materials, machines and
finishing options. The system logic cannot be update, for example users
cannot define a new wall thickness level or add a new material property or
add their own criteria.
iii. Explaining the reasons for not being able to make a selection from the
database if a selection could not be made.
iv. Ranking processes and materials together because they influence each
other.
175
2- The involvement of users in the development stage of the AMDSS which is an

important issue so that the developed selector represents their needs and not
only the point of view of the developer.
7.3 Possible Future Forms of AMDSS
The two versions of the questionnaire asked users to select between four possible
future forms of the developed AMDSS: 1- a standalone application, 2- CAD plug-
in, 3- Online application and 4- others (to be specified by the user). Users can
select more than one answer.
In the first version, standalone was chosen four times, online was chosen eight
times, CAD plug-in was chosen six times. In the second version, stand-alone was
chosen two times, online was chosen four times and CAD plug-in was chosen four
times. Combining the two results give the selection of stand-alone six times, the
selection of online twelve times and the selection of CAD plug-in ten times. The
preferences in order are: 1- online, 2- CAD plug-in and 3- stand alone.
7.3.1 Standalone Application
The AMDSS has been developed in the Matlab environment. Matlab has a
deployment tool that compiles the developed program so that it can work as a
stand-alone application outside Matlab. The deployment tool starts a new empty
project file that the developer adds to it all the .m files (Matlab coding files) and
the .fig files (Matlab figure files) that form the complete program. The final result
is an executable file that can be opened on any computer. Within the developed
project Matlab added a compiler installer file called MCRInstaller.exe. Users
have to run the installer at any computer only once before running the executable
file. Another issue is that the developed AMDSS uses the visual query builder
toolbox embedded in Matlab. The toolbox connects to the database through
ODBC data source administrator. Users have to define the database used in the
program in windows before running the executable file as shown in Figure 7.1 .
Certainly, the used method to transform the Matlab program to a stand-alone

application is not the best option because when a user needs to run the developed
program he has first to download the Matlab compiler on his machine and also
add the database file using ODBC source administrator. Software companies can
take the framework and the programming logic using the provided IDEF0 and
176
flowcharts and develop an AM selector using any advanced versatile

programming language.
Figure 7.1: Adding ODBC Data Source in Windows
7.3.2 CAD Plug-in
Many users asked to have a CAD plug-in that could take the information required
from the drawing file and analyze it. The program can work in the CAD
environment as a start point. For example, Smith and Rennie (2010) have
developed a knowledge system contained within a CAD environment
(Solidworks) to select AM materials and processes. The screening is based on
geometrical part features. The system screens the feature tree of a part, find
features and compare each feature values against the saved values of the feature
that an AM material and process is suitable for. It is an iterative process that is
repeated until all the features are checked and at the end users have a list of
materials that could manufacture the part. Users can render the part so they can
view an estimation of the part.
In the case of the developed AMDSS, the information that could be retrieved is
length, width, height and wall thickness of a given part. A menu could be added in
the CAD software that could capture all the information needed. Next, the
AMDSS starts within the CAD and users make their selections. After selecting the
177
process and materials, the part can be rendered within the CAD again so users can
imagine how the part will look like.
7.3.3 Online Application
It is very obvious that the future trend is moving more and more towards an online
world not only in AM process selection but in every field and domain. The idea of
accessing the information from any place and from any computer or mobile phone
is interesting for users.
In the literature there are some researchers that have used the idea of internet
based rapid prototyping but this was for providing users with direct remote access
to RP facility over the internet (sometimes called RP Tele manufacturing). The
activities of these web-based RP-RM platforms include: process selection, price
quotation, .STL viewer, RP data re-processing, job scheduling, remote control of
AM machines, security management and applying new technologies. A good
review of this direction of research could be found in Lan (2009). There is no
focus on process selection in these types of studies; their focus is mainly on the
system architectures.
Few researchers addressed the idea of the online selection of additive

manufacturing. The available online application for selection are: 1- The RP
selector developed by the IVF Sweden 2005 (IVF,2005), 2- The IRPDMS
developed by (Mahesh et al.,2005), although the webpage is not available, 3- The
decision support system for RP selection developed by (Lan et al.,2005) and 4-
The KARMA project (ongoing FP7 project) (KARMA,2010).
178
Chapter 8. Conclusions and Future Work
The aim of this research was to develop an additive manufacturing decision

support system that is capable of evaluating and selecting the most appropriate
AM processes to meet the requirements of the desired component. This aim has
been achieved by developing the AMDSS, which is a general AM selector system
that helps users to select AM process chains, materials, intermediate material,
finishing methods and machines. This chapter discusses the conclusions that can
be drawn out of this study and suggests some possible future work.
8.1 Review of Objectives
This section describes the achievement of the five main objectives of the research.
1. Exploring different additive manufacturing processes and examining the

most up to date information about new technologies in this field, and the
most widely used by industry.
To develop the system, different AM processes and process chains were

examined. The database developed (described in chapter three) included 14 of the
most popular processes and process chains in the industry.
2. Exploring different additive manufacturing materials, machines and

finishing options.
To develop the system, different AM materials, machines and finishing options

were examined. The database developed (described in chapter three) included 133
materials, 10 intermediate materials, 52 machines and 44 finishing paths.
3. Using decision analysis tools to make selection among different layer

manufacturing technologies and processes.
Chapter three discusses the use of IDEF0 and flowcharts to develop and describe
the first version of the developed selector. In addition, chapter six discusses the
also the use of the IDEF0 and flowcharts to develop the second version of the
developed selector.
179
4. Developing a system that is capable of assisting decision makers in:
I. Determining the feasibility of additive manufacturing processes for

a particular application.
II. Selecting the most suitable process.
The AMDSS was developed using KBS + Database architecture and was capable
of determining if the processes were able to manufacture certain component or not
and if not the AMDSS gives reasons for users why AM is not a suitable process.
In addition, the processes and materials was ranked using SMART method
according to the criteria of users so users can select from the final list the process
and material that are suitable for their components . In addition, users can select
intermediate materials, finishing options and machines. Chapter three and chapter
six discuss the selection of the first and the second version of the AMDSS
respectively.
5. Testing the developed system by applying verification and validation

methods.
Testing the system was done through different methods: 1- verifying of the system
to be sure that it is working correctly by checking system logic and comparing
calculations done by the system with calculations done in spreadsheets and this
was explained in chapter four, 2- validating of the system with three case studies
in chapter four, 3- testing the system with users through interviewing users and
making two versions of questionnaires described in chapter five (first version)and
six (second version) and 4- making four users to validate the system with their
components described in chapter five.
8.2 Conclusions
The following conclusions can be drawn out of this research:
1. AM selectors could be very useful for inexperienced and average users in

order to help them to select processes, materials, machines and finishing
paths if they are developed in a manner that represent the users needs.
180
2. The updatability issue in any AM selector is very important and prevents

the system from becoming obsolete. It has been achieved by using the
flexible if-then rules.
3. Knowledge Based System (KBS) and Database (DB) architecture is

considered a versatile architecture for the development of an updatable
additive manufacturing decision support. This architecture could also be
used for any decision support system and not only for AM selections.
4. The involvement of users in the development stage of AM selector is

essential so the developed system represents the point view of users and
not only developers.
5. Users prefer to have an online AM selector and also a CAD plug-in rather
than a standalone application.
8.3 Future Work
The possible future work includes:
1. Developing a standalone system professionally to be a simpler, more user-

friendly and more attractive selector system.
2. Integrating the developed system into different CAD environments so

designers are comfort with the use of such a system. For example, adding
the possibility to read .stl file instead of entering the data manually so the
system could define the part width, length, height and minimum wall
thickness.
3. Developing an online system based on the developed AMDSS using the

three-tier architecture. The online system should have an online database
that could be updated by experts so the data in the database become
accurate and users can rely on.
4. Linking other modules to the system such as developing optimum building

scenarios for different processes using different materials. A general AM
process planning is required.
5. Adding a library that contains known solutions to known problems and use
case based reasoning approach for the new problem to find a similar
181
problem and in result a similar solution. If this work could be done online,
there could an online library that contains solution for different problems.
This could be a very good pool for new users.
6. Developing a system that includes AM processes and conventional

processes.
7. This work has addresses only the problem of the AM selection problem
but there are other two types of problems that still need more work: 1-
capital investment support and 2- process planning support. A system that
could incorporate all the three types of problems will be a good step
through a complete AM solution.
182
Appendix A. First Questionnaire Responses
1- User profile
This part contained three questions.
Question 1:
How often do you need to select additive manufacturing process or material?
a. Weekly
b. Monthly
c. Sporadically
d. Other:_______________________
Answers:
The answers to this question are summarized in Table A-1.
Advanced Users Average Users All Respondents

Weekly 4/4 Weekly 1/9 Weekly 5/13
Monthly 0/4 Monthly 3/9 Monthly 3/13
Sporadically 0/4 Sporadically 5/9 Sporadically 5/13
Table A-1: Responses to User Profile Part Q1.
Question 2:
Have you used any knowledge based system for selecting additive process or
material before? If yes please name it.
Answers:
Two advanced users answered yes and the other two answered no. One of the
users who answered yes mentioned that he has used Matweb and AM materials
selector. These two selectors are general selectors and not specialized in AM. The
second user stated that he has used two AM selectors: the first one he did not
remember the name, the second one is Proto Selection. All the nine average
users answered no. The answers to this question are summarized in Table A-2.
183
Yes 2/4 Yes 0/9 Yes 2/13
No 2/4 No 9/9 No 11/13
No Answer 0/4 No Answer 0/9 No Answer 0/13
Table A-2: Responses to User Profile Part Q2.
Question 3:

b. Material
d. Machines
Answers:

Own Experience 2/4 Own Experience 2/9 Own Experience 4/13
Product Needs 1/4 Product Needs 5/9 Product Needs 6/13
Cost/Lead Time 1/4 Cost/Lead Time 1/9 Cost/Lead Time 2/13
Service Bureaus 0/4 Service Bureaus 1/9 Service Bureaus 1/13
Recommendations Recommendations Recommendations
Table A-3: Responses to User Profile Part Q3a.
b. Material

Cost 0/4 Cost 1/9 Cost 1/13
Table A-4: Responses to User Profile Part Q3b.
184

In-house capability 1/4 In-house capability 0/9 In-house capability 1/13
Table A-5: Responses to User Profile Part Q3c.
d. Machines

Build Size 2/4 Build Size 0/9 Build Size 2/13
Build Time 0/4 Build Time 1/9 Build Time 1/13
Availability 1/4 Availability 2/9 Availability 3/13
Knowledge 0/4 Knowledge 2/9 Knowledge 2/13
Service Bureau 0/4 Service Bureau 1/9 Service Bureau 1/13
No Difference 1/4 No Difference 0/9 No Difference 1/13
Table A-6: Responses to User Profile Part Q3d.
2- AMDSS Screens
This part of the questionnaire asked about the main AMDSS screens. It contained
eight questions that addressed eight main screens: 1-process screen, 2-material
properties screen, 3- criteria weighting screen, 4- final ranking screen, 5- process
update screen, 6- material update screen, 7-machine update screen, and 8-
finishing options update screen. Each question contained one or more sections
like: information section, terminology section, screening methods section, and
scoring results section. Each section contained one or more sub-questions.
185
Question 1: Process screen
The first question asked about the process screen shown in Figure A-1. It
contained two sections which were: information section and terminology section.

b. For your application, do you have data for all the presented fields?
Figure A-1: Process Selection Screen.
Answers:

Yes 4/4 Yes 8/9 Yes 12/13
No 0/4 No 1/9 No 1/13
Table A-7: Responses to AMDSS Screens Part Q1a (Information Section).
186

Yes 1/4 Yes 7/9 Yes 8/13
No 3/4 No 2/9 No 5/13
Table A-8: Responses to AMDSS Screens Part Q1b (Information Section).

Yes 0/4 Yes 0/9 Yes 0/13
No 4/4 No 7/9 No 11/13
Table A-9: Responses to AMDSS Screens Part Q1c (Information Section).
Two advanced users said yes without specifying specific field, the other two said
yes with the first one would like to add part volume and the second one would like
to add delivery time. One average user did not answer, six did not want to add
anything, one wanted to define the build platform shape, and one wanted to add
overhung feature and make a decision on the support method for every thin
feature. The answers to this question are summarized in Table A-10.

Yes 4/4 Yes 2/9 Yes 6/13
No 0/4 No 6/9 No 6/13
Table A-10: Responses to AMDSS Screens Part Q1d (Information Section).
ii. Terminology Section
1- Do you think that the terminology used in the following drop down menu is
easy to understand? Use the scale give below each term.
[1= difficult to understand and 5 =easy to understand.]

187
Min wall thickness
(Very-Thin 0-0.5mm) / (Thin-Average 0.5-2mm) / (Average-Wide (0.2mm)
1 2 3 4 5
Accuracy
(Loose <0.1mm) / (Average 0.1-0.25mm) / (Tight>0.25mm)
1 2 3 4 5
Surface Finish
(Average-Rough>12.5 um) /(Good-Average 3.2-12.5 um) / (Excellent 0-0.32

um)
1 2 3 4 5
Quantity
(One) / (Low2-20) / (Medium 20-25) / (High>50)
1 2 3 4 5
If you think the terminology is difficult to understand please suggest alternative

phrasing.
Answers:
188
Rating
Users 1 2 3 4 5
Min wall Average 1 1 1 6

thickness Advanced 1 3
All 1 2 1 9
Average 1 1 6
Accuracy Advanced 1 3
All 2 1 9
Average 1 1 5
Surface finish Advanced 3 1
All 4 1 6
Average 1 2 5
Quantity Advanced 1 3
All 1 3 9
Table A-11: Responses to AMDSS Screens Part Q1 (Terminology Section).
Question 2: Material properties screen
The second question asked about the material properties screen shown in Figure
A-2. It contained two sections which were: information section and screening
method section. .

189
Figure A-2: Material Properties Screen.
Answers:

Yes 3/4 Yes 9/9 Yes 12/13
No 0/4 No 0/9 No 0/13

Yes 0/4 Yes 4/9 Yes 4/13
No 3/4 No 5/9 No 8/13
190

Yes 0/4 Yes 0/9 Yes 0/13
No 3/4 No 9/9 No 12/13
Three advanced users said no and the fourth left this question blank. One of them
commented to rename sanitary conditions to medical consideration. Another One
suggested having alternative units for selections. Three average users said yes and
six Average users answered no. The three who answered yes have different
comments: one of them suggested adding flexural modulus, another one suggested
adding biocompatible, environmental impact, and flexural modulus, the third one
suggested adding porous, and hygroscopic (ability of the material to hold water).

Yes 0/4 Yes 3/9 Yes 3/13
No 3/4 No 6/9 No 9/13
Table A-15: Responses to AMDSS Part Screens Q2d (Information Section).
ii. Screening Method Section
1- Do you prefer the material screening method to be :
a. A generalized screening method like by selecting material class

(Metals, Plastics, Ceramics, Composites, Wood. .etc)
b. A properties screening method by giving values to one or more
material properties.
c. I would like to choose every time I make a selection between the two
screening methods.
191
Answers:

Generalized 2/4 Generalized 3/9 Generalized 5/13
Properties 1/4 Properties 0/9 Properties 1/13
Choose Choose every Choose every
0/4 6/9 6/13
every time time time
Table A-16: Responses to AMDSS Screens Part Q1 (Screening Method Section).
2- According to the material properties screening method do you think that it

should be based on:
a. Minimum values.
b. Maximum values
c. Both

Min 2/4 Min 2/9 Min 4/13
Max 0/4 Max 0/9 Max 0/13
Both 1/4 Both 5/9 Both 6/13
Table A-17: Responses AMDSS Screens Part Q2 (Screening Method Section).
Question 3: Criteria weighting screen
The third question asked about the criteria weighting screen shown in Figure A-3.
It contained one section which is the information section.
Information Section

192
Figure A-3: Criteria Weighting Screen.
Answers:

Yes 4/4 Yes 7/9 Yes 11/13
No 0/4 No 0/9 No 0/13

Yes 0/4 Yes 5/9 Yes 5/13
No 3/4 No 2/9 No 5/13
193
Yes 0/4 Yes 0/9 Yes 0/13
No 4/4 No 6/9 No 10/13
Two advanced users said yes, and two said no. The first user with positive answer
commented to add building time and the second one commented to add elongation
and impact strength. Four average users said yes, three said no and two did not
answer. The first average user who said yes wanted to add cost and modulus. The
second user wanted to add cost of materials. The third one suggested adding
relative ranking of indicative cost (high, medium, low) of process material and
also to add lead time. Finally the fourth user also suggested adding cost. The
answers to this question are summarized in Table A-21.

Yes 2/4 Yes 4/9 Yes 6/13
No 2/4 No 3/9 No 5/13
Question 4: Final ranking screen
The fourth question asked about the final ranking screen shown in Figure A-4. It
consisted of one section which is the scoring results section.
Figure A-4: Final Ranking Screen.

194
Scoring Results Section
a. Are the scoring results clear and understandable? If no please comment.
Answers:
a. Are the scoring results clear and understandable? If no please comment.

Yes 3/4 Yes 6/9 Yes 9/13
No 1/4 No 2/9 No 3/13
Table A-22: Responses to AMDSS Screens Part Q4a (Scoring Results Section).
Two advanced users did not want to add or remove anything. The other two, one
suggested to add indicating costing and the other one suggested to add info on
material such as resin, nylon..etc. Six average users answered yes, one answered
no and two left this question blank. The first average user who said yes suggested
removing the score. The second one suggested adding a summary of each final
entry. The third and fourth user suggested adding cost. The fifth suggested adding
colour to this screen. Finally the sixth user suggested showing the score
calculation. The answers to this question are summarized in Table A-23.

Yes 2/4 Yes 6/9 Yes 8/13
No 2/4 No 1/9 No 3/13
Table A-23: Responses to AMDSS Screens Part Q4b (Scoring Result Section).
Question 5: Process update screen
The fifth question asked about the process update screen shown in Figure A-5. It
contained one section which is the information section.
195
Information Section

Figure A-5: Process Update Screen.
Answers

Yes 3/4 Yes 6/9 Yes 9/13
No 0/4 No 2/9 No 2/13
196

Yes 1/4 Yes 7/9 Yes 8/13
No 2/4 No 1/9 No 3/13

Yes 0/4 Yes 0/9 Yes 0/13
No 3/4 No 8/9 No 11/13
Two advanced users said yes, one said no and one did not answer this question.
The first user who said yes suggested adding process time and the second user
suggested adding part volume. Seven average users answered no and two users
did not answer this question. The answers to this question are summarized in
Table A-27.

Yes 2/4 Yes 0/9 Yes 2/13
No 1/4 No 7/9 No 8/13
Question 6: Material update screen
The sixth question asked about the material update screen shown in Figure A-6. It
contained one section which is the information section.
197
Figure A-6: Material Update Screen
Information Section

Answers:

Yes 3/4 Yes 6/9 Yes 9/13
No 1/4 No 1/9 No 2/13
198

Yes 0/4 Yes 5/9 Yes 5/13
No 4/4 No 3/9 No 7/13

Yes 0/4 Yes 0/9 Yes 0/13
No 4/4 No 8/9 No 12/13
Three advanced user said yes and one said no. The first user suggested adding
material cost. The second user suggested adding more properties like for example
chemical properties. The third user suggested adding speed because different
materials can take longer time on the same machine. Three average users said yes,
five answered no and one did not answer this question. The first user who said yes
suggested adding flexural modulus, the second user suggested adding cost and the
third user suggested adding electrical resistance and porosity. The answers to this
question are summarized in Table A-31.

Yes 0/4 Yes 3/9 Yes 3/13
No 3/4 No 5/9 No 8/13
199
Question 7: Machine update screen
The seventh question asked about the Machine update screen shown in Figure A-
7. It contained one section which is the information section.
Figure A-7: Machine Update Screen.
Information Section

Answers:

Yes 4/4 Yes 6/9 Yes 10/13
No 0/4 No 2/9 No 2/13
200

Yes 2/4 Yes 3/9 Yes 5/13
No 1/4 No 1/9 No 2/13

Yes 0/4 Yes 0/9 Yes 0/13
No 3/4 No 9/9 No 12/13
Three advanced users said no, and the fourth would like to add beam diameter
because this is related to accuracy. One average user answered yes and he would
like to add machine availability. Five users answered no and three did not answer.

Yes 1/4 Yes 1/9 Yes 2/13
No 3/4 No 5/9 No 8/13
Question 8: Finishing options update screen
The eighth question asked about the finishing options update screen shown in
Figure A-8. It contained one section which is the information section.
201
Figure A-8: Finishing Options Update.
Information Section

Answers:

Yes 4/4 Yes 5/9 Yes 9/13
No 0/4 No 0/9 No 0/13

Yes 4/4 Yes 4/9 Yes 8/13
No 0/4 No 1/9 No 1/13
202

Yes 0/4 Yes 0/9 Yes 0/13
No 3/4 No 5/9 No 8/13
Two advanced users said yes, and the other two said no. the first user who said yes
suggested characterizing processes and quantifying levels of finishing. In addition,
adding the time the finishing method could take so this could help in costing. The
Second one suggested adding information about the surface the finishing is
applied to like: raw metal or paint or lacquer. Furthermore, specifying surface
reflectivity (matt, satin, gloss).
Two average users said yes, two said and five users left this question blank. The
first user who answered yes commented to add finishing cost, and the second one
suggested adding the surface finish level that you get. The answers to this question
are summarized in Table A-39.

Yes 2/4 Yes 2/9 Yes 4/13
No 2/4 No 2/9 No 4/13
This part of the questionnaire consisted of twelve questions. The main purpose of
this part was to analyze the different aspects of the developed AMDSS and to
understand the strength and weakness of it.
203
Question 1:
The Decision Support System (DSS) has been designed to choose process first
than material. Do you think that this is appropriate? Please comment on your
answer.
Answers

Select process first 1/4 Select process first 1/9 Select process first 2/13
Select material first 2/4 Select material first 4/9 Select material first 6/13
Both paths 1/4 Both paths 4/9 Both paths 5/13
Table A-40: Responses to General Questions Part Q.
Question 2:
Did you find the choice of the finishing methods that you want useful? Please
comment.
Answers
The first advanced user thought the choice was useful. The second one suggested
adding more machinery processes for metal parts. The third one thought choice
was useful but quite general and maybe useful for average users but not enough
information for advanced users. The fourth one thought that quantification and
characterization against conventional surface finishes would be a good option like
for example saying that a specific surface finish is like injection plastic surface
finish.
Three average users thought that the choice was useful and the other six have
different opinions: the first user thought that adding costing and ranking of the
finishing options would be a good aspect, the second one thought that more
information should be provided like the quality of the surface finish, time and
cost, the third one suggested to specify a link surface finishes to applications, the
fourth one proposed to add the surface finish level that the user can get from each
surface finish option, the fifth one suggested to have more details about the
process performance, the sixth one agreed with the idea of the fifth about having
204
more details and specifications. The answers to this question are summarized in
Table A-41.

Yes 2/4 Yes 3/9 Yes 5/13
No 2/4 No 6/9 No 8/13
Table A-41: Responses to General Questions Part Q2.
Question 3:
Did you find the DSS helpful when you have applied it to your case study for
selecting :
ii. Materials
iii. Machines
Answers

Yes 2/4 Yes 6/9 Yes 8/13
No 2/4 No 2/9 No 4/13
Table A-42: Responses to General Questions Part Q3 (i).
ii. Materials
205
Yes 3/4 Yes 7/9 Yes 10/13
No 1/4 No 1/9 No 2/13
Table A-43: Responses to General Questions Part Q3 (ii).
iii. Machines

Yes 2/4 Yes 7/9 Yes 9/13
No 2/4 No 1/9 No 3/13
Table A-44: Responses to General Questions Part Q3 (iii).
The answers to this question are summarized in Table A-45

Yes 2/4 Yes 6/9 Yes 8/13
No 2/4 No 2/9 No 4/13
Table A-45: Responses to General Questions Part Q3 (iv).
Question 4:
manufacturing process, is the information provided clear?
Answers
206
Yes 4/4 Yes 6/9 Yes 10/13
No 0/4 No 1/9 No 1/13
Question 5:
manufacturing material, is the information provided clear?
Answers

Yes 4/4 Yes 6/9 Yes 10/13
No 0/4 No 1/9 No 1/13
Question 6:
Did you find that the DSS provides sufficient information on the ranking and the
choices?
Answers

Yes 3/4 Yes 7/9 Yes 10/13
No 1/4 No 1/9 No 2/13
Question 7:
How important is the ability of updating the process chain, material, finishing and
machines in an additive manufacturing DSS?
207
a. Not Important
b. Important
c. Very Important
Answers

Not Important 0/4 Not Important 0/9 Not Important 0/13
Important 1/4 Important 2/9 Important 3/13
Very Important 1/4 Very Important 0/9 Very Important 1/13
Ext. Important 2/4 Ext. Important 4/9 Ext. Important 6/13
Question 8:
What elements do you like to add in the DSS?
Answers
Advanced Average All

Building Time 1 2 3
Costing 2 4 6
Surface finish characterization 1 0 1
Help/Examples/ Welcome screen 1 3 4
More processes+ Materials 1 1 1
Question 9:
Answers
208
DSS length 1 0 1
Simplicity/easiness of use 2 6 8
Ranking system 1 2 3
Possibility of being updated 0 1 1
Good source of information 0 1 1
Question 10:
Answers

No ranges for material properties 1 0 1
Lack of help/guidance 1 7 8
Graphical interface 2 0 2
Lack of ranking score breakdown 0 1 1
Question 11
How do you rate the following elements in the DSS:
Elements Rating

Intermediate material selection Weak Fair Good Very Good Excellent
209
Answers
Users select from the rating options: weak, fair, good, very good and excellent.
Table A-53 shows both the advanced and average users rating together for the
different AMDSS elements. Not all the users have rated all the elements, so some
elements do not have thirteen answers.
Elements Rating
Users Weak Fair Good Very Good Excellent
Process Average 0 1 5 2 0
selection Advanced 0 0 2 1 1
All 0 1 7 3 1
Material Average 0 1 5 2 1
All 0 2 6 3 2
Intermediate Average 0 1 3 1 1
material Advanced 0 1 1 2 0
selection All 0 2 4 3 1
Finishing Average 0 2 4 2 0
All 1 3 5 3 0
Machines Average 0 1 3 2 0
All 0 1 4 5 0
Process Average 0 1 4 1 2
update Advanced 0 1 1 1 1
All 0 2 5 2 3
Material Average 0 1 3 2 1
All 0 3 5 2 1
Finishing Average 0 2 2 2 1
All 1 3 3 3 1
Machine Average 0 1 1 2 1
All 0 1 4 2 2
Average 0 3 4 1 0
AMDSS clarity Advanced 0 0 2 1 1
All 0 3 6 2 1
AMDSS Average 1 0 3 3 0
length Advanced 0 1 1 1 1
All 1 1 4 4 1
210
Table A-54 shows the percentage of the user selections dividing the five ratings to
two groups which are: 1- weak to fair and 2- good to excellent. The percentage
was calculated by dividing on thirteen which is the total number of users. Some
users did leave the answer blank.
Elements Rating
Users Weak to Fair Good to Excellent

Average 1 11.11% 7 77.78%
Process selection Advanced 0 0.00% 4 100.00%
All 1 7.69% 11 84.62%
Average 1 11.11% 7 77.78%
Material selection Advanced 1 25.00% 3 75.00%
All 2 15.38% 10 76.92%
Intermediate Average 1 11.11% 5 55.56%
material selection Advanced 1 25.00% 3 75.00%
All 2 15.38% 8 61.54%
Average 2 22.22% 5 55.56%
Finishing selection Advanced 1 25.00% 3 75.00%
All 3 23.08% 8 61.54%
Average 2 22.22% 5 55.56%
Machines selection Advanced 0 0.00% 4 100.00%
All 2 15.38% 9 69.23%
Average 1 11.11% 7 77.78%
Process update Advanced 1 25.00% 3 75.00%
All 2 15.38% 10 76.92%
Average 1 11.11% 6 66.67%
Material update Advanced 2 50.00% 2 50.00%
All 3 23.08% 8 61.54%
Average 2 22.22% 5 55.56%
Finishing update Advanced 2 50.00% 2 50.00%
All 4 30.77% 7 53.85%
Average 1 11.11% 4 44.44%
Machine update Advanced 0 0.00% 4 100.00%
All 1 7.69% 8 61.54%
Average 3 33.33% 5 55.56%
DSS clarity Advanced 0 0.00% 4 100.00%
All 3 23.08% 9 69.23%
Average 1 11.11% 6 66.67%
DSS length Advanced 1 25.00% 3 75.00%
All 2 15.38% 9 69.23%
Table A-54: User Ratings Percentage.
211
Question 12:
Looking long term, do you think that the best form for this DSS is:
c. A CAD Plug-in.
Answers

Stand alone 0 Stand alone 4 Stand alone 4
Online 3 Online 5 Online 8
CAD plug-in 2 CAD plug-in 4 CAD plug-in 6
Other 0 Other 0 Other 0
212
Appendix B. Second Questionnaire Responses
1- User Profile
Only the users who test the AMDSS for the first time answered the user profile
part. The users who have already tested the first version of the AMDSS did not
answer this part because they have already filled this information before. Two
users have tried the first version so they did not answer and both are classified as
advanced users. The remaining four users have answered this part. This part of the
questionnaire consists of three questions.
Question1:
How often do you need to select additive manufacturing process or material?
a. Weekly
b. Monthly
c. Sporadically
d. Other:_______________________
Answers:
The answers to this question are summarized in Table B-1.

Weekly 1/1 Weekly 1/3 Weekly 2/4
Monthly 0/1 Monthly 1/3 Monthly 1/4
Sporadically 0/1 Sporadically 1/3 Sporadically 1/4
Table B-1: Responses to User Profile Part Q1.
Question2:
Have you used any knowledge based system for selecting additive process or
material before? If yes please name it
Answers:
213
Yes 0/1 Yes 0/3 Yes 0/4
No 1/1 No 3/3 No 4/4
Question3:

b. Material
d. Machines
Answers:

Application/ Application/ Application/
1/1 0/3 1/4
Accuracy Accuracy Accuracy
Experience/ Experience/ Experience/
0/1 3/3 3/4
Asking Experts Asking Experts Asking Experts
This part of the questionnaire consists of twelve questions. The questions and the
answers of the users are described below.
Question1:
Would you like to add any information to any of the AMDSS screens?
Answers:
Two of the advanced users answered this question with yes while the third user
answered with no. The first user suggested adding more machines and adding the
ability of the selection by engineering materials. There are differences between the
214
AM materials and engineering materials. For example the ABS used in FDM
process is not the same ABS used in injection plastic. He suggested that the
system could be able to find the closest material to an engineering material. The
second user thought that some simplifications are required for the average users.
Two of the three average users answered this question with no. the third user
answered yes and suggested to have a home button in every screen that direct
users to the start screen with one click. The answers to this question are
summarized in Table B-4.

Yes 2/3 Yes 1/3 Yes 3/6
No 1/3 No 2/3 No 3/6
Table B-4: Responses to General Questions Part Q1.
Question 2:
Would you like to remove any information to any of the AMDSS screens?
Answers:
The answers to this question are summarized in Table B-5. The user, who
answered with yes, suggested removing from the database some SLA materials
because the list from his point of view is very exhaustive.

Yes 1/3 Yes 0/3 Yes 1/6
No 2/3 No 3/3 No 5/6
Question3:
Did you find the AMDSS helpful when you have applied it to your case study for
selecting:

II. Materials
III. Machines
215
Answers:


Yes 2/3 Yes 2/3 Yes 4/6
No 1/3 No 0/3 No 1/6
Table B-6: Responses to General Questions Part Q3 (I).
II. Materials

Yes 2/3 Yes 2/3 Yes 4/6
No 0/3 No 0/3 No 0/6
Table B-7: Responses to General Questions Part Q3 (II).
III. Machines

Yes 2/3 Yes 3/3 Yes 5/6
No 0/3 No 0/3 No 0/6
Table B-8: Responses to General Questions Part Q3 (III).


Yes 2/3 Yes 2/3 Yes 4/6
No 0/3 No 0/3 No 0/6
Table B-9: Responses to General Questions Part Q3 (IV).
216
Question 4:
How important is the ability of updating the process chain, material, finishing and
machines to you in an additive manufacturing DSS?
a. Not Important
b. Important
c. Very Important
Answers:

Not Important 0/3 Not Important 0/3 Not Important 0/6
Important 1/3 Important 0/3 Important 1/6
Very Important 1/3 Very Important 0/3 Very Important 1/6
Ext. Important 1/3 Ext. Important 3/3 Ext. Important 4/6
Question 5:
What elements do you like to add to the AMDSS?
Answers:

Matching Eng. Materials with
1/3 0/3 1/6
AM Materials
Costing 1/3 0/3 1/6
Mechanical Properties
1/3 0/3 1/6
Variations versus orientation
Nothing to Add 0/3 2/3 2/6
No Answers 0/3 1/3 1/6
217
Question 6:
Answers:

DSS Improvements 2/3 0/3 2/6
Simplicity 1/3 0/3 1/6
Surface Finish Selection 0/3 1/3 1/6
User Assistance 0/3 1/3 1/6
No Answer 0/3 1/3 1/6
Question 7:
Answers:

Nothing 3/3 2/3 5/6
No Answer 0/3 1/3 1/6
218
Question 8:
How do you rate the following elements in the AMDSS:
Elements Rating
Intermediate
Weak Fair Good Very Good Excellent
material selection
219
Answers:
Elements Rating
Users Weak Fair Good Very Good Excellent
Process Average 2
selection Advanced 2 1
All 2 3
Material Average 2
selection Advanced 2 1
All 2 3
Intermediate Average 1 1
material Advanced 3
selection All 4 1
Finishing Average 2
selection Advanced 1 1 1
All 1 1 1 2
Machines Average 1 1
selection Advanced 1 1 1
All 1 2 2
Process Average 1 1
update Advanced 2 1
All 2 2 1
Material Average 2
update Advanced 2 1
All 2 1 2
Finishing Average 1 1
update Advanced 3
All 3 1 1
Machine Average 2
update Advanced 2 1
All 2 1 2
Average 1 1
AMDSS clarity Advanced 2 1
All 3 2
AMDSS Average 2
length Advanced 1 2
All 1 2 2
Users select from the rating options: weak, fair, good, very good and excellent.
Table B-14 shows both the advanced and average users rating for the different
AMDSS elements. One average user has left this question blank.
220
Table B-15 shows the percentage of the user selections dividing the five ratings to
two groups which are: 1- weak to fair and 2- good to excellent. The percentage
was calculated by dividing on six which is the total number of users.
Elements Rating
Users Weak to Fair Good to Excellent

Average 2 66.67%
Process selection Advanced 3 100%
All 5 83.33%
Average 2 66.67%
Material selection Advanced 3 100%
All 5 83.33%
Intermediate Average 2 66.67%
material selection Advanced 3 100%
All 5 83.33%
Average 2 66.67%
Finishing selection Advanced 1 33.33% 2 66.67%
All 1 16.67% 4 66.67%
Average 2 66.67%
Machines selection Advanced 1 33.33% 2 66.67%
All 1 16.67% 4 66.67%
Average 2 66.67%
Process update Advanced 3 100%
All 5 83.33%
Average 2 66.67%
Material update Advanced 3 100%
All 5 83.33%
Average 2 66.67%
Finishing update Advanced 3 100%
All 5 83.33%
Average 2 66.67%
Machine update Advanced 3 100%
All 5 83.33%
Average 2 66.67%
DSS clarity Advanced 3 100%
All 5 83.33%
Average 2 66.67%
DSS length Advanced 3 100%
All 5 83.33%
Table B-15: User Ratings Percentage
221
Question 9:
Looking long term, do you think that the best form for a DSS of this type is:

c. A CAD Plug-in.
Answers:

Stand alone 2 Stand alone 0 Stand alone 2
Online 2 Online 2 Online 4
CAD plug-in 2 CAD plug-in 2 CAD plug-in 4
Other 0 Other 0 Other 0
Table B-16: Responses to General Questions Part Q9
Question 10:
If you have used the first version of this system, how do you compare the 1 st
version to the 2nd version with regard to:
I. The process selection
a- Same
Comment:
II. The material selection:
a- Same
Comment:
222
III. Finishing selection
a- Same
Comment:
IV. Machine selection
a- Same
Comment:
Answers:
Same 2nd version better 1st version better

Process Selection 2
Material Selection 2
Finishing Selection 1 1
Machine Selection 1 1
Table B-17: Responses to General Questions Part Q10
Question 11:
Please comment on the following capabilities of the AMDSS:
III. Determining minimum and maximum values of the material properties.
223
Answers:

Good Improvement 2/3 Good Improvement 1/3 Good Improvement 3/6
Essential Option 1/3 Essential Option 0/3 Essential Option 1/6
Very Useful 0/3 Very Useful 1/3 Very Useful 1/6
Table B-18: Responses to General Questions Part Q11 (I)

Table B-19: Responses to General Questions Part Q11 (II).
III. Determining minimum and maximum values of the material properties.

Table B-20: Responses to General Questions Part Q11 (III).
224
Table B-21: Responses to General Questions Part Q11 (IV).
Question 12:
Do you have any suggestions for improving the AMDSS?
Answers:

Add Material Class 1/3 Add Material Class 0/3 Add Material Class 1/6
Appearance 0/3 Appearance 1/3 Appearance 1/6
Nothing 0/3 Nothing 1/3 Nothing 1/6
225
References
Amano, R. S. and Rohatgi, P. K. (2011) 'Laser engineered net shaping process for SAE 4140 low
alloy steel', Materials Science and Engineering: A, Vol. 528, (22-23), pp. 6680-6693.
Arcam (2012a) Available at: http://www.arcam.com/technology/faq/electron-beam-melting-

ebm.aspx (Accessed: March 2012).
Arcam (2012b) Available at: http://www.arcam.com/technology/ebm-materials.aspx (Accessed:

April 2012).
Armillotta, A. (2006) 'Assessment of surface quality on textured FDM prototypes', Rapid

Prototyping Journal, Vol.12, (1), pp. 35-41.
Armillotta, A. (2007) 'Selection of layered manufacturing techniques by an adaptive AHP decision

model', Robotics & Computer-Integrated Manufacturing 24, pp. 450-461.
ASTM. (2012) ASTM Standard F2792-12a,"Standard Terminology for Additive Manufacturing

Technologies", ASTM International, West Conshohocken, PA, 2012, DOI: 10.1520/F2792-
10 , www.astm.org.
Badev, A., Abouliatim, Y., Chartier, T., Lecamp, L., Lebaudy, P., Chaput, C. and Delage, C. (2011)
'Photopolymerization kinetics of polether acrylate in the presence of ceramic fillers used
in sterolithography', Journal of Photochemistry and Photobiology A: Chemistry, Vol.222,
pp. 117-122.
Bernard, A. (1999) 'Knowledge based system for the choice of rapid prototyping process',
Proceedings of the 10th solid freeform fabrication Symposium, Austin, Texas, pp. 39-45.
Bernard, A., Deglin, A. and Ris, G. (2003) 'An original approach for the memorisation and the
generation of rapid product development processes', Rapid Prototyping Journal, Vol.9,
(2), pp. 58-67.
Bibb, R. (1999) The development of a rapid prototyping selection system for small companies.
PhD thesis. University of Wales.
Byun, H.-S. and Lee, K. H. (2005) 'A decision support system for the selection of a rapid
prototyping process using the modified TOPSIS method', International Journal of
Advanced Manufacturing Technology, Vol.26, pp. 1338-1347.
Campbell, R. I. and Bernie, M. R. N. (1996) 'Creating a database of rapid prototyping system

capabilities', Journal of Materials Processing Technology, Vol.6, pp. 163-167.
Chartier, T., Badev, A., Abouliatim, Y., Lebuady, P. and Lecamp, L. (2012) 'Stereolithography
process: influence of the rheology of silica suspensions and the medium on
polymerization kinetics-cured depth and width', Journal of the European Ceramic
Society, Vol.32, pp. 1625-1634.
Cheah, C. M., Chua, C. k., Lee, C. W., Feng, C. and Totong, K. (2005) 'Rapid prototyping and
tooling techniques: a review of applications for rapid investment casting', International
Journal of Advanced Manufacturing Technology, Vol.25, pp. 308-320.
226
Chen, Y., Hsu, C.-Y., Liu, L. and Yang, S. (2012) 'Constructing a nutrition diagnosis expert system',
Expert Systems with Applications, Vol.39, (2), pp. 2132-2156.
Choi, J.-W., Kim, H.-C. and Wicker, R. (2011) 'Multi-material stereolithography', Journal of
Materials Processing Technology, Vol.211, (3), pp. 318-328.
Choia, J.-W., Medinaa, F., Kima, C., Espalina, D., Rodrigueza, D., Stuckerc, B. and Wicker, R.
(2011) 'Development of a mobile fused deposition modeling system with enhanced
manufacturing flexibility', Journal of Materials Processing Technology, Vol. 211, pp. 424-
432.
Chua, C. K., Howe, C. T. and Hoe, E. K. (1998) 'Integrating rapid prototyping and tooling with
vacuum casting for connectors', The International Journal of Advanced Manufacturing
Technology, Vol.14, pp. 617-623.
Custompart.net (2012) Available at: http://www.custompartnet.com/ (Accessed: March 2012).
Dahotre, N. B. and Harimkar, S. P. (2008) 'Laser-Based Rapid Prototyping Processes chp9 in Laser
Fabrication and Machining of Materials', in Springer US, pp. 353-411.
Das, S., Wohlert, M., Beaman, J. J. and Bourell, D. L. (1999) 'Processing of titanium net shapes
SLS/HIP', Materials & Design, Vol.20, (2), pp. 115-121.
Derby, B. (2011) 'Inkjet printing ceramics: from drops to solid', Journal of the European Ceramic
Society, Vol.31, pp. 2543-2550.
Dias, R. T., Mateus, A., Mitchell, G. R. and Bartolo, P. J. (2012) Computer modeling and simulation
of reaction injection moulding: filling and curing stages. 2012 Taylor & Francis Group:
London.
Dippenaar, D. J. and Schreve, K. (2012) '3D printing tooling for vacuum- assisted resin transfer
moulding', International Journal of Advanced Manufacturing Technology, DOI
10.1007/s00170-012-4034-2.
Durr, H. and Kaschka, U. (1998) Proceedings of European Conference on Rapid Prototyping and
Manufacturing. Aachen,
Edwards, W. and Barron, F. H. (1994) 'SMARTS AND SMARTERS: improved simple methods for
multi attribute utility measurement', Organizational Behavior and Human Decision
processes, Vol. 60, pp. PP 306-325.
Fathi, S., Dickens, P. and Hague, R. (2012) 'Jetting stability of molten caprolactam in an additive
inkjet manufacturing process', The International Journal of Advanced Manufacturing
Technology, VOL.59.
Freiberg, M., Striffler, A. and Puppe, F. (2012) 'Extensible Prototyping for pragmatic engineering
of knowledge-based systems', Expert Systems with Applications, Vol.39, (11), pp. 10177-
10190.
Galantucci, L. M., Lavecchia, F. and Percoco, G. (2009) 'Experimental study aiming to enhance
the surface finish of fused deposition modeled parts', CIRP Annals - Manufacturing
Technology, Vol.58, (1), pp. 189-192.
227
Ghazy, M. and Dalgarno, K. (2009) 'A Framework for Layered Manufacturing Process Selection',
9th International Conference on Production Engineering, Design, and Control (PEDAC).
Alexandria, Egypt,
Ghazy, M. and Dalgarno, K. (2011) 'Development of layer manufacturing decision support

system', 16th European Forum on Rapid Prototyping and Manufacturing, AEPR. Paris
France, June 2011.
Ghazy, M., Dalgarno, K. and Saleh, O. (2010) 'Rapid tooling process for sand casting patterns',
Computer Aided Production Engineering (CAPE 21). Edinburgh,UK, April 2010.
Ghazy, M., Dalgarno, K. and Saleh, O. (2011) 'Finishing process for nylon casting patterns made
for SLS', Computer Aided Production Engineering (CAPE 22). Edinburgh, Uk, April 2011.
Gibson, I. (2002) Software Solutions for Rapid Prototyping. Professional Engineering: London.
Gibson, I., Rosen, D. W. and Stucker, B. (2010) Guidelines for process selection. Chp.12 in
Additive Manufacturing Technologies. Springer Science + Business Media.
Giner-Santonja, G., Beltran, P. A. and Ferragut, J. N. (2012) 'The application of the analytic
network process for the assessment of best available techniques', Journal of Cleaner
Production, Vol.25, pp. 86-95.
Grida, I. and Evans, J. R. G. (2003) 'Extrusion free forming of ceramics through fine nozzles',
Journal of the European Ceramic Society, Vol.23, pp. 629-635.
Hague, R., Mansour, S. and Saleh, N. (2003) 'Design opportunities with rapid manufacturing',
Assembly Automation, Vol.23, (4), pp. 346-356.
Hague, R., Mansour, S. and Saleh, N. (2004) 'Material and design considerations for rapid
manufacturing', International Journal of Production Research, Vol.42, (22), pp. 4691-
4708.
Hanemann, T., Bauer, W., Knitter, R. and Woias, P. (2006) "Rapid Prototyping and Rapid Tooling
Techniques for the Manufacturing of Silicon, Polymer, Metal and Ceramic Micro devices.
Chapter 4 in " MEMS/NEMS Handbook - Techniques and Applications. Volume 3.
Leondes, Cornelius T (eds). Springer.
Hanumaiah, N., Ravi, B. and Mukherjee, N. P. (2006) 'Rapid hard tooling process selection using
QFD-AHP methodology', Journal of Manufacturing Technology Management, Vol.17, (3),
pp. 332-350.
Hanumaiah, N., Ravi, B. and Mukherjee, N. P. (2007) 'Rapid tooling manufacturability evaluation
using fuzzy-AHP methodology', International Journal of Production Research, Vol.45, (5),
pp. 1161-1181.
Hanumaiah, N., Subburaj, K. and B.Ravi. (2008) 'Computer aided tooling process selection and
manufacturability evaluation for injection mold development', Computers in Industry,
Vol.59, pp. 262-276.
Hermann, A. and Allen, J. K. (1999) ASME Design Engineering Technical Conference. Las Vegas,
Nevada,
Hon, K. K. B. and Gill, T. J. (2003) 'Selective Laser Sintering of SiC/Polyamide Composites', CIRP
Annals - Manufacturing Technology, Vol.52, (1), pp. 173-176.
228
Hopkinson, N., Hague, R. and Dickens, P. M. (2006) Rapid Manufacturing: An Industrial
Revolution for the Digital Age. John Wiley & Sons,Ltd.
Hornberger, L. E. (1993) Rapid Prototyping Program. Santa Clara,California: Santa Clara

University
IDEF (2012) Available at: http://www.idef.com/ (Accessed: June 2012).
Ilyas, I., Taylor, C., Dalgarno, K. and Gosden, J. (2010) 'Design and manufacture of injection
mould tool inserts produced using indirect SLS and machining processes', Rapid
Prototyping Journal, Vol.16, (6).
IVF (2005) RP Selector. Available at: http://extra.ivf.se/rp-selector/ (Accessed: 10 March 2012).
Jones, S. and Yuan, C. (2003) 'Advances in shell moulding for investment casting', Journal of
Materials Processing Technology, Vol.135, pp. 258-265.
Kappes, S. (1997) 'Putting your IDEF0 model to work', Business Process Management, VOL 3, (2),
pp. PP 151-161.
KARMA (2010) Knowledge Based Process planning and Design for Additive Layer Manufacturing
(KARMA Project). Available at: http://cordis.europa.eu/projects/94835_en.html
(Accessed: March 2012).
Kascha, U. and Auerbach, P. (2000) 'Selection and evaluation of rapid tooling process chains with
protol', Rapid Prototyping Journal, Vol.6, (1), pp. 60-65.
Kendal, S. L. and Creen, M. (2007) 'Types of Knowledge-Based Systems Chp.2 in "An Introduction
to Knowledge Engineering"', in Springer London, pp. 26-88.
Khalate, A. A., Bombois, X., Babuka, R., Wijshoff, H. and Waarsing, R. (2011) 'Performance
improvement of a drop-on-demand inkjet printhead using an optimization-based
feedforward control method', Control Engineering Practice, 19, (8), pp. 771-781.
Khrais, S., Al-Hawari, T. and Al-Araidah, O. (2011) 'A fuzzy logic application for selecting layered
manufacturing techniques', Expert Systems with Applications, Vol.38, pp. 10286-10291.
Lai, M.-C., Huang, H.-C. and Wang, W.-K. (2011) 'Designing a knowledge-based system for
benchmarking: A DEA approach', Knowledge-Based Systems, Vol.24, (5), pp. 662-671.
Lamikiz, A., Snchez, J. A., Lpez de Lacalle, L. N. and Arana, J. L. (2007) 'Laser polishing of parts
built up by selective laser sintering', International Journal of Machine Tools and
Manufacture, Vol.47, (12-13), pp. 2040-2050.
Lan, H. (2009) 'Web-based rapid prototyping and manufacturing systems: A review', Computers
In Industry, Vol.60, pp. 643-656.
Lan, H., Ding, Y. and Hong, J. (2005) 'Decision support system for rapid prototyping process
selection through integration of fuzzy synthetic evaluation and an expert system ',
International Journal of Production Research, Vol. 43, (1).
Laoui, T. (2010) An approach to develop a rapid manufacturing knowledge-based environment -

Bartolo et al. (eds). Taylor & Francis Group: London.
229
Leong, K. F., Chua, C. K., Chua, G. S. and Tan, C. H. (1998) 'Abrasive jet deburring of jewellery
models built by stereolithography apparatus (SLA)', Journal of Materials Processing
Technology, Vol.83, (1-3), pp. 36-47.
Levy, G. N., Schindel, R. and kruth, J. P. (2003) 'Rapid manufacturing and rapid tooling with layer
manufacturing (LM) technologies, state of the art and future perspectives', CIRP Annals,
Manufacturing Technology, Vol.52, (2), pp. 589-609.
Li, B. M., Xie, S. Q. and Xu, X. (2011) 'Recent development of knowledge-based systems, methods
and tools for One-of-a-Kind Production', Knowledge-Based Systems, Vol.24, (7), pp.
1108-1119.
Lu, J., Chen, S., Wang, W. and van Zuylen, H. (2012) 'A hybrid model of partial least squares and
neural network for traffic incident detection', Expert Systems with Applications, Vol.39,
(5), pp. 4775-4784.
Lu, W., Lin, F., Han, J., Qi, H. and Yan, N. (2009) 'Scan Strategy in Electron Beam Selective
Melting', Tsinghua Science & Technology, Vol. 14, Supplement 1, pp. 120-126.
Mahesh, M., Fuh, J. Y. H., Wong, Y. S. and Loh, H. T. (2005) 'Benchmarking for decision making in
rapid prototyping systems', Proceedings of the IEEE, International conference on
automation science and engineering Edmonton, Canada,
Mahesh, M., Loh, H. T., Wong, Y. S. and Fuh, J. Y. H. (2003) 'Integrated decision support system
for selection of RP processes', The fourteenth Annual International Solid Freeform
Fabrication Symposium University of Texas, USA,
Masood, S. H. and Al-Alawi, M. (2002) 'The IRIS rapid prototyping system selector for educational
and manufacturing users', International Journal of Engineering Education, Vol.18, (1), pp.
66-77.
Masood, S. H. and Song, W. Q. (2004) 'Development of new metal/polymer materials for rapid
tooling using Fused deposition modelling', Materials & Design, 25, (7), pp. 587-594.
Masood, S. H. and Soo, A. (2002) 'A rule based expert system for rapid prototyping system
selection', Robotics & Computer-Integrated Manufacturing, Vol.18, pp. 267-274.
Materialise (2012) Available at: http://www.materialise.com/vacuum-casting (Accessed: March

2012).
Melchels, F. P. W., Feijen, J. and Grijpma, D. W. (2010) 'A review on stereolithography and its
applications in biomedical engineering', Biomaterials, Vol.31, (24), pp. 6121-6130.
MIT (2012) Available at: http://web.mit.edu/tdp/www/whatis3dp.html (Accessed: March 2012).
Moeskopf, E. and Feenstra, F. (2008) 'Chp5. Introduction to Rapid Prototyping ', in Reverse
Engineering Springer Series in Advanced Manufacturing, pp. 99-117.
Mognol, P., Rivette, M., Fegou, L. and Lesprier, T. (2007) 'A first approach to choose between
HSM,EDM and DMLS processes in hybrid rapid tooling', Rapid Prototyping Journal, Vol.
13, (1).
Muller, H. (1999) 'The decision dilemma assessment and selection of rapid prototyping process
chains', Proceedings of the 8th European conference on Rapid Prototyping and
Manufacturing. Nottingham, UK,
230
Muller, H., Bauer, J. and Klingenberg, H. H. (1996) 'Computer based rapid prototyping system
selection and support', Proceedings of the Time Compression Technologies Conference.
The Heritage Motor Center, Gaydon, Uk.,
Munguia, J. (2009) RMADS: Development of a concurrent rapid manufacturing advice system.

PhD. thesis. Technical University of Catalonia.
Munguia, J., Lloveras, J., llorens, S. and Laoui, T. (2010) 'Development of an AI-based system
manufacturing advice system', International Journal of Production Research, Vol.48, (8),
pp. 2261-2278.
Murr, L. E., Gaytan, S. M., Ramirez, D. A., Martinez, E., Hernandez, J., Amato, K. N., Shindo, P. W.,
Medina, F. R. and Wicker, R. B. (2012) 'Metal Fabrication by Additive Manufacturing
Using Laser and Electron Beam Melting Technologies', Journal of Materials Science
& Technology, Vol.28, (1), pp. 1-14.
Myers, M. D. and Newman, M. (2007) 'The qualitative interview in IS research: Examining the
craft', Information and Organization, Vol. 17, (1), pp. 2-26.
Nagahanumaiah, Subburaj, K. and B.Ravi. (2008) 'Computer aided rapid tooling selection and
manufacturability evaluation for injection molding development', Computers In Industry,
Vol.59, (2-3), pp. 262-276.
Noguera, R., Lejeune, M. and Chartier, T. (2005) '3D fine scale ceramic components formed by
ink-jet prototyping process', Journal of the European Ceramic Society, Vol.25, pp. 2055-
2059.
Olugu, E. U. and Wong, K. Y. (2012) 'An expert fuzzy rule-based system for closed-loop supply
chain performance assessment in the automotive industry', Expert Systems with
Applications, Vol.39, (1), pp. 375-384.
Pal, D. K. and Ravi, B. (2007) 'Rapid tooling route selection and evaluation for sand and
investment casting', Virtual and Physical Prototyping Journal, Vol 2, (4), pp. 197-207.
Pal, D. K., Ravi, B. and Bhargava, L. S. (2007) 'Rapid tooling route selection for metal casting using
QFD-ANP methodology', International Journal of Computer Integrated Manufacturing,
Vol 20, (4), pp. 338-354.
Parasad, P. S. R., Reddy, A. V., Rajesh, P. K., Ponnambalam, p. and Prakasan, K. (2006) 'Studies on
reology of ceramic inks and spread of ink droplets for direct ceramic inkjet printing',
Journal of Materials Processing Technology, Vol.176, pp. 222-229.
Petrovic, V., Jorda, O., Martin, A., Rannar, L.-E. and Dejanovic, S. (2011) 'Knowledge assisted
rapid manufacturing (KARMA Project)', Proceedings of AEPR'11, 16th European Forum on
Rapid Prototyping and Manufacturing. Paris, France,
Pham, D. T. and Dimov, S. S. (2003) 'Rapid Prototyping and Rapid Tooling- the key enablers for
rapid manufacturing', Proceedings of the Institution of Mechanical Engineering: Journal
of Mechanical Engineering Science, Vol. 217, (part C).
Phillipson, D. K. (1997) 'Rapid prototyping machine selection programme', 6th European

Conference on Rapid Prototyping and Manufacturing. Nottingham, UK,
231
Pratt, M. J., Bhatt, A. D., Dutta, D., Lyons, K. W., Patil, L. and Sriram, R. D. (2002) 'Progress
towards an international standard for data transfer in rapid prototyping and layered
manufacturing', Computer-Aided Design, Vol.34, (14), pp. 1111-1121.
Rahmati, S., Rezaei, M. R. and Javad, A. (2009) 'Design and manufacture of a wax injection tool
for investment casting using rapid tooling', Tsinghua Science & Technology, Vol.18, (38),
pp. 108-115.
Rao, R. V. and Padmanabhan, K. K. (2007) 'Rapid prototyping process selection using graph
theory and matrix approach', Journal of Materials Processing Technology, Vol.194, pp.
81-88.
Sahin, S., Tolun, M. R. and Hassanpour, R. (2012) 'Hybrid expert systems: A survey of current
approaches and applications', Expert Systems with Applications, Vol.39, (4), pp. 4609-
4617.
Smith, P. and Rennie, A. (2008) 'Development of an additive layer manufacturing (ALM) selection
tool for direct manufacturing of products', The Ninetieth Annual International Solid
Freeform Fabrication Symposuim, Austin, Texas, pp. 507-518.
Smith, P. and Rennie, A. E. W. (2010) 'Computer Aided Material Selection (CAMS) for Additive
Manufacturing Materials. ', Virtual and Physical Prototyping Journal, Vol.5, (4), pp. 209-
213.
Soe, S. P. (2012) 'Quantitative analysis on SLS part curling using EOS P700 machine', Journal of
Materials Processing Technology , http://dx.doi.org/10.1016/j.jmatprotec.2012.06.012.
Sood, A. K., Ohdar, R. K. and Mahapatra, S. S. (2010) 'Parametric appraisal of mechanical

property of fused deposition modelling processed parts', Materials & Design, 31,
(1), pp. 287-295.
Su, B., Dhara, S. and Wang, L. (2008a) 'Green ceramic machining: A top-down approach for the
rapid fabrication of complex-shaped ceramics', Journal of the European Ceramic Society,
Vol.28, (11), pp. 2109-2115.
Su, B., Dhara, S. and Wang, L. (2008b) 'Green ceramic machining: a top-down approach for the
rapid fabrication of complex shaped ceramics', Journal of the European Ceramic Society,
Vol.28, pp. 2109-2115.
Tang, H.-H., Chiu, M.-L. and Yen, H.-C. (2011) 'Slurry-based selective laser sintering of polymer-
coated ceramic powders to fabricate high strength alumina parts', Journal of the
European Ceramic Society, Vol.31, (8), pp. 1383-1388.
Thian, S. C. H., Tang, Y., Tan, W. K., Fuh, J. Y. H., Wong, Y. S., Loh, H. and Lou, L. (2008) 'The
manufacture of micromould and microparts by vacuum casting', The International
Journal of Advanced Manufacturing Technology, Vol 38, (9-10), pp. 944-948.
Tomeckova, V. and Halloran, J. (2012) 'Cure depth for photopolymerization of ceramic

suspensions', Journal of the European Ceramic Society, Vol.30, pp. 3023-3033.
Valentan, B., Brajilih, T., Drstvensek, I. and Balie, J. (2011) 'Development of a part- complexity
evaluation model for application in additive fabrication technologies', Journal of
Mechanical Engineering, Vol.57, (10), pp. 709-718.
232
VG (2012) Available at: http://www.vg-kunst.de/en/information/dictionary/dictionary.html
(Accessed: September 2012).
Vouk, D., Malus, D. and Halkijevic, I. (2011) 'Neural networks in economic analyses of
wastewater systems', Expert Systems with Applications, Vol.38, (8), pp. 10031-10035.
Wohlers, T. (2010) Wohlers Report 2010:Additive Manufacturing state of the industry annual
worldwide progress report.
Wong, Y. S., Fuh, J. Y. H., Loh, H. T. and Mahesh, M. (2002) Rapid Prototyping and Manufacturing
(RP&M) benchmarking, Chp3. in Software Solutions for RP, (Ed) Gibson I. PEP Ltd: UK.
XU, F., Wong, Y. S. and Loh, H. T. (2000) 'Toward generic models for comparative evaluation and
process selection in rapid prototyping and manufacturing', Journal of Manufacturing
Systems, Vol.19, (5), pp. 283-296.
Yasa, E., Kruth, J. P. and Deckers, J. (2011) 'Manufacturing by combining Selective Laser Melting
and Selective Laser Erosion/laser re-melting', CIRP Annals - Manufacturing Technology,
Vol. 60, (1), pp. 263-266.
Yves-Christian, H., Jan, W., Wilhelm, M., Konrad, W. and Reinhart, P. (2010) 'Net shaped high
performance oxide ceramic parts by selective laser melting', Physics Procedia, Vol.5, Part
B, pp. 587-594.
Zadeh, L. A. (1965) 'Fuzzy sets', Information and Control, 8, (3), pp. 338-353.
Zemnick, C., Woodhouse, S. A., Gewanter, R. M., Raphael, M. and Piro, J. D. (2007) 'Rapid
prototyping technique for creating a radiation shield', The Journal of Prosthetic
Dentistry, Vol.97, (4), pp. 236-241.
Zhang, Y. and Liu, H. (2009) 'Application of Rapid Prototyping Technology in Die Making of Diesel
Engine', Tsinghua Science & Technology, Vol.14, Supplement 1, (0), pp. 127-131.
Zhao, J., Cao, W., Ge, C., Tan, Y., Zhang, Y. and Fei, Q. (2009) 'Research On Laser Engineering Net
Shaping of thick-wall nickel based alloy parts', Rapid Prototyping Journal, Vol. 15, (1), pp.
24-28.
Zuo, X. (2012) 'Experimental Research on Vacuum Casting of micro-gears', Advanced Materials

Research, 383-390, pp. 1714-1718.
233

Ghazy 12 PDF

Uploaded by

Copyright:

Available Formats

Ghazy 12 PDF

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Ghazy 12 PDF

Uploaded by

Copyright:

Available Formats

Development of an Additive

Manufacturing Decision Support

A thesis submitted to the Faculty of Science,

School of Mechanical and Systems Engineering

Additive manufacturing (AM) technology describes a set of processes capable of

I want to express my gratitude to the Egyptian Factories that offered me case

Furthermore, I would like to thank my University in Egypt: The Arab Academy

Chapter 2. Literature Survey ...................................................................... 6

Chapter 3. Additive Manufacturing Decision Support System (AMDSS)

Chapter 4. Verification and Validation.................................................... 82

Chapter 5. User Feedback and Evaluation ............................................ 101

Chapter 6. Additive Manufacturing Decision Support System Further

Chapter 7. Discussion ............................................................................... 169

Chapter 8. Conclusions and Future Work............................................. 179

Appendix A. First Questionnaire Responses ........................................... 183

Table 2-1: Different Attempts for Additive Manufacturing Selection. .......................................... 32

Figure 2.1: Additive Manufacturing Different Applications (Wohlers,2010) ................................... 8

ABS Acrylonitrile Butadiene Styrene

Additive Manufacturing (AM) is defined by ASTM as the process of joining

1.2 Research Problem

Capital investment support. What technology should a user buy to fulfill

Process planning support. What are the parameters that need to be

The growing numbers of AM technologies and the quick advance in their

1.3 Research Aim and Objectives

The aim of this research is to develop an additive manufacturing decision support

1. Exploring different additive manufacturing processes and examining the

2. Exploring different additive manufacturing materials, machines and

3. Using decision analysis tools to make selection among different layer

4. Developing a system that is capable of assisting decision makers in:

I. Determining the feasibility of additive manufacturing processes for

II. Selecting the most suitable process.

5. Testing the developed system by applying verification and validation

1.4 Research Road Map

In order to develop an additive manufacturing decision support system, nine main

1- Analyzing available additive manufacturing (AM ) selector systems.

2- Identifying shortcomings of the available selector systems.

3- Identifying system target specifications.

4- Developing the system.

5- Verifying and validating the system.

6- Obtaining feedback from users.

7- Improving the system based on feedback.

8- Obtaining feedback for the improved system from users.

9- Launching the final version.

1.5 Thesis Outline

This chapter gives an overview about additive manufacturing technologies (AM).

2.1 Additive Manufacturing Overview

Additive Manufacturing (AM) is the process of producing 3D physical parts from

The following steps describes a basic AM process (Cheah et al.,2005, Moeskopf

2- Converting the CAD model into STL format (standard triangulation

6- Post processing which includes cleaning, finishing the model and

2.2 Additive Manufacturing Application Areas

2.2.1 Rapid Prototyping

Rapid Prototyping (RP) is using AM technology to produce prototypes. The

2.2.2 Rapid Manufacturing

Rapid Manufacturing (RM) is using AM technology to produce end product parts

2.2.3 Rapid Tooling

Rapid tooling (RT) is using AM technology to produce a tool or a die. RT can be

Figure 2.1 shows different AM applications. Rapid prototyping (visual aids,

Direct part Visual aids,