Engineering Design
Engineering Design
1. INTRODUCTION....................................................................................................................... 6
1.1 BACKGROUND.......................................................................................................................... 6
1.2 ENGINEERING DESIGN - A FORM OF INFORMATION PROCESSING................................................. 6
1.3 RESEARCH IN ENGINEERING DESIGN.......................................................................................... 7
1.4 SCOPE AND MOTIVATION OF THIS THESIS .................................................................................. 8
1.5 STRUCTURE OF THIS THESIS...................................................................................................... 9
2. SCIENTIFIC FRAMEWORK FOR RESEARCH IN DESIGN THEORY............................ 11
2.1 DEFINITION OF SCIENTIFIC THEORY AND DISCUSSION ON HOW THEORIES CAN BE USED............ 11
2.2 RESEARCH PROCESS MODEL FOR DESIGN THEORY................................................................... 12
3. CONTEXT OF THIS THESIS ................................................................................................. 14
3.1 INDUSTRIAL CONTEXT: THE DESIGN PROCESS ......................................................................... 14
3.1.1 Introduction to the design process roadmap..................................................................... 15
3.1.2 Putting together a project specific design process from the activities in the roadmap....... 16
3.1.3 Description of the generic activities in the design activity collection ................................ 17
3.1.3.1 Project control and decomposition........................................................................................................... 17
3.1.3.2 Design object analysis............................................................................................................................. 17
3.1.3.3 Problem formulation ............................................................................................................................... 18
3.1.3.4 Decoupling (conflict resolution) .............................................................................................................. 19
3.1.3.5 Concept Generation and Selection........................................................................................................... 20
3.1.3.6 Trade-off................................................................................................................................................. 20
3.1.3.7 Implementation....................................................................................................................................... 21
3.2 ACADEMIC CONTEXT: CONTEMPORARY DESIGN THEORY RESEARCH....................................... 22
3.2.1 Motivation and Scope of selected design methods ............................................................ 22
3.2.1.1 Altshullers motivation and scope............................................................................................................ 22
3.2.1.2 Clausings motivation and scope ............................................................................................................. 23
3.2.1.3 Suhs motivation and scope..................................................................................................................... 23
3.2.1.4 The Workshop Design Konstruktion (WDK) Schools motivation and scope............................................ 24
3.2.1.5 Discussion on scope of design methods ................................................................................................... 24
3.2.2 Approaches to develop new design methods ..................................................................... 24
3.2.2.1 Altshullers approach.............................................................................................................................. 24
3.2.2.2 Clausings approach................................................................................................................................ 25
3.2.2.3 Suhs approach ....................................................................................................................................... 26
3.2.2.4 The WDK Schools approach .................................................................................................................. 26
3.2.2.5 Discussion on approaches to develop design methods .............................................................................. 27
3.2.2.6 Approach followed to develop the method presented in this thesis ........................................................... 28
4. EXPANDED FRAMEWORK OF INFORMATION IN DESIGN .......................................... 29
4.1 RESEARCH QUESTION ............................................................................................................. 29
4.2 RESEARCH METHOD ............................................................................................................... 29
4.3 LITERATURE STUDIES ............................................................................................................. 29
4.4 EMPIRICAL STUDY OF HOW ENGINEERING INFORMATION IS SYNTHESIZED IN A DESIGN PROJECT30
4.5 RESULTS ................................................................................................................................ 33
4.6 HYPOTHESES H1 AND H2 ....................................................................................................... 37
4.7 TESTING HYPOTHESES H1 AND H2......................................................................................... 37
4.7.1 Case study background ................................................................................................... 37
4.7.2 Effectively searching for potential design solutions.......................................................... 38
4.7.3 Design of the Depth Charge Initiator............................................................................... 39
4.7.3.1 Problem definition .................................................................................................................................. 39
4.7.3.2 Top level design...................................................................................................................................... 39
4.7.3.3 Decomposing the initiator (FR1) ............................................................................................................. 40
4.7.3.4 Environmental Analysis .......................................................................................................................... 40
4.7.3.5 Searching for DPs in the environmental phases ....................................................................................... 41
4.7.3.6 Choosing DPs. ........................................................................................................................................ 43
4.7.3.7 Design of sub-systems............................................................................................................................. 44
4.7.4 Final comments on the case study .................................................................................... 45
4.8 CONCLUSION.......................................................................................................................... 46
5. MANAGING INFORMATION IN THE FRAMEWORK ...................................................... 47
5.1 RESEARCH QUESTION............................................................................................................. 47
5.2 RESEARCH METHOD............................................................................................................... 47
5.3 THEORY DEVELOPMENT......................................................................................................... 47
5.3.1 Finding problem sources in a design object ..................................................................... 48
5.3.2 Proposing an information system for field service............................................................ 51
5.3.3 Tracing the impact of an Engineering Change Order (ECO)............................................ 51
5.3.4 Tracing design decisions.................................................................................................. 53
5.4 COMPUTER SYSTEMS THAT REALIZE SOME OF THE FUNCTIONS IN THE ABOVE THEORY............ 54
5.4.1 Background ..................................................................................................................... 54
5.4.2 Design information system based on axiomatic design ..................................................... 56
5.4.2.1 Analyzing the design process and information flows ................................................................................ 56
5.4.2.2 Finding existing tools that satisfy the requirements ................................................................................. 57
5.4.2.3 Implementation - Case study ................................................................................................................... 57
5.4.3 QCAD - Quality Computer Aided Design......................................................................... 64
5.4.4 Discussion on the differences between the two computer implementations........................ 66
5.5 CONCLUSIONS ........................................................................................................................ 67
6. UTILIZING PRINCIPLE-BASED METHODS AND TOOLS TO EFFECTIVELY
GENERATE THE DESIGN OBJECT WITHIN THE FRAMEWORK.................................... 69
6.1 HYPOTHESIS - H3 ................................................................................................................... 70
6.2 THEORY GENERATION............................................................................................................ 70
6.3 USING AXIOMATIC DESIGN AND TIPS IN THE FRAMEWORK...................................................... 71
6.4 EXAMPLE: AN ENGINEERING CHANGE ORDER UTILIZING AXIOMATIC DESIGN AND TIPS WITHIN
THE FRAMEWORK......................................................................................................................... 73
6.5 CONCLUSIONS ........................................................................................................................ 76
7. LIMITATIONS ON THE APPLICABILITY OF THE FRAMEWORK............................... 78
7.1 RESEARCH QUESTION............................................................................................................. 78
7.2 HYPOTHESIS........................................................................................................................... 78
7.3 TESTING HYPOTHESIS H4........................................................................................................ 78
7.4 CONCLUSIONS ........................................................................................................................ 79
7.5 SAAB SERVICE PARTNERS BUSINESS PLAN. ........................................................................... 81
8. CONTRIBUTIONS, CONCLUSIONS AND FUTURE WORK.............................................. 82
8.1 SUMMARY OF THIS RESEARCH PROJECTS CONTRIBUTIONS TO THE ACADEMIC COMMUNITY.... 82
8.2 SUMMARY OF THE EFFECTS ON INDUSTRY AND KNOWLEDGE TRANSFER ................................. 82
8.3 CONCLUSIONS ........................................................................................................................ 83
8.4 PROPOSED FUTURE WORK BASED ON THIS THESIS ................................................................... 84
8.4.1 Future research ............................................................................................................... 84
8.4.2 Towards the Thinking Design Machine (TDM)................................................................. 85
9. REFERENCES.......................................................................................................................... 87
PART II APPENDICES
APPENDIX 1: SCIENTIFIC METHOD
APPENDIX 2: AXIOMATIC DESIGN
APPENDIX 3: TIPS: THE THEORY OF INVENTIVE PROBLEM SOLVING
APPENDIX 4: EXCERPT FROM SAAB SERVICE PARTNERS BUSINESS PLAN 1994-1998
LIST OF EQUATIONS
EQUATION 4-1.................................................................................................................................. 32
EQUATION 4-2.................................................................................................................................. 33
EQUATION 4-3.................................................................................................................................. 39
EQUATION 4-4.................................................................................................................................. 40
EQUATION 4-5.................................................................................................................................. 43
EQUATION 4-6.................................................................................................................................. 44
EQUATION 4-7.................................................................................................................................. 44
EQUATION 4-8.................................................................................................................................. 45
EQUATION 4-9.................................................................................................................................. 45
EQUATION 5-1.................................................................................................................................. 57
EQUATION 6-1.................................................................................................................................. 73
EQUATION 6-2.................................................................................................................................. 74
EQUATION 6-3.................................................................................................................................. 74
EQUATION 6-4.................................................................................................................................. 74
LIST OF FIGURES
FIGURE 1-1 IMPORTANCE OF EARLY DESIGN DECISIONS...................................................................... 6
FIGURE 1-2 PROCESSING INFORMATION IN THE DESIGN PROCESS........................................................ 7
FIGURE 2-3. MODEL OF THE RESEARCH PROCESS FOR DESIGN THEORY............................................. 13
FIGURE 3-1. THE DESIGN PROCESS ROADMAP (FROM [TATE96])...................................................... 16
FIGURE 3-2 FEEDBACK CONTROL LOOP DEPICTING THE SYNTHESIS AND ANALYSIS DURING THE
CONCEPT GENERATION AND SELECTION ACTIVITY (FROM [WILS79])......................................... 20
FIGURE 4-1 MANUFACTURING REQUIREMENTS ................................................................................ 30
FIGURE 4-2 GRAVITATION COMPONENTS DEPEND ON ORIENTATION OF TUBE.................................... 32
FIGURE 4-3 INFORMATION FRAMEWORK.......................................................................................... 34
FIGURE 4-4 INITIATOR...................................................................................................................... 38
FIGURE 4-5 ANALYZING THE ENVIRONMENT .................................................................................... 41
FIGURE 4-6 SUPPLYING ELECTRICITY............................................................................................... 44
FIGURE 4-7 GENERATING ARMING CONDITION 1............................................................................... 45
FIGURE 5-1 HIERARCHIES OF FRS, DPS, AND PVS............................................................................ 48
FIGURE 5-2 UNSATISFIED TOP LEVEL FR.......................................................................................... 49
FIGURE 5-3 TRACING UNSATISFIED FRS ........................................................................................... 49
FIGURE 5-4 FINDING AN FRS CORRESPONDING DP.......................................................................... 50
FIGURE 5-5 FINDING A DPS CORRESPONDING PV............................................................................ 50
FIGURE 5-6 PROPOSED ENGINEERING CHANGE ORDER IN DP HIERARCHY. ....................................... 52
FIGURE 5-7 RELATING FR AND DP HIERARCHIES THROUGH THE DESIGN MATRIX. ............................ 52
FIGURE 5-8 IMPACT OF AN ENGINEERING CHANGE ORDER ................................................................ 53
FIGURE 5-9 CAD MODEL OF SUBSYSTEM IN PACKAGING MACHINE (FROM TETRA PAK).................... 55
FIGURE 5-10 DESIGN PROCESS MODEL SHOWING THE MAIN DESIGN PHASES (ADAPTED FROM P.
ANDERSSON, TETRA PAK)......................................................................................................... 56
FIGURE 5-11 RELATING THE DESIGN PROCESS TO INFORMATION INFRASTRUCTURE (ADAPTED FROM P.
ANDERSSON, TETRA PAK)......................................................................................................... 56
FIGURE 5-12 FR TREE FOR PART OF PACKAGING MACHINE............................................................... 58
FIGURE 5-13 DP TREE FOR PART OF PACKAGING MACHINE............................................................... 58
FIGURE 5-14 DESIGN INFORMATION ABOUT PART OF PACKAGING MACHINE REPRESENTED IN A FORM
OF FUNCTION MEANS TREE (ADAPTED FROM M. SJGREN, TETRA PAK) .................................... 60
FIGURE 5-15 CAD MODEL AT A-LEVEL SHOWING THE PRODUCT MODEL AS WELL AS INFORMATION
GENERATED AT A HIGH LEVEL OF ABSTRACTION. (FROM TETRA PAK) ....................................... 61
FIGURE 5-16 CAD MODEL AT B-LEVEL SHOWING THE PRODUCT MODEL AS WELL AS INFORMATION
GENERATED AT A SUB-SYSTEM LEVEL OF ABSTRACTION. (FROM TETRA PAK)............................ 62
FIGURE 5-17 CAD MODEL OF MECHANISM IN PACKAGING MACHINE WITH REFERENCES TO THE FR/DP
STRUCTURES IN FIGURE 5-12 AND FIGURE 5-13 AS WELL AS THE FUNCTION MEANS TREE IN
FIGURE 5-14. (ADAPTED FROM M. SJGREN, TETRA PAK) ........................................................ 63
FIGURE 6-18 DESIGN PROCESS AS A FEEDBACK LOOP (FROM [WILS79]) .......................................... 70
FIGURE 6-19 TIPS AND AXIOMATIC DESIGN IN THE FRAMEWORK..................................................... 72
FIGURE 6-20 CHECKING DPS AGAINST CONSTRAINTS....................................................................... 72
FIGURE 6-21 SCHEMATIC DRAWING OF A SWITCH............................................................................. 73
FIGURE 6-22 DIFFERENT DESIGN CONCEPTS THAT INCREASE THE CONTACT AREA OF A SWITCH........ 76
LIST OF TABLES
TABLE 1-1. SOME WORK IN DESIGN THEORY .................................................................................... 10
TABLE 6-1 IDENTIFYING COMPLEMENTARY PROPERTIES BETWEEN AXIOMATIC DESIGN, TIPS AND
THE FRAMEWORK...................................................................................................................... 71
TABLE 6-2 INVENTIVE PRINCIPLES FOR SOLVING THE TECHNICAL CONTRADICTION IN THE SWITCH ... 75
6
1. Introduction
1.1 Background
The research area of product development and manufacturing is currently receiving increasing attention
to address industry efforts to reduce lead time, cut development and manufacturing costs, increase
quality and product performance, and lower total life cycle cost.
In this context, focus has shifted from improving the performance during the later stages of the product
development cycle to the front end phases where product development take place at a higher level of
abstraction. This shift is motivated by studies and experience showing that design decisions made
during the early stages of the product development cycle have the largest impact on total cost of the
product (figure 0-1). It is often claimed that up to 80% of the products total cost is committed during
the concept development stage (for example, see [FRED94]).
Costs
Design Testing Conception Process
Planning
Production
100%
0
%
Committed
Cash
flow
Figure 0-1 Importance of early design decisions
1.2 Engineering design - a form of information processing
Engineering design is an information processing activity. The process begins by inputting information
about needs and constraints and ends with a complete description of the system that will satisfy these
needs (along with a description of how to realize such system) (Figure 0-2).
7
Design Process
Information
about
needs and
constraints
Sufficient information
to generate a system
that meets the needs
and constraints.
Figure 0-2 Processing information in the design process
During the design process, the designer must access existing information and generate new information.
Existing information is found both within and beyond the engineers current realm of knowledge. For
example, information may exist in the designers immediate surroundings (e.g., in the company) or
elsewhere (e.g., with the customer, with a vendor, or at a university). When this information has been
entered into the design, we are beginning to build up knowledge (systematized information) about the
design object and the process to manufacture it.
Thus, to perform engineering design that quickly leads to high quality products that can be realized at
low cost, the company must have:
access to information, and
competence and tools enabling them to quickly systematize the information into knowledge
about the design object along with knowledge about how to realize it.
Furthermore, in order to build a sustainable competitive advantage in the product development process,
the company must have one or more of the following
useful proprietary information that is easily accessible to the companys designers,
superior methods of accessing and utilizing public information, and
competence and tools to more effectively systematize.
1.3 Research in engineering design
Systematic research in engineering design began in Germany during the 1850s. Some of these efforts
(up until 1983) have been compiled in a list by Bjrnemo [BJR83 p. 29]. More recent contributions in
the field of engineering design have been added to Bjrnemos list, demonstrating that research in
engineering design is an active research field that has spread from Germany to most industrialized
nations around the world (Table 0-1).
8
To date, most research in engineering design has focused on design methods. As a result, a number of
design methods are now being taught and practiced in both industry and academia. However, most of
these methods do not provide prescriptive advice where to access information nor how to store such
information once the designer has acquired it. Furthermore, there is a lack of theory on how to reuse
information that has been systematized and stored.
Accordingly, there is a need for generalizable theory on an information infrastructure that will support
various design theories enabling designers in industry to more effectively utilize these design methods.
Such an information infrastructure will also enable academia to identify new research areas that are
complementary to those traditionally pursued in the field of engineering design. Furthermore, theories
on information infrastructures will add valuable content to design courses taught at universities.
1.4 Scope and motivation of this thesis
The purpose of thesis is to initiate the effort to establish the information infrastructure for axiomatic
design [SUH90]. The work is primarily focused on answering the following questions:
What are sources of information that the designer can use during the axiomatic design
process?
How can information obtained during the design process be of benefit during later stages of
the design objects life cycle?
A new information framework was developed to answer the first question, and an extension of the
theory presented by Suh in [SUH90] was developed to answer the second question.
It was shown that this framework applies not only to engineering problems, but also to problems
outside the field of engineering. Furthermore, the framework and theory that was developed can most
likely be fully implemented in a computerized information system. Based on the information
framework, it was demonstrated how other design methods (such as Altshullers theory of inventive
problem solving [ALTS88]) can be used to enhance the designers performance within the framework.
It is expected that the information infrastructure for axiomatic design presented here will facilitate the
following:
strategic work to build sustainable competitive advantages in product development for
industry choosing to work with axiomatic design,
identification of research questions related to the development of software supporting
axiomatic design, and
teaching axiomatic design to both students and practitioners.
9
1.5 Structure of this thesis
The contents of this thesis are as follows:
Section 2 - Scientific method: A framework for conducting scientifically valid research in engineering
design is presented and related to the thesis.
Section 3 - Overview of the types of design theories and a general model of the design process:
Shows the context of this thesis in terms of other theories in design theory and methodology. This
section also contains a model of the design process that provides a useful reference for relating the
various theories and activities to the design process.
Section 4 - Expanded framework of information in design: Introduces an expanded framework of
information in design, based on Suhs concept of domains. This framework is shown to be beneficial in
identifying and generating design solutions, as well as for capturing design information.
Section 5 - Managing information in the framework: A theory on how the information generated in
the framework can be utilized to predict the impact of engineering change orders, search for problems
in the design object and feedback information on the performance of previous designs to the design
engineers.
Section 6 - Relating Theory of Inventive Problem Solving to the framework: Here it is shown how
Altshullers theory is complementary to the framework in generating design information.
Section 7 - Applicability of the framework: This section demonstrates that the framework is not
strictly limited to engineering problems by presenting an example of how the framework has been
successfully used during business planning.
Section 8 - Contributions, conclusions and future work: The contributions of this thesis to academia,
industry and the general transfer of knowledge are presented. Conclusions and proposed future work
(including how the knowledge developed here could be used in developing a thinking design machine)
are also discussed.
10
Table 0-1. Some work in design theory
1
Author Theory or Method Country Appr. year
Altshuller Theory of Inventive Problem Solving [ALTS88] Soviet 1956
Andreasen Chromosome model [ANDR92] Denmark 1992
Bach Die Maschinelemente Germany 1881
Boothroyd and Dewhurst DFM/DFA [BOOT89,BOOT94] USA 1983
Clausing QFD [HAUS88, COHE95]
Total Quality Development [CLAU94]
USA 1988
Cross Engineering Design Methods UK 1989
Dixon and Poli Engineering Design and Design for Manufacturing
[DIXO95]
USA 1995
Erkens Beitrge zu Konstruktionserziehung Germany 1928
Hansen Konstruktionswissenshaft - Grundlagen und
Methoden
Germany 1974
Hubka & WDK School Design Science [HUBK92] Europe 1973
Kesselring Die starke Konstruktion Germany 1942
Kesselring Technische Kompositionslehre Germany 1954
Koller Eine Algoritmisch-physikalisch orienterte
Konstruktionsmethodik
Germany 1973
Leyer Maschinenkonstruktionslehre Germany 1963-71
Matousek Konstruktionslehre des allgemeinen Maschinenbaus Germany 1957
Nieman Machinelemente Germany 1950
Olsson Systematisk Konstruktion Sweden 1976
Pahl and Beitz Engineering Design a Systematic Approach
[PAHL88]
Germany 1977
Pugh Total Design [PUGH91, PUGH96] UK 1985
Redtenbacher Prinzipen der Mechanik und des Maschinenbau Germany 1852
Riedler Maschinenzeichnen Germany 1913
Reuleaux Konstruktionslehre fr den Maschinenbau Germany 1854
Reuleaux Teoretische Kinematik: Grndzge einer Theorie
des Maschinenwesens
Germany 1875
Rodenacker Methodisches Konstruieren Germany 1970
Roth Aufbau und handhabung von
Konstruktionskatalogen
Germany 1974
Sohlenius et al Prodevent (Orderstyrd, kundanpassad
produktframtagning) [SOHL76]
Sweden 1976
Suh Axiomatic Design [SUH90] USA 1978
Taguchi Jikken Keikakuho (Eng. trans. System of
Experimental Design) [PHAD89]
Japan 1977-78
Ullman The Mechanical Design Process USA 1986
Ulrich and Eppinger Product Design and Development [ULRI95] USA 1995
VDI-GKE VDI Guideline 2221: Systematic Approach to the
design of technical systems and products
Germany 1973
Wgerbauer Die Technik des Konstruierens Germany 1943
Yoshikawa General Design Theory Japan 1980
Zwicky The Morphological Method of Analysis and
Construction
USA 1948
11
2. Scientific framework for research in design theory
The purpose of this section is to provide a scientific framework for the research presented in this thesis.
It was found that there is a shortage of literature covering the use and validity of research methods and
scientific frameworks for design theory research. For this reason, it was necessary to investigate and
derive an acceptable, research method or framework for engineering design theory research. This
section summarizes the results of this study and outlines a scientific framework for the design theory
research (Figure 0-3). Complete results of this study can be found in appendix 1.
The material presented in this section does not indicate that this is the only way to do scientifically valid
research in design theory. However, it is concluded to be one scientifically valid way of doing research
in this area.
2.1 Definition of scientific theory and discussion on how theories can be used
2
In this thesis a scientific theory is defined as a theory comprising fundamental knowledge areas in the
form of perceptions and understandings of different entities, and the relations between these
fundamental areas. These perceptions and relations are combined by the theorist to produce special
consequences (which can be, but are not necessarily, predictions of observations) [FLL93 p. 75-78].
Fundamental knowledge areas (such as mathematical expressions, categorizations of phenomena or
objects, models, etc.) are more abstract than observations of real-world data. Such knowledge, and
relations between knowledge constitute a theoretical system. A theoretical system may be one of two
types (depending on the way in which the fundamental knowledge areas are treated): They may be
treated as either
axioms, or
hypotheses.
Fundamental knowledge that are not or cannot be tested, yet are generally accepted as true, are treated
as axioms. If the fundamental knowledge areas are being tested, they are treated as hypotheses. In this
case, the consequences which result are used to support or to refute the hypotheses. So, the distinction
between these two theoretical systems is that the axiomatic theory is concerned with derived knowledge;
while the hypothetico-deductive theory is concerned with testing the validity of hypotheses.
1
This table is neither exhaustive nor categorized, it merely provides the reader with examples of
research in engineering design.
2
This section is based on material presented in [FLL93].
12
In either case, the fundamental knowledge areas are combined with information about particular real-
world situations to derive logical consequences. The consequences themselves are not the hypotheses.
Rather, they can then be used to fulfill three purposes:
to test,
to predict, or
to explain.
Hypotheses lead to consequences which are used either to test or to make predictions (which could be
tested). Axioms lead to consequences which are used to explain or to make predictions (which are
subsequently not tested). The use of these logically derived consequences to test or to predict is termed
applied research. The use of the consequences to explain observed events or facts is pure (or basic)
research. The purpose of pure research is to provide true knowledge about the world that indirectly
will contribute to mans ability to control and change this world [FLL93 p. 185].
2.2 Research process model for design theory
In design theory, the role of a design researcher is to develop new design theories and to verify these. In
Figure 0-3, a process for such activities is presented.
The purpose of the model in Figure 0-3 is to make explicit one scientifically acceptable research
process in design theory research that will serve as the scientific framework for this thesis.
The input to the process is a research question and the output is a theory that can be tested or used for
explanation and prediction. The main phases of the research process are data gathering, theory
development, and theory validation. Each phase comprises one or more activities that are performed to
generate knowledge. In some research projects, an individual researcher works through all phases. In
other projects, the work is initiated by one researcher, and the results of this phase are passed to a
different researcher who then continues the research by working in a subsequent phase.
Each phase and its activities are described in more detail in appendix 1.
Data Gathering
Hypotheses
Generation
Experience
and results
Hypothesis
selection Criteria
(strength, safety,
simplicity)
Researchers
Subconscious and
conscious thought
processes
Hypotheses
Acceptable research
standards
Research
question
Generate or find
information
Testing
Conclu-
sions
Researcher,
logic
Laboratory
equipment or
reality
Observations,
Available facts
Establishing
relationships
between the
inputs
(Theory
generation)
Fundamental
knowledge in
several different
areas
Theory
logic
Deducing
consequences
from the
theory
researcher
researcher
Special
Consequences
Data Gathering Theory Development Theory Validation
logic
Knowledge
about specific
case
Explanations and
Predictions of events
or phenomena
Figure 0-3. Model of the research process for design theory
14
3. Context of this thesis
3
The purpose of section 3 is to provide the reader with the industrial and academic contexts for this
thesis. The industrial context is the industrial design (product development) process, while the academic
context is the existing state of the art in contemporary design theory research.
3.1 Industrial context: The design process
4
The design process transforms the a customers perception of a problem into the design object, i.e., any
problem solution that the customer finds satisfactory. Designers are able to transform customer
perception to design object through their use of design tools/methods, knowledge of discipline-specific
information, and a given set of available resources.
Based on discussions with industry, an acceptable design process model that would serve as a context
for this thesis, would fulfill the following requirements:
1. Emphasize the importance of making good front-end design decisions.
2. Easy to articulate and understand the results of the approach. Communication of
requirements and design solutions among designers, customers, and other stakeholders is
critical to the success of the projects.
3. Able to segregate functional and physical space (requirements and solutions). The process
must begin by allowing designers to clearly define the requirements in the functional space.
4. Support abstraction, i.e., it is able to handle high-level requirement issues with end users
and address realization details with implementors.
5. Helps designers manage complexity by encapsulating unnecessary detail.
6. Accommodates existing systems, organizations and facilities. Most of the product
development projects rarely start from scratch with a complete clean sheet of paper.
Ignoring these existing factors will lead to unrealistic design processes.
7. Robust, i.e., able to handle a wide variety of customer needs and problem spaces.
8. Analysis and synthesis processes that are easy and intuitive to users.
9. Support iteration of activities, since many activities are either repeated or re-done during a
design project.
3
The research for section 3 was conducted in close cooperation with Derrick Tate of the department of
Mechanical Engineering at MIT.
4
Much of the material in this section is accepted for publication in [TATE96].
15
Based on these requirements, a general design process model needed to be flexible enough to cover all
instances of the design process, yet specific enough to address specific design activities. To find such a
model, a study of existing design process models
5
was conducted. Some models were very broad, and
this not helpful in specific design activities. Other models were so restrictive that they were only valid
in special cases, if at all.
This apparent contradiction was solved by segmenting the design process into a collection of generic
activities and making the relationships between these activities explicit (an activity is defined as the
transformation of inputs to outputs). From such a collection of activities, the designer can structure a
unique design process. Using this idea, it is possible to describe the activities with sufficient specificity
to be useful (for example, to support the designer in selecting design tools and methods for each
activity, to clearly identify the decision points, and to create a good information infrastructure for the
process). However, this collection of design activities is still general enough to assemble any design
process.
The design activity collection presented in this section came about as an abstraction and generalization
of several industrial design processes that were studied during this thesis research and in cooperation
with Tate [TATE97].
3.1.1 Introduction to the design process roadmap
The design activity collection consists of several distinct activities with clear starting and end points.
These activities can be conducted in several different sequences. Each design project will have its own
unique sequence, depending on its specific status, scope, and goals. Figure 0-1, shows all activities that
are currently in the design activity collection and the possible links between them. From this roadmap it
is possible to assemble a project specific process that can begin with any activity. The arrows entering
an activity show what information is required to perform the activity. Arrows that do not originate from
or end in one of the activities in this roadmap interface with activities other than those in the design
process, e.g., customer surveys.
5
The models reviewed were found in [HUBK92, PAHL88, PUGH91, WILS80].
16
Project control
& decomposition
Existing design,
CNs, Cs
Design object
analysis
Problem
formulation
FRs DPs
(coupled DM
or unmet Cs)
Concept
generation &
selection
Decoupling
DP values
FRs DPs
(uncoupled or
decoupled DM
& met Cs)
Trade-off
FRs Cs
FRs DPs
(uncoupled or
decoupled DM
& met Cs)
FRs DPs
(uncoupled or
decoupled DM
& met Cs)
FRs DPs
(coupled DM
or unmet Cs)
DP values
Implementation
Customer Needs,
Con strain ts,
existing design
Unsatisfactory result
of trade-off
Plan for new design,
higher-level DPs,
CNs, Cs,
Decomposition or
failed conceptualization
Solution to decompose or
report of failed decoupling
Results
Axiomatic Design terminology [29, 30]
Cs: Constraints
FRs: Functional Requirements
DPs: Design Parameters
DM: Design Matrix
Figure 0-1. The design process roadmap (from [TATE96]).
3.1.2 Putting together a project specific design process from the activities in the roadmap
A design process (project) generally begins with the project control and decomposition activity (at the
left side of Figure 0-1). A design process that yields a design object ends with either an implementation
or a trade-off activity (at the right of Figure 0-1), where a design object is described in terms of its
physical parameters and their implementation (drawings, models, etc.), and manufacturing processes.
Design processes that fail to yield a design object end in the project control and decomposition activity,
when a decision to terminate the project is made.
The specific path which is followed is the responsibility of the design team. Each time an activity is
completed, a decision is made about how to proceed with the design process. The preferred outcome of
the project is to successfully complete implementation as quickly as possible. By understanding the
desired outcomes of the various activities as described in Figure 0-1 (FRs, DPs, uncoupled/decoupled
DM, etc.) and comparing these with the actual outcomes, decisions about how to proceed with the
design process can be based firmly on concepts of independence and minimum information content
(maximum probability of success) [SUH90]. Often, the choice to perform one activity over another will
allow the design team to reach its goal in fewer steps. For example, the choice to decouple a design
object, as opposed to conducting the trade-off activity, will (if successfully done) allow the designer to
quickly implement a design satisfying the independence axiom. Trade-off, on the other hand, is a much
more involved, iterative process which may not always yield a design object that satisfies all functional
requirements.
17
3.1.3 Description of the generic activities in the design activity collection
This section contains a description of each activity shown in the design process roadmap (Figure 0-1).
Each description explains the activitys purpose, inputs, outputs, and typical questions that arise during
the activity.
3.1.3.1 Project control and decomposition
Project control and decomposition places controls on the design project and establishes its scope.
During a design project, this activity is revisited multiple times. It will be performed when it is
necessary either to plan and decompose for work at a more detailed level of the design, or to plan
activities in response to an inability to solve a previous problem.
During project control, possible courses of actions are evaluated and decided upon. The scope of the
design project and other project management issues such as a budget, milestones, etc. are also
established. Typical questions that arise during this activity include the following:
What resources are available to solve the problem?
Will a project be a new design or an adaptation of an existing design?
How can the project be decomposed into sub-problems?
The inputs of project control and decomposition can vary. At the beginning of a clean slate project,
the inputs are customer needs and constraints. At the start of a re-design, or evolutionary design
project, the inputs will include customer needs, constraints, and additionally a representation of the
existing design object. When this activity occurs as a part of an on-going design project, the inputs will
be descriptions of the design object at the current level of abstraction and a description of any problems
which have occurred.
Outputs of the project control and decomposition activity are project goals, constraints, and plans
describing how to put subsequent activities together in a sequence.
3.1.3.2 Design object analysis
The design object analysis activity presupposes an existing design object requiring an analysis of
functionality or feasibility. The analysis may follow one or more specific approaches, such as
axiomatic design [SUH90], functional analysis (value engineering) [MILE72], or Theory of Inventive
Problem Solving [ALTS88]. Design object analysis is often a central activity in feasibility studies. The
results of this activity can be fed back into the project control and decomposition activity to allow a
detailed planning and control of the design process. Design object analysis can also be conducted to
18
identify areas for improvement in existing designs. Typical questions that arise during this activity
include the following:
What are the FRs of this design?
What improvements can be made to the design object?
Does the design object meet all its constraints?
How do I reduce cost, number of parts and assembly time?
Are there couplings in the design?
Are off-the-shelf, or other existing, solutions acceptable for this project?
Is a new solution needed for the problem?
In design object analysis, there are two inputs: an understanding of the customer needs, and an existing
design object that must be analyzed. The output of this activity is a description of the design object in
terms of its functions, constraints, physical implementation, and functional dependencies (or
independence).
3.1.3.3 Problem formulation
Problem formulation is performed when the designers objective is to create a new or innovative
solution to a design problem. This activity requires designers to have the flexibility to select a new or
innovative solution. A new solution may be required because a previous design problem that was never
solved. Or, a new solution could be sought because customer needs have changed and an analysis of the
existing solution has shown that the current solution was unacceptable to the customer. Typical
questions of problem formulation include the following:
What does the customer want?
What is the customer willing to pay for?
What is the ideal function?
What are the constraints (e.g., those imposed by the company) on an acceptable solution?
Problem formulation involves translating information from the customerin terms of customer needs
and constraintsinto functional description of a design object which will meet these needs. During this
activity, the functional description will be produced (for problems not at the highest level of the
hierarchy). Accordingly, it is necessary to consider decisions made at the previous level about the
physical embodiment of the design object.
19
Inputs for problem formulation are customer needs and constraints pertaining to the current level of the
design hierarchy. In addition, if the design is not at the highest level of abstraction, then decisions about
the physical embodiment of the design (from previous levels of the hierarchy) are required. The outputs
of problem formulation are a set of FRs and constraints which then serve as input for the
conceptualization activity.
3.1.3.4 Decoupling (conflict resolution)
Decoupling is one possible activity that will follow the problem analysis activity. This will occur when
problem analysis shows that the design is coupled or fails to meet its constraints. Otherwise, the trade-
off activity will follow. The aim of the decoupling activity is to achieve an uncoupled or decoupled
design that meets all constraints. This is achieved by applying one or more problem solving strategies
specific to conceptual design (e.g., Suhs theorems, Altshullers principles, Su-field analysis, etc.). The
preferred result of this activity is a design that is acceptable for implementation or further
decomposition. However, the decoupling activity does not always lead to a solution suitable for
implementation. In such instances the result must be used for a new project control and decomposition
activity and an alternative course should be selected.
Very often, the decoupling activity is performed when the designer must improve the design objects
performance, yet is not free to make major changes to the design object. Typical questions that this
activity is intended to answer include:
How can this design object be modified to resolve the problems we are experiencing?
How can the design parameters be better integrated to meet the constraints?
Inputs to the decoupling activity are descriptions of a design object or concept that fail to satisfy the
independence axiom [SUH90], or a design object that fails to satisfy the constraints. In the first case,
the design object does not perform its functions satisfactorily. In the second case, the design object may
perform its functions satisfactorily, but the physical solution is unacceptable for some other reason
(e.g., it is too big or too heavy, it does not meet some legal requirement, etc.).
The output of this activity is normally a description of a design object (or concept), that is functionally
uncoupled or decoupled and meets all its constraints. In cases when a successful result is not produced,
the output will be a coupled design or a design that does not meet its constraints. A report will also be
produced explaining why the activity failed.
20
3.1.3.5 Concept Generation and Selection
The concept generation and selection activity always follows the problem formulation activity. The
objectives of concept generation and selection are:
to develop concepts that satisfy the specifications derived in problem formulation, and
to determine which of these concepts to implement.
Hence, this activity is supported by concept generating (synthesis) tools and analysis tools. Suh
illustrates this activity as a feedback control loop (see figure 0-2), and concludes that analysis is an
important part of concept generation because if we cannot analyze a design solution, then we cannot
rapidly generate the best design since we cannot distinguish good design from bad design. In the
absence of a criterion for selecting a good design, we cannot make good design decisions. Concept
generation tools include brainstorming, databases, etc., while analysis tools include design axioms,
group decision making, and other design rules.
G
H
X Y
Y/X=G/(1+GH)
G = Synthesis capability
H = Analysis Ability
Figure 0-2 Feedback control loop depicting the synthesis and analysis during the concept
generation and selection activity (from [WILS79])
Typical questions addressed during this activity include the following:
Given a set of functions, what are possible solutions? How can these functions be realized?
How many ways can I solve this problem? How creative can I be?
Will this concept work? How do I choose between concepts?
How can these design parameters be physically integrated while achieving their functions?
Inputs to the conceptualization activity are a set of functional requirements (FRs) and constraints.
Outputs are a set of FRs, constraints, design parameters (DPs) (including physical implementation in
the form of sketches or models), and design matrices (DMs) that show the functional dependencies in
the design object.
3.1.3.6 Trade-off
The objective of the trade-off activity is to get the best possible performance out of a design object that
does not satisfy the independence axiom (see implementation activity for design objects that do satisfy
this). A typical situation in which a trade-off activity is conducted is when there is an existing design
21
that has been found to have some inherent problems, yet it is impossible to anything but small changes
to some DP values. Accordingly, this activity entails setting the design parameters to values that best
achieve the desired functionality given the circumstances. Typical questions answered during this
activity include the following:
How should the values of these parameters be set to provide the best possible functionality?
What is the best functionality I can get with this design object?
Inputs to this activity are functional requirements, constraints, design parameters, and design matrices.
Outputs from this activity are functional requirements, constraints, design matrices, and design
parameters (set at their optimal values).
3.1.3.7 Implementation
Implementation is the final activity for any design project that has yielded a design object that satisfies
the independence axiom and accordingly should achieve all functional requirements. A typical question
for this activity is:
How should the values of these parameters be set to provide optimal functionality?
Inputs to the implementation activity are functional requirements, constraints, design parameters, and
design matrices.
Outputs from this activity are functional requirements, constraints, design matrices, and design
parameters that are tuned to provide optimal functionality for this design object.
22
3.2 Academic context: Contemporary design theory research
6
Several contemporary European and US research programs were reviewed to understand
the motivation behind conducting research in design theory,
the scope of such research, and
how these methods were developed.
The research programs examined were those put forth by Altshuller [ALTS88], Clausing [CLAU94],
Suh [SUH90], and the WDK school (Hubka, Eder and Andreasen) [HUBK92].
These programs represent a cross-section of contemporary research in design. They are widely taught
at universities, actively researched around the world and writings on these programs are readily
available in English.
3.2.1 Motivation and Scope
7
of selected design methods
3.2.1.1 Altshullers motivation and scope
Altshuller recognized the need for a scientific approach to invention after listening to scientists and
inventors speak of design as sudden enlightenment. They complained that it was impossible to control
the creative process much less understand what it is and how it comes about. Such discussions made
Altshuller question why everything but creativity should be open to scrutiny and why this process,
unlike all others, should not be subject to control. According to Altshuller, failure to control the creative
process results in many inventions coming too late, frequent mistakes, and inventors dreaming up
unrealistic solutions. [ALTS88 p. ix-x].
Accordingly, Altshuller strived to make creativity a controlled process. Here, creativity refers to the
activity of generating new designs, not selecting from among existing designs. Creativity is required
when formulating problem statements identifying and analyzing key conflict areas, and applying
solution guidelines to the specific situation at hand.
Altshullers program is intended to enhance the engineers thinking during innovative work thereby
contributing to the overall design process. However, the scope of his method of creative or innovative
thinking is general, despite being mainly aimed at engineers [ALTS88 p. xi]. Applying this method in
6
This section is a modified subset of [TATE95].
23
the context of a design project should provide benefits in the form of a reduced number of iterations and
better solutions (based on Altshullers definition of what is a good product). Furthermore, Altshuller
describes how solve technical conflicts in his algorithm for solving inventive problems.
3.2.1.2 Clausings motivation and scope
The driving force behind Clausings work has been a very broad oneto improve industrial
performance. It appears as if Clausings approach is to use results of different product design research
to improve the overall product development process. He incorporates the work of other researchers into
his framework, which he calls total quality development".
In Clausings own words, Total Quality Development is the modern way of developing new products
that will be competitive in the global economy. It combines the best engineering, the best management,
the best strategy, and especially, the best teamwork. The resulting improvements are greatly reduced
development time, a reduction in all costs, higher quality, and increased product variety. Combined,
these improvements greatly increase customer satisfaction [CLAU94 p. 3].
3.2.1.3 Suhs motivation and scope
Suhs primary motivation for developing axiomatic design is education; i.e., educating designers to
make good design decisions. Suh seeks to establish an academic discipline for design and
manufacturing [SUH78a, SUH78b, SUH90 p. 21-22], i.e., fundamental, correct principles and methods
that will guide decision making in design. According to Suh, without such discipline the ad hoc nature
of design can not be improved [SUH90 p. 5]. To be effective the student must be taught to see the big
picture and [be taught] the ability to conceptualize a solution, as well as how to optimize an existing
product or process [SUH90 p. 22].
Suhs view of the scope of design may be summarized by the following: Design, as the epitome of the
goal of engineering, facilitates the creation of new products, processes, software, systems, and
organizations through which engineering contributes to society by satisfying its needs and aspirations
[SUH90 p. 5]. In contrast to Clausing, Suh does not describe how to connect design activities to the
companys general activities; rather, Suhs theories and methods are focused on decision making in the
design process.
In his book [SUH90], Suh primarily considers designs from three fields: manufacturing process design,
product design, and organizational design. Although the bulk of his personal experience in applying
7
The definition of scope which we are using is the following: extent covered [MERR89 p. 650].
24
axiomatic design is limited to these three areas, he recognizes the potential for its application in other
fields. Industrial use and acceptance of axiomatic design has been growing in a variety of fields. Recent
applications of the theory have included product design, manufacturing process design, the design of
software configuration control systems, organizational design, and corporate planning.
3.2.1.4 The Workshop Design Konstruktion (WDK) Schools motivation and scope
Hubka and Eder focus on systematic design methods, and independent auditing in order to improve
the efficiency of the designer [HUBK92 p. iii]. According to these researchers, design become more
efficient by scientifically reducing or eliminating waste of labor, time, or materials [HUBK92 p. 45].
Looking at the beginnings of the design process, Hubka and Eder state that design can be considered
broad enough to include defining needs and product planning as well as the narrow view of designing
[HUBK92 p. 49]. Clearly though, they do not feel that this is necessarily within the scope of design.
The design sequence begins when an engineering design team receives a set of requirements (either from
a customer or another sponsor) [HUBK92 p. 74]. The end point of the design process is a description
of a technical system, specifically a full and complete description of an optimal product (i.e. a
technical system) [HUBK92 p. 46].
3.2.1.5 Discussion on scope of design methods
In general, the European schools (such as WDK) of design tend to separate out and concentrate on a
portion of the total product development activity. This is also explained by Clausing in the following:
The undergraduate engineering curriculum typicallyincludes one or two design courses. These
concentrate on creative concepts and feasibility, the assurance of a first-order compatibility with the
laws of nature. Let us call this partial design [CLAU94 p. 5]. However, this term partial design is
unsatisfactory because it is misleading. The activity is not partial in the sense that it is incomplete;
rather it covers only a small, well-defined fraction of design. It is proposed to use the Germanic word
konstruktion for this narrow, detailed activity [TATE95].
3.2.2 Approaches to develop new design methods
This section describes, in more detail, how the four theories and methods discussed above were
developed.
3.2.2.1 Altshullers approach
Altshuller identified a need for new methods to manage the creative process. Such methods would be
capable of radically reducing the number of empty trials [ALTS88 p. 3] during a trial-and-error
approach. Furthermore, the creative process needed to be recognized to permit the effective application
25
of new methods. All this required a scientifically based theory (for the solution of inventive tasks)
capable of being implemented in practice [ALTS88 p. 3].
In order to develop such a theory, three requirements were established. If these requirements were
satisfied, it would be possible to guarantee a solution to any technical problem. The requirements were
1) information about the whole of physics, 2) tables linking the type of problem to the respective
physical effects, and 3) control of psychological factors that inhibit the thinking of the inventor
[ALTS88 p. 35].
Altshuller begun work on an algorithm
8
for the solution of inventive problems [ALTS88 p. 36] in 1946.
He studied the experience of inventive creativity from a fundamental point of view and brought out the
characteristic features of good solutions (i.e., those characteristics that distinguished them from bad
solutions). As a result of these studies, Altshuller discovered that the solution of inventive problems
turned out to be good if it overcame the technical contradiction
9
contained in the problem presented, and
bad if the technical contradiction was not revealed and eliminated [ALTS88 p. 40].
3.2.2.2 Clausings approach
As was described in section 0, the motivation behind this approach is pragmatic; if a technique works
(that is, improves the design process or design object), it is more important to put it into use than to
understand exactly why it works. Thus this school consists almost entirely of methods, not theories.
In developing his approach to design, Clausing has been using two primary sources: personal
experience from industry and benchmarking the best practices around the world. He then integrates the
best components into a holistic approach to design [CLAU94 p. xix].
This approach of developing a design method is different in that it is totally goal oriented (to improve
industrial performance). Clausing doesnt make any claims to be scientific in his approach, but
implicitly claims that it works better than any other approach (that he is aware of) in an industrial
setting. Clausings contributions are mainly: 1) pragmatically analyzing the different design methods
and placing them in the context of the total development process in a corporation, and 2) integrating the
best design theories and methods with management and strategy to form a cohesive approach to design.
8
Altshuller defines an algorithm as any sufficiently clear program of action [ALTS88 p. 36].
9
A technical contradiction exists if [when using] certain methods [to improve] one part (or one
parameter) of a technical system, it is inadmissible for an other part (or other parameter) to deteriorate
in the process [ALTS88 p. 28].
26
By its evolutionary nature, this program will continually change and improve as Clausing continues to
search for new components that complement or improve his approach.
3.2.2.3 Suhs approach
Suh started the development of his program by asking the following: Given a set of functional
requirements for a given product, are there generally applicable axioms which yield correct decisions in
each step of manufacturing (i.e., starting from the design stage to the final assembly and inspection
stages) so as to devise an optimal manufacturing system? [SUH78a]
A heuristic approach was used to develop the axioms. This approach involved positing an initial set of
axioms that were subject to trial and evaluation in manufacturing case studies. This evaluation would
then be used in order to expand, redefine, and refine the original set of axioms, until the process
converged on a comprehensive set of axioms [SUH90]. Based on such a set of axioms, many specific
methods for analysis and problem solving could be developed [SUH90 p. 171]. Out of this exercise
evolved twelve hypothetical axioms which later have been reduced into two and a set of corollaries and
theorems [SUH90 p. 20].
Suh started his search for design axioms by observing that that there are both good and bad design
solutions. Accordingly, there are features or attributes that distinguish each type of solution. The first
axiom defines an acceptable design as one where the design parameters and functional requirements are
related such that a specific design parameter can be adjusted to satisfy its corresponding requirement
without affecting other functional requirements. The second axiom states that the best design (of several
proposed) is the one that has the lowest information content (i.e., the highest probability of success)
[SUH90 p. 47-8].
3.2.2.4 The WDK Schools approach
In contrast with the other programs discussed here, this school, as presented by Hubka and Eder,
primary focuses on developing descriptive models both for technical systems and the design process
[HUBK92 p. 71-102]. When such descriptive theories are established, it would be desirable if the
[prescriptive] statements (of advice and compulsion) could be derived from the descriptive [theories]
[HUBK92 p. 116].
Based on this general procedural model of the design process, a procedural plan for a specific situation
could be derived and adapted from the ideal model [HUBK92 p. 59].
27
Andreasen, has further evolved Hubka and Eders work, based on his belief that designers are, in
general, unable to describe large parts of their work, because it takes place in unnamed patterns of
ideas, rapid experimental patterns of association, and partly sub-consciously. Therefore, Andreasen
determined that the task of design research must be to create the conceptual framework and the
patterns of thought. In order to support design of mechanical systems, Andreasen concludes that the
design theory
10
must be based on a theory of the design process and a theory of mechanical systems
[ANDR92a p. 1-2]. He also believes that if we are to make progress in design science, we have to
create a theoretical apparatus so that we can discuss design and attempt to derive laws, models and
methods [ANDR92a p. 10].
3.2.2.5 Discussion on approaches to develop design methods
Both Altshuller and Suh established a set of principles or axioms from which a variety of methods or
algorithms to solve specific problems can be developed. Both also attempt to define what is a good
solution or design, and interestingly, they arrive at very similar definitions independently of one
another.
Based on their principles or axioms, Altshuller and Suh have developed different but complementary
approaches to arrive at a good design. Altshuller developed a system of methods to separate
contradictory properties through clever synthesis and integration of parameters, while Suh developed a
metric and analysis rule that warns the designer if he or she is creating a bad design.
The approach that Hubka and Eder followed to develop the WDK school is based on observation and
systematizing what designers already do, complemented with a theory for modeling technical systems.
This, appears to yield an after-the-fact approach to design. As such, it will only provide a scientific
description of what the designers do without stating whether this is the right thing to do. The models of
the technical system will be used by the designer to describe how the technical system will function;
however these models will not provide any fundamental reason why it will or will not work.
Furthermore, the WDK school of Hubka and Eder does not appear to directly define what constitutes a
good designsomething that is central to both Altshuller and Suh. Instead, Hubka and Eder refer to the
ISO 9000 definition of quality, which is the totality of those properties and characteristics (of a
product or an activity) that relate to its suitability to fulfill the stated requirements [HUBK92 p. 21].
The quality is evaluated against a set of criteria, and a composite quality number (representing e.g.,
technical and economical value) is calculated. Hubka and Eder recognize that this method has
10
Theory here is defined as a system of concepts, rules, axioms and models.
28
problems, but implicitly defend it in that all methods have their disadvantages. In comparison to Suh
and Altshuller it appears as a weakness of this school to lack a clear definition of what constitutes a
good design.
Clausing provides an important technology transfer link from the academic research community to the
community of users in industry. Clausings unique approach applies a concept selection method, that
combines useful features from several different methods to create a holistic method. Then
benchmarking is used to ensure that the new method is superior indeed.
Clausings way of evolving a design approach requires a number of properties: a broad network of
contacts that can provide information on new developments; more focus on pragmatic value than
scientific value; and an open mind which is not committed to any individual component of the approach,
but rather is willing to replace components with new ones that are better. Perhaps Clausing is the only
researcher in this field who has found a way to satisfy the following challenge by Ullman: If only we
could use a sound design [method] to approach the problem of designing a theory... [ULLM91 p.
801]. Even if Clausings approach by some would be called unscientific, it is nevertheless effective.
The differences in approaching the development of knowledge in this field is captured by the following
statement where Sohlenius expands on thoughts presented by Von Karman: The engineer creates what
has never been, the scientist analyses what is and the engineering scientist analyses what is, imagines
what should be, creates what has never been, and analyses the results of the creation [SOHL90].
3.2.2.6 Approach followed to develop the method presented in this thesis
The research presented in this thesis is prescriptive in its nature and is primarily based on Suhs work,
and therefore follows a similar approach to that of Suh. This is especially evident in sections 5 and 6
where Suhs and Altshullers axioms and principles are used to derive further theorems and theories.
However, some descriptive research is done in section 4 to establish how designers work when
following the axiomatic design method. The results of this research was used to create a prescriptive
information framework presented in section 4. Section 7 also provides some descriptive research to
demonstrate that the prescriptive models derived in section 4 were applicable outside the field of
engineering.
29
4. Expanded Framework of Information in Design
4.1 Research question
What are sources of design information in the context of axiomatic design?
4.2 Research method
This research has followed the process outlined in chapter 2. Information gathering was conducted
through literature studies and studies of how designers work when following the axiomatic design
method. This information was then used as a basis for forming hypotheses that would answer the
research question. Finally, the hypotheses were tested in an industrial case study.
4.3 Literature studies
The scope of this literature study has been to answer the following question: What outside sources of
innovation are available to the designer? In the context of axiomatic design, innovation is interpreted as
generating the design information necessary to map from the functional to the physical domain or from
the physical to the process domain. Identifying these sources of innovation will facilitate the
development of an extended framework around the domain model proposed by Suh [SUH90 p. 128].
Such a framework will enable the designer to effectively utilize available sources of design information
during the axiomatic design process.
Utterback has found that, often, the most radical innovations are made by industry outsiders rather
than the leading companies in a particular industry [UTTE94 p. 160]. The reason for this phenomenon
has to do with risk and commitment. Outsiders have everything to gain and nothing to loose with the
introduction of radical innovations. In contrast, industry insiders are often committed to an existing
technology due to large investments that cannot be easily divested (for a detailed discussion on
commitment and its implications on corporate strategy, see Ghemawat [GHEM92]).
Von Hippel presents a more detailed categorization of the functional sources of innovation: users,
manufacturers, and suppliers [HIPP88 p. 4]. Users are defined as those who benefit from using an
innovation. Manufacturers are those who benefit from manufacturing the innovation. Finally, suppliers
are those who benefit from supplying the materials or components necessary to build the invention
[HIPP88 p. 3]. Von Hippel concludes that innovation is likeliest to occur in the category where the
expected rent from the innovation is highest [HIPP88 p. 117]. In order to acquire the desired design
information von Hippel proposes several mechanisms:
30
Field service organizations can gather information about user modifications on the products they
service.
Sales departments can obtain information on new product needs, ideas and prototype solutions.
Marketing research groups can seek possible sources of new product solution data as well as need
data.
These mechanisms mainly address the information transfer from users to manufacturers. A novel way
to effectively transfer design information between suppliers and manufacturers is the JIT II system (Just
in Time II) proposed and implemented by Lance Dixon [DIXO94]. With JIT II a manufacturer provides
its suppliers employees with almost completely free access to the manufacturers plant in order to
facilitate a high level of interaction with the manufacturers engineers. Dixon points out that there are
risks associated with allowing outside people to influence the design [DIXO94 p. 69]. However, with
careful management, these risks can be outweighed by the benefits. According to Dixon there are clear
benefits with having consistent access to in-house supplier expertise-even when the supplier does not
ultimately win a program. JIT II may also spawn innovation as design engineers press suppliers for
solutions. Managed with care, it is possible for a JIT II customer to maintain inflow of leading-edge
technologies to [its design groups] [DIXO94 p. 70-71].
4.4 Empirical Study of how engineering information is synthesized in a design project
This section describes a design project from the medical industry where using Suhs method of
axiomatic design [SUH90] was used to design a new medical device. A total of four designers from the
company were involved. The researchers role was to facilitate the application of axiomatic design
during the project and learn how the designers worked when trying to apply axiomatic design.
The task was to design a manufacturing system that would insert a small rubber disk into a plastic tube
(sealed at one end) as illustrated in figure 0-1. Upon completion of the assembly sequence, the rubber
disk should touch the membrane seal without trapping any air. Besides cost constraints, designers were
not allowed to puncture the membrane seal or tube during the manufacturing process.
Before
assembly
a ssem b
After
assembly
Figure 0-1 Manufacturing Requirements
31
The functional requirements (FRs) for this product were established as follows:
FR1: position the disk, and
FR2: remove all air trapped between the disk and the seal.
Next, the design group had to identify design parameters (DPs) to satisfy these FRs. For this study, the
researcher was most interested in identifying the types of information that designers generate along with
the possible sources of such information.
In generating DP1, it was reasoned that some force was needed to position the disk. The design group
proceeded to generate a list of all potential forces. This list included mechanical force,
pressure/vacuum, centrifugal force, gravitation, magnetic force, and electric force. It appeared as if the
engineers followed two approaches in generating this list of DPs: brainstorming and structured
reasoning. During brainstorming, the designers listed any force that came to mind, without criticizing
any of the suggestions. Using structured reasoning, the engineers carefully analyzed what was known
about the design and its environment, and then tried to identify potential sources of force that existed in
this environment. When analyzing the feasibility of each DP (primarily comparing them to constraints),
it became clear that the DPs generated from the environment were likely to have both lower cost and
lower information content (as per axiom 2 [SUH90]). In contrast, when DPs are generated from outside
the existing environment, it is often necessary to create a new [technical] system, which may have a
high information content. Following this logic, gravitation was chosen for DP1 as it is always present
(zero information content) and it is free.
A number of ways to remove trapped air were also generated for DP2. These included making a small
hole to release air through the disk, once inserted, temporarily changing the disks geometry to enable
air to escape, and positioning a small straw between the disk and inside of the tube, through which the
trapped air could escape (once the disk is positioned, the straw would be removed). Evaluating these
alternatives, it was estimated that the straw was the most complex solution, and that the hole violated
one of the constraints. Thus, temporary changing the geometry (size) of the disk was chosen as DP2.
Again, a DP from the existing environment (the disk) was chosen rather than introducing a new device
to perform the function.
At this point, the design must be analyzed (Equation 0-1) to ensure it satisfies the independence
axiom:
32
Equation 0-1
FR1 Position Disk
FR2 Remove Air
X X
O X
DP1 Gravitation
DP2 Change disk geometry
'
1
]
1
'
This is a decoupled design. It shows that the manufacturing system must first change DP2 (disk
geometry), and then implement DP1 (gravitation) in order to successfully achieve FR1 (positioning of
the disk). Since it is possible to satisfy the independence axiom with this design, the design group
proceeded to identify the process variables.
g
g
Figure 0-2 Gravitation components depend on orientation of tube
The following options for PV1 (how to generate gravity) were identified: locate on different planet,
locate in orbit, use centrifuge to enhance or reduce gravitation, change orientation of tube to reduce the
gravitation component toward the bottom of the tube. Only one of the proposed concepts was deemed
as feasible by the design group: PV1 orientation of tube (Figure 0-2). The tube has to be oriented
vertically to maximize the effect of gravitation in positioning the disk.
Suggestions for PV2 (how to temporarily alter the disks geometry) included: using a tool to hold the
disk from the sides, hitting the disk, Shrinking the disk by freezing it and the re-heating it to resume the
original shape. Of these proposed PVs, only room temperature existed in the environment, but several
different ways of freezing the disk could be identified in the customer domain, i.e., at the company site.
The group decided to select temperature as PV2. This meant introducing a freezing system (e.g., a
freezer or liquid nitrogen) to freeze (and shrink) the disks, drop them into the tubes, and then utilize the
existing room temperature to warm the disks up so that they would retake their original shape.
33
Finally, the proposed DP-PV mapping was analyzed (Equation 0-2) to ensure the independence axiom
was satisfied:
Equation 0-2
DP1 Gravitation
DP2 Change disk geometry
X O
O X
PV1 Orientation of tube
PV2 Change disk temperature
'
1
]
1
'
'
1
]
1
'
12
Due to the nature of this product, specific numbers as well as some specific requirements and
constraints that were part of the original design report can not be included. However, this does not
make any difference to the conclusions presented here.
40
Equation 0-4
FR21 Provide force to launch device
FR22 Send device in desired direction
FR23 Convey force to entire device
X O O
O X O
O O X
DP21 Explosive
DP22 Barrel
DP23 Chassis
'
1
]
1
1
1
'
These equations show that the independence axiom is satisfied, i.e., the independence of the FRs is
maintained and the design process continues. However, design of the launcher (including the explosive,
barrel and chassis) is already complete and will not be further decomposed here.
At this level, a number of constraints are introduced that will apply to all DP choices later in the design
process.
Constraints
C
1
= Safety
C
2
= Weight
C
3
= Position of the center of gravity
C
4
= Outside measures (geometry) has to fit within chassis
C
5
= Environmental endurance
4.7.3.3 Decomposing the initiator (FR1)
FR1 must now be decomposed into the lower level FRs. The decomposition must take into
consideration that an electrical system was chosen as DP1.
FR11 Provide electricity
FR12 Activate AC1
FR13 Activate AC2
FR14 Activate AC3
FR15 Send Ignition Signal
'
'
1
]
1
1
1
1
1
1
'
'
1
]
1
'
A concept realizing this solution is shown in Figure 0-6. Gas pressure enters the rear end of the depth
charge. The gas is led to a chamber where an impact piston is forced to hit one end of a battery. This
impact should suffice to break a glass ampoule containing an electrolyte. When the electrolyte diffuses,
the battery becomes active.
FR12 Generate Arming condition 1
The design equations resulting from decomposing FR12 (generate AC1) are:
Equation 0-7
FR121 Sense launch
FR122 Activate circuit after leaving barrel
X O
X X
DP121 Rod sensing barrel
DP122 Switch activated by rod
'
1
]
1
'
45
Equation 0-8
FR1211 Push rod toward barrel
FR1212 Extend rod when leaving barrel
FR1213 Prevent rod from moving back after launch
X O O
X X O
O O X
DP1211 Gas pressure
DP1212 Piston
DP1213 Latch mechanism
'
1
]
1
1
1
'
A mechanism to integrate this design is shown in Figure 0-7. Also in this case, the gas pressure is led
into a chamber that has a piston. On the low pressure side of the piston, there is a rod sliding along the
barrel. When launching, gas pressure builds up in the chamber forcing the rod towards the barrel.
When the depth charge leaves the launcher, the rod can move further out. This causes an electric switch
to close. In order to ensure that the switch does not open again later, a latch mechanism to hold it in
place must be introduced. This latch mechanism could be either mechanical or electrical.
Gas
Pressure
Barrel
Rod
Depth charge
moving this way
to exit launcher
Figure 0-7 Generating arming condition 1
FR15 Provide Initiation Signal
The decomposition of FR15 (provide initiation signal) is
Equation 0-9
FR151 Sense target impact
FR152 Send signal to detonator
X O
X X
DP151 Accelerometer
DP152 Switch activated by accelerometer
'
1
]
1
'
As this design can be realized using standard off-the-shelf components, it is not discussed further.
4.7.4 Final comments on the case study
The result of this case study was a more reliable and robust system where the part count was reduced
from more than 350 to less than 100. This system is now in serial production.
46
4.8 Conclusion
The case study introduced in section 4.7 shows that the framework was useful for a designer following
the axiomatic design method. Therefore, hypothesis H1 can not be rejected. The case study also shows
that the designer was not impaired by the framework. Therefore, hypothesis H2 must be rejected.
In conclusion, there is now evidence supporting the hypothesis that the framework provides an
accurate representation of the sources of design information in the context of axiomatic design.
Furthermore, it must be concluded that the framework (at least in this special case) is a useful tool that
provides the designer with an effective framework for structuring, searching and managing design
information.
5.
47
Managing Information in the framework
The first part of this section contains a development of new theory on how to utilize design information
once it has been captured. The second part shows the results of two projects aimed at creating computer
systems that realize some of the functions outlined in the theory development (one developed
independently of this research project [GOLD94] the other in close cooperation with this research
project).
5.1 Research Question
How can design information, obtained using the proposed framework, be of benefit during later
stages of the design objects life cycle?
5.2 Research Method
A theory on how to utilize design information in the framework during later stages of the design
objects life cycle is developed. The feasibility of creating a computer system able to implement this
theory is also discussed.
5.3 Theory Development
Information captured in the functional, physical and process domains of the framework follows the
format of axiomatic design. The following can be stated about the information captured in the
framework:
FRs, DPs and PVs are organized in hierarchical structures in the functional, physical and
process domains respectively [SUH90 p. 37-38].
At each level of the hierarchy, the FRs, DPs, and PVs form characteristic vectors [SUH90
p. 54].
A characteristic vector at any specific level of the design hierarchy in one domain
(functional, physical or process domain) is related to the characteristic vectors at the same
level in the other domains. The relationships are described by the [A] and [B] matrices
[SUH90 p. 129].
A characteristic vector at a certain level of abstraction is related to the characteristic
vectors at the higher and lower levels of abstraction in the same domain. These
relationships are found in the hierarchical structure of the functional, physical and process
domains.
48
This section will show how this knowledge can be put into practice. The goal is to develop a
prescriptive theory for utilizing design information to develop a computerized system. Such system
would support the search for problem sources in a design object, facilitate field service of the design
object, enable effective management of engineering change orders, and allow design decisions to be
traced.
What
How
What
How
FR1
FR11 FR12
FR13 FR14
DP1
DP11 DP12
DP13 DP14
PV1
PV11 PV12
PV13 PV14
Why Why
Figure 0-1 Hierarchies of FRs, DPs, and PVs.
Figure 0-1 shows a simple information structure and the relationships between the elements of this
structure. Moving from left to right in the structure takes the designer from what to how [SUH90
p. 25-26]. Going from right to left helps the designer answer the question why. Vertical movements
bring the designer to different levels of abstraction. The lower levels in the hierarchy contains the most
specific information while the highest levels contains the most abstract information. This simple
structure, and the implications of moving within the structure will be used as a basis for the theory
developed in this section.
5.3.1 Finding problem sources in a design object
When a system, subsystem, or prototype malfunctions, the framework can be very useful in determining
whether the problem is rooted in the product or process.
49
Figure 0-2 Unsatisfied top level FR
Assume that the top level FR (i.e., FR1) of a design is not satisfied (Figure 0-2). Since every FR is
decomposed (unless it is a leaf FR at the lowest level of abstraction), the first step in localizing the
source of the problem is to identify the FRs at the next lower level in the hierarchy below FR1. Next, it
is necessary to investigate which of these FRs is not met. Since FR1 is achieved when all the FRs in the
hierarchy below FR1 are achieved, it is assumed that if FR1 is not achieved, at least one lower level FR
must also not be achieved. This process is then repeated until an unsatisfied leaf FR is found (Figure 0-
3).
Figure 0-3 Tracing unsatisfied FRs
When an unsatisfied leaf FR is found, its corresponding DP must be investigated. This DP is found in
the same place of the DP hierarchy as the corresponding FRs place in the FR hierarchy (see Figure 0-
4).
50
FRs
DPs
Figure 0-4 Finding an FRs corresponding DP
The result of this investigation will show whether the DP meets its specifications.
In the cases when the DP meets its specifications, the designer can conclude that the manufacturing
process performs as intended, but the DP does not perform as intended, i.e., the problem is in the
product. This means that the DP is not appropriate for the FR, and a new design must be sought. This
leads to the following theorem:
Theorem A: If a design parameter (DP) meets its specifications, but fails to provide the intended
function, then the DP must be replaced by another DP. Consequently, its corresponding process
variable (PV) must also be changed (the consequence is derived from theorem 5 [SUH90]).
DPs PVs
Figure 0-5 Finding a DPs corresponding PV
51
The problem is in the process when the DP does not meet its specifications. This occurs when the
corresponding PV is incorrect or has not been adjusted correctly (see Figure 0-5). Theorem B provides
insight into this situation:
Theorem B: When a design parameter (DP) does not meet its specifications, its corresponding
process variable (PV) must be either replaced or adjusted.
This investigation of the information structures that were developed during the design process, made it
possible to derive two theorems that apply when locating flaws in a design. These theorems provide the
designer with a rationale for changing the specifications of the design object (the DPs) (and accordingly
the process (PVs) intended to realize the new DPs) or only changing the process (PVs).
5.3.2 Proposing an information system for field service
An analog reasoning to the analysis in section 0 can be used to derive how the information generated
during the design process can be used to create an information system that will support (field) service
13
of a design object.
A field service information system would guide the technician to the lowest level FR that is not satisfied
(this process was illustrated in Figure 0-2 through Figure 0-4). When this FR has been identified, the
system would present information about and prescribe changes to the specific DP (or DPs if the design
is decoupled) that were implemented to generate the function.
Besides supporting the service department, such a system would facilitate statistical analysis of the
failure rate of various systems and components in the field (or elsewhere). This information can be fed
back to the design department for consideration in future design projects (this feedback idea is also
disussed by Stadler, Nordlund et al, in Ericsson Telecoms vision for hardware testing in 1999
[STAD93]).
5.3.3 Tracing the impact of an Engineering Change Order (ECO)
The information structure can also be used to predict the impact of an engineering change order (ECO).
An ECO is a decision to change one of the DPs in the hierarchy, without changing DPs at levels above
this DP. Clearly, if a DP is changed, all DPs in the structure below it may also change since they were
decompositions of the higher level DP that was changed.
52
Figure 0-6 shows a DP hierarchy which contains one DP that potentially should be changed. This
section explains how engineers can investigate the impact this proposed change will have on the rest of
the design.
DP1
Figure 0-6 Proposed engineering change order in DP hierarchy.
All available information about the design is located in the FR, DP and PV hierarchies, as well as the
design matrices. The design matrices contain information about how FRs and DPs, as well as DPs and
PVs are related. Figure 0-7 illustrates how the design matrices are used to trace the impact of an ECO.
X O
X X
1
]
1
Figure 0-7 Relating FR and DP hierarchies through the design matrix.
The design matrix in Figure 0-7, shows that the design in this example is decoupled at this level of
abstraction. This means that a change to DP
a
(gray in this figure) will affect both FR
a
and FR
b
. DP
a
will be set to satisfy FR
a
, but will also impact FR
b
as is illustrated by the off-diagonal element in the
design matrix. In order to ensure that FR
b
is still satisfied, it is likely that DP
b
must also be adjusted.
13
Field service is defined as servicing a design object that has been delivered to a customers site,
whereas service is a more generic term, not referring to where the design object is located.
FR
a
FR
b
DP
b
53
X O
X X
1
]
1
DP
b
Figure 0-8 Impact of an engineering change order
Since DP
b
is decomposed further (i.e., is made up of more specific DPs), the designer must go to the
lowest level of the hierarchy to find which specific DPs must be changed in order to ensure that FR
b
is
satisfied.
The ECO should not impact higher levels of the design because DPs at higher abstraction levels of the
design remain unchanged. Higher level DPs impose a number of constraints on all lower level DPs.
These constraints ensure that an ECO that is within the constraints does not impact the higher level DPs
in the hierarchy.
Theorem C: An engineering change order can only affect the design parameters at the same level
and lower abstraction levels in the design hierarchy. As a consequence of this, process variables
(PVs) corresponding to the changed DPs may also need to be changed (the consequence is derived
from theorem 5 [SUH90]).
Theorem D: When the design is decoupled or coupled, an engineering change order will impact
more than one FR. Consequently, it is likely that more than one DP must be changed as a part of the
engineering change order.
5.3.4 Tracing design decisions
As was shown in Figure 0-1, the relationship between adjacent design domains can be described as a
what - how relationship. The domain to the left contains what should be achieved, while the domain
to the right contains information about how this will be achieved. This implies a motion from left to
right across the domains during development of a design object.
54
Moving across the design domains from the right to the left is possible when a design object already
exist. This process traces what a PV or DP was supposed to achieve. Through this process one can
learn why a specific PV or DP became part of the process or design object (see Figure 0-1).
Based on this observation, it should be possible to devise a system for capturing design history. Some
initial studies on how to capture design history related to axiomatic design have been presented by
Lindholm [LIND96] and Malmqvist [MALM95].
Lindholms paper focuses on capturing rationale and history of decisions made during the design
process and facilitate discussions in the design teams. Lindholm outlines the types of information
(especially about functional requirements (FRs) and design parameters (DPs)) that should be captured
as part of the design history. Finally, he suggests that it would be useful to link FRs and DPs to CAD
drawings.
Malmqvists paper proposes a computerized approach to capturing design history. This approach is a
combination and extension of Function Means Trees [ANDR80] and the Chromosome Model
[ANDR92]. The Extended Function Means Tree serves as a tool to guide the synthesis process while
the Chromosome Model provides a consistent product modeling template capable of capturing design
rationale as the design process progresses. Malmqvist recommends that designers work with these two
models concurrently to ensure that both models contain up-to-date design information. Malmqvist
provides examples where this approach has been tested. He also suggests linking this system to CAD or
word processing programs.
5.4 Computer systems that realize some of the functions in the above theory
This section summarizes the results of two projects aimed at capturing design information. The first
project was performed by engineers at Tetra Pak Research and Development AB in Lund, Sweden, with
theory support from this research project. This work is part of an on-going product development project
at Tetra Pak. The second project by Goldis [GOLD94] was conducted in parallel to and completely
independent from this research project. Goldis system is similar to that developed in the first project
and is discussed here to address the feasibility of implementing information infrastructures to support
design methods in computer software.
5.4.1 Background
In industry, design information is normally captured in the form of specifications, design meeting
protocols and models. The models are either physical models (prototypes, etc.), or abstract models, e.g.,
55
drawings or computer models (like the CAD model of a subsystem in a packaging machine shown in
Figure 0-9).
Figure 0-9 CAD model of subsystem in packaging machine (from Tetra Pak)
Such models (both physical and abstract) only capture information from the physical domain in the
proposed framework. Accordingly, when presented with pure CAD models, individuals other than the
original designer have difficulty determining the exact functions of each component and problems
establishing the functional relationships between the components.
To enable designers to utilize the theory developed in the earlier parts of this section, a new type of
design information system is required. The first step in designing such a system is to determine its
functional requirements and constraints. These are presented below:
FR1: Capture information prescribed by the proposed framework, and
FR2: Model the design object as the design process progresses.
C1: Must be closely linked to the companys product development process.
Section 0 discusses a realization of a design information system based on axiomatic design [SUH90]
while section 0 discusses a realization of a design information system based on total quality
development (TQD) [CLAU94].
56
5.4.2 Design information system based on axiomatic design
5.4.2.1 Analyzing the design process and information flows
The design process used in this project is illustrated in Figure 0-10.
Design
initiative
Define
design
problem
Search/create
overall
functional
principles
Define
sub-
problem
Search/create
sub-
functional
Principles
Assemble sub-
functional
principles into
overall principle.
time
Iterate until completely decomposed
Figure 0-10 Design process model showing the main design phases
(adapted from P. Andersson, Tetra Pak)
The first task in this project was to investigate the design information produced during different phases
in the design process model. Next, this information was related to the framework and a CAD product
modeling tool in order to understand how information flows during the design project. The result of this
analysis is presented in Figure 0-11.
Design
initiative
Define
design
problem
Search/create
overall
functional
principles
Define
sub-
problem
Search/create
sub-
functional
Principles
Assemble sub-
functional
principles into
overall principle.
A fuzzy
need
Express
need in
engineer-
ing terms
Make logical
schemes,
sketches,and
assumptions
Evaluate
and
decompose
concept
Make logical
schemes,
sketches,and
assumptions
Create
helicopter
view
Customer
requirements
DP1
DP11 DP12
Constraints
C1
Solution generated
constraints
C1
C11
C12
...
CAD
model
time
Design
Phases
Information
Iterate until completely decomposed
FR1
FR11 FR12
Environment
Figure 0-11 Relating the design process to information infrastructure
(adapted from P. Andersson, Tetra Pak)
57
5.4.2.2 Finding existing tools that satisfy the requirements
In this project, extensive, in-house software development resources could not be expended on a new
information system. Therefore, it was desirable to find existing software solutions that would satisfy the
FRs (ideally, software that was already used within the company in order to avoid new investments).
After evaluating several available software packages, it was concluded that Microsoft Project and a
word processor would enable designers to capture the information prescribed by the framework.
Specifically, Microsoft Project has the ability to create the large tree structures, required by the
framework.
To model the design object, the team selected Pro Engineer, a commercially available CAD software.
Equation 0-1 shows the design equation for this system.
Equation 0-1
FR1 Capture information according to framework
FR2 Model design object
X O
O X
DP1 MS Project
DP2 Pro Engineer
'
1
]
1
'
The equation shows that this is an uncoupled design, implying that the two software packages can be
treated as independent modules and can be replaced by other software packages independently of one-
another. This makes the system easy to upgrade and change.
5.4.2.3 Implementation - Case study
The feasibility of the proposed system was tested during the development of a new packaging machine
at Tetra Pak. This case study shows how design information for one of the subsystems of this
packaging machine was captured. The function of this subsystem is to form the package during the
filling process.
To begin, the highest abstraction level functional requirements (FRs) and their corresponding design
parameters (DPs) were established by the project management. Next, these FRs were decomposed to
the next level of the hierarchy by the project management. At this level, the FRs were assigned to
project engineers who were responsible for realizing the desired function. One such FR (FR 3.5:
form/define the package during filling) and its subsequent decomposition are illustrated in Figure 0-12.
The corresponding DP hierarchy is presented in Figure 0-13.
58
FR 3.5
Form/define
the package
during filling
FR 3.5.1
Support/enclose
package back
FR 3.5.2
Support/enclose
package front
FR 3.5.3
Support/enclose
package sides
FR 3.5.4
Support/enclose
package bottom
FR 3.5.5
Support/enclose
package top
FR 3.5.1.1
Ability to slide
up/down
FR 3.5.7
Hold
package
FR 3.5.1.2
Locked in
horizontal pos.
FR 3.5.1.3
Locked in
vertical position
FR 3.5.3.1
90 degree
drive
FR 3.5.6
90 Degree
main drive
FR 3.5.6.1
Hold forming
unit horizontal
FR 3.5.6.2
Hold
motor/gear
FR 3.5.4.1
90 degree
drive
Figure 0-12 FR tree for part of packaging machine
DP 3.5
A 90 degree
rotating forming
mechanism and a
pump system
defines the package
DP 3.5.1
Movable back-
plate that follows
the package back
DP 3.5.2
PP Hinge
front
DP 3.5.3
PP Hinge
sides
DP 3.5.4
PP Hinge
bottom
DP 3.5.5
PP plate
mounted
in the
rotating unit
DP 3.5.7
Vacuum cups
DP 3.5.1.2
Arm
DP 3.5.3.1
Inner
gear mechanism
DP 3.5.6
Servo motor &
gear + outer
gear mechanism
DP 3.5.6.1
PP Hinge sides
DP 3.5.6.2
Bracket
DP 3.5.4.1
Inner
gear mechanism
Figure 0-13 DP tree for part of packaging machine
Figure 0-14 was created to capture the design information in Figure 0-12 and Figure 0-13, in a format
that was found to be easier for the engineers to read.
Note that either representation formats are acceptable provided the properties of the information
structure (i.e. the vertical and horizontal relationships in the framework) are maintained to enable
utilization of the theory developed in the early parts of this section.
59
In parallel to the development of these information structures, CAD models of the design object were
also developed. Figure 0-15 shows a CAD model for a packaging machine at the highest level of
abstraction (A-level layout). The top level FRs are inserted on the model, in this case they are: To
form, fill and seal a package. The applicable constraints, as well as their sources (e.g., the projects
steering committee, the market or the package) are identified. Some of these constraints are then
translated into geometries shown in the drawing on the right (e.g., machine length).
In accordance with the framework, DPs are selected and illustrated on the drawing, first at a conceptual
level then at a more detailed level (called ultimate solution in the drawing). Finally, as was shown in
the framework, each selection of a DP introduces new constraints that apply at lower levels. These are
captured in the left part of the figure under the heading Solution generated constraints.
60
FR 3.5
WHAT
Form/define
the package
during filling
DP 3.5
HOW
A 90 degree
rotating forming
mechanism and a
pump system
defines the package
FR 3.5.1
WHAT
Support/enclose
package back
DP 3.5.1
HOW
Movable back-
plate that follows
the package back
FR 3.5.2
WHAT
Support/enclose
package front
DP 3.5.2
HOW
PP Hinge front
FR 3.5.3
WHAT
Support/enclose
package sides
DP 3.5.3
HOW
PP Hinge sides
FR 3.5.4
WHAT
Support/enclose
package bottom
DP 3.5.4
HOW
PP Hinge
bottom
FR 3.5.5
WHAT
Support/enclose
package top
DP 3.5.5
HOW
PP platemounted
in the rotating
unit
FR 3.5.1.1
Ability to slide
up/down
DP 3.5.1.1
FR 3.5.7
WHAT
Hold package
DP 3.5.7
HOW
Vacuum cups
FR 3.5.1.2
Locked in
horizontal pos.
DP 3.5.1.2
Arm
FR 3.5.1.3
Locked in
vertical position
DP 3.5.1.3
FR 3.5.3.1
90 degree drive
DP 3.5.3.1
Inner
gear mechanism
FR 3.5.6
WHAT
90 Degree
main drive
DP 3.5.6
HOW
Servo motor &
gear + outer
gear mechanism
FR 3.5.6.1
Hold forming
unit horizontal
DP 3.5.6.1
PP Hinge sides
FR 3.5.6.2
Hold motor/gear
DP 3.5.6.2
Bracket
FR 3.5.4.1
90 degree drive
DP 3.5.4.1
Inner
gear mechanism
Figure 0-14 Design information about part of packaging machine represented in a form of
function means tree (adapted from M. Sjgren, Tetra Pak)
Customer requirements:
-Continuously form and fill
packages
-Few and simple parts should be
used
Figure 0-15 CAD model at A-level showing the product model as well as information generated at a high level of abstraction. (From Tetra Pak)
Figure 0-16 CAD model at B-level showing the product model as well as information generated at a sub-system level of abstraction.
(from Tetra Pak)
63
Figure 0-16 shows the next level of abstraction for a subsystem of the same packaging machine. This
subsystems functional requirement is to form a tube where the position of the sealing is defined and
secured. Its corresponding design parameter (called Principle solution in the figure) is to guide the
edges with accurate tolerance with respect to the LS-sealing. This DP is visualized in the right part of
the figure. The constraints that apply at this level are found under the heading Constraints. The
specific constraints generated at this level as a result of the chosen solution are found under the heading
Solution generated constraints.
Eventually, the design progresses to a level where all subsystems can be implemented. At this point it is
possible to extract drawings of each subsystem and equip them with references to the functional
requirements and design parameters that were documented in the tree structures. An example of such a
drawing is found in Figure 0-17. This figure shows the implementation of the function means tree in
Figure 0-14. This is also the same information as is found in the FR and DP trees in Figure 0-12 and
Figure 0-13.
Figure 0-17 CAD model of mechanism in packaging machine with references to the FR/DP
structures in Figure 0-12 and Figure 0-13 as well as the function means tree in Figure 0-14.
(Adapted from M. Sjgren, Tetra Pak)
3.5.6
3.5.6.2
3.5.3.
3.5.3
3.5.4
3.5.4.1
3.5.6
3.5.6.
64
5.4.3 QCAD - Quality Computer Aided Design
QCAD - a system for coordinating design information among different parts of the product
development process - was developed and implemented in software by Yale Goldis [GOLD94]. It was
primarily developed for use with the total quality development (TQD) approach which follows the
EQFD method [CLAU94]. QCAD combines the concept of domains [SUH90] and QFD analysis
[CLAU94] with the product model programming in [ICAD93]. This combination yields an information
system able to link functions with the hardware parts that realize those functions [GOLD94 p. 55].
The information system supports the designer to develop three hierarchies (function hierarchy,
hardware hierarchy, and process hierarchy). Information relevant to each hierarchy is captured as
specified below:
Functional hierarchy [GOLD94 p. 74-85].
Functions - describe what need the product meets.
Constraints - based on the use environment or customer needs
House of quality (HoQ) - provides a tool to combine engineering metrics with abstract
requirements
Function diagrams - based on design structure matrices [ULRI95], and showing
information flows between functions in order to determine the relationships between
these functions.
Concept generation - a list of the ideas generated that can be reused in the future
Concept selection - uses the metrics in the HoQ to evolve concepts in the Pugh Concept
Selection matrix [PUGH91].
Concept review - a matrix displaying the relationships between functions and parts,
enabling the product development team to avoid interactions that are too complex
[GOLD94 p. 84].
65
Hardware hierarchy [GOLD94 p. 85-93]
Product structure diagram - based on design structure matrices [ULRI95] showing
information flows in a set of design activities.
House of quality (HoQ) - provides a tool to relate hardware design parameters to
engineering metrics identified in the HoQ in the functional hierarchy.
Hardware parameters - variables such as dimensions, material choice, geometries, etc.
Product parameter design - optimized parameter values.
Materials and process selection - uses the Pugh Concept Selection matrix [PUGH91]
to select materials and processes for piece-parts.
Concept review - evaluates the relationship between the hardware and process to avoid
unnecessary complexity.
Process hierarchy [GOLD94 p. 93-95]
Bill of materials - itemization of all parts in an assembly
Process parameters - detailed values of the production process, e.g., assembly
sequences, speeds or feeds of a milling operation.
Process parameter design - the optimal setting of parameters for the production process
House of Quality (HoQ) - translating hardware characteristics to process requirements
These information structures can be displayed in the same computer window as a CAD model of a
product, enabling the designer to view more information than is possible in a pure CAD model.
Goldis points out that QCAD is very useful not only during the design process, but also when the
designer wants to change a function, as the system will then identify all hardware affecting this function
[GOLD94 p. 74]. This is achieved through the function diagram and concept review matrices. The
function diagrams indicate functions affected by or affecting the function that the designer wishes to
change. The concept review matrix shows all hardware parts affecting those functions that may be part
of the change.
During redesign, the designer can study the existing hardware implementation of a function to generate
new ideas [GOLD94 p. 74]. QCAD facilitates this in two ways: by showing the current design, and by
providing a list with all ideas generated during concept generation (a part of the functional hierarchy).
66
5.4.4 Discussion on the differences between the two computer implementations
The main difference between the two projects described in sections 0 and 0 is that they are based on
two different design processes: Tetra Paks system (TPS) is based on axiomatic designs process
[SUH90] while QCAD is based on EQFD [CLAU94].
QCAD uses function diagrams (design structure matrices) to make relationships between functions
explicit. Then, concept review matrices are used to find the hardware component related to these
functions. Using these two matrices, QCAD can provide a link from functions, to hardware and
processes, but not necessarily the other way.
Goldis concludes that this system can identify what hardware and processes must be changed when the
designer changes a function. This is consistent with theorem 5 in axiomatic design. This theorem states
that when a given set of FRs is changed, the design solution given by the original DPs cannot satisfy
the new set of FRs. Consequently, a new design solution must be sought.
It seems straightforward to implement a function in the current version of QCAD that would realize the
theory discussed in sections 0 and 0 (finding problem sources and providing feedback to the product
development team). However, since the links between the hierarchies in QCAD only go in one direction
(from the functional hierarchies to the hardware and process hierarchies) it is not clear how the theory
on engineering change orders (maintaining the functional requirements but changing the design
parameter) described in section 0 could be easily implemented in QCAD. Some small modifications,
especially on how relationships are captured should increase QCADs ability to also support axiomatic
design and the proposed framework.
The big advantage of QCAD over TPS is that it is implemented in one computer software system
thereby eliminating the need to manually link information structures with the CAD models.
Furthermore, QCAD currently implements many more functions than TPS.
67
5.5 Conclusions
A new theory was developed to show how design information (developed using the proposed
framework) can be of benefit during later stages of the design objects life cycle. This theory shows
how this design information can be used to accomplish the following:
Guide a search for problem sources in a design object,
facilitate field service of the design object,
enable effective management of engineering change orders, and
allow design decisions to be traced.
Four new theorems were stated that would support the designer in this work:
Theorem A The design parameter meets is specification but do not deliver the function
If a design parameter (DP) meets its specifications, but fails to provide the
intended function, then the DP must be replaced by another DP. Consequently, its
corresponding process variable (PV) must also be changed.
Theorem B The design parameter does not meet its specification
When a design parameter (DP) does not meet its specifications, its corresponding
process variable (PV) must be either replaced or adjusted.
Theorem C Impact of an engineering change order in the design hierarchies
An engineering change order can only affect the design parameters at the same
level and lower abstraction levels in the design hierarchy. As a consequence of
this, process variables (PVs) corresponding to the changed DPs may also need to
be changed.
Theorem D Impact of an engineering change order
When the design is decoupled or coupled, an engineering change order will impact
more than one FR. Consequently, it is likely that more than one DP must be
changed as a part of the engineering change order.
Two projects that implement information infrastructures to support design methods were described, one
based on axiomatic design, the other on EQFD. The first project was conducted by one industry in
close cooperation with this research project. The second project was completely independent from this
research project.
68
The results from these projects clearly demonstrate that it is practical and useful to develop a system
capable of capturing more information than pure CAD models developed during design processes. They
also show that it is feasible for such system to utilize the theory developed in this section.
Both projects also showed that it is possible to concurrently develop and link information captured in
the FR, DP, and PV hierarchies with CAD models. Such linkage helps designers visualize the design
object and facilitates its implementation in systems and components. A CAD system with the ability to
link models to the tree structures can provide visual models (rather than text models) that would guide
the search for problem sources in a design object and facilitate field service of the design object. The
system would also visually represent all components that are affected by an engineering change order.
6.
69
Utilizing principle-based methods and tools to effectively generate the design object within the
framework
Thus far, this thesis has introduced an extended information framework around Suhs domain concept,
and shown where sources of information in this framework are located. Furthermore, it has been
demonstrated how information generated and captured in this framework can be used to improve
various design related activities, such as engineering change orders and problem searching.
In a previous paper [NORD94], the theory of inventive problem solving [ALTS88, SUSH95], was
proposed as a complement to axiomatic design [SUH90]. Specifically, this section investigates how
Suhs independence axiom [SUH90] and the theory of inventive problem solving can be used within the
framework introduced in section 4, under the hypothesis that these theories are complementary.
The following statements about the independence axiom and the theory of inventive problem solving are
the basis for the hypothesis:
The independence axiom
14
[SUH90 p. 9] should be used during pre-CAD design to analyze a
proposed set of design parameters in order to avoid a functionally coupled design.
The theory of inventive problem solving (TIPS)
15
among other things contains laws of
engineering system evolution [ALTS88 p. 223-231] and a collection of problem solving
techniques. The problem solving techniques include generalized knowledge for generating a
desired function by identifying a set of appropriate physical, chemical, and geometrical effects
[ALTS88 p. 309-314] and a set of 40 principles for solving technical contradictions [ALTS88
p. 151-168].
14
For a more in-depth description of axiomatic design, see appendix 2, or [SUH90]
15
For a more in-depth description of TIPS, see appendix 3, or [ALTS88, SUSH95, MAZU96a,
MAZU96b]
70
6.1 Hypothesis - H3
H3: Working within the proposed framework, the theory of inventive problem solving provides a
synthesis tool complementary to the analysis rule provided by the independence axiom within the
proposed framework. More specifically, when dealing with the design of a mechanical system in the
proposed framework, Altshullers principles for resolving technical contradictions can sometimes be
applied to resolve a situation where a design parameter (DP) or a process variable (PV) does not
meet a constraint.
6.2 Theory Generation
The design process can be modeled as a feedback control loop (Figure 0-18) where X represents the
input and Y represents the desired output. G represents the synthesis capability and H represents the
analysis capability. Wilson et al [WILS79] show that when GH is very large, the gain is equal to 1/H.
According to Suh [SUH90 p. 7], when there is a lack of analysis capability, the [designer] can not
rapidly generate the best design since a good design can not be distinguished from a bad one.
Furthermore, Suh points out that in the absence of a criterion for selecting a good design, we can not
make good design decisions [SUH90 p. 7]. The design axioms were developed to provide this analysis
capability.
Figure 0-18 Design Process as a feedback loop (from [WILS79])
The framework proposed in section 4 provides the engineer with an infrastructure to capture
information that is generated during the design process. The framework also provides a context within
which the analysis and synthesis methods and tools can be placed.
Table 0-1 summarizes the contribution of each component of the theory to the over-all design process.
This table shows that the design axioms provide analysis support, TIPS provides synthesis support and
the framework provides an information infrastructure. Figure 0-19 incorporates TIPS and the design
G
H
X Y
+
-
Y
X
G
1 GH
G
GH
H for GH >> 1
1
+
71
axioms in the proposed framework. The remaining portion of this section explores in greater detail the
role of TIPS in the framework.
6.3 Using axiomatic design and TIPS in the framework
After top-level FRs have been established, there are several information generating activities the
designer must perform. First, DPs and PVs must be generated. As was shown in section 4, the designer
should investigate both the environment and the customer domain for potential DPs. To effectively
search for DPs, the designer can also employ Altshullers proposed database that matches technical
functions with physical, chemical, and geometrical effects
17
.
Next, the designer must select an appropriate set of DPs from those generated. All DPs selected must
meet all constraints as well as satisfy the design axioms.
A technical contradiction exists whenever a DP does not meet a constraint. It is important to distinguish
between a technical contradiction (from TIPS) and a functional coupling (from axiomatic design).
Technical contradiction are modeled as contradictions between two technical parameters, while
functional couplings are modeled as an interdependence between FRs and DPs. Functional coupling
may exist without a technical contradiction and vice versa.
16
The framework does provide some support for synthesis in that it prescribes certain areas in which
the designer should look for potential DPs and PVs. However, this support is not in the form of rules or
principles.
17
Such databases exist in commercially available software packages.
Table 0-1 Identifying complementary properties
between axiomatic design, TIPS and the framework.
Design
axioms
TIPS Frame
work
Decision-making support
(Analysis)
Yes No No
Synthesis Support No Yes No
16
Information
Infrastructure
No No Yes
72
World
Environment
(Super system)
Customer
Domain
Functional
Domain
Physical
Domain
FRs
Constraints
DPs
Information
Information
PVs
Process
Domain
Customer
needs
TIPS
TIPS
TIPS
TIPS
and
Design
axioms
TIPS
and
Design
axioms
Figure 0-19 TIPS and axiomatic design in the framework
18
To resolve a technical contradiction (DP-Constraint contradiction, or a PV-constraint contradiction)
Altshuller postulated 40 inventive principles. Figure 0-20 illustrates how DPs are checked against the
constraints (Cs). This activity takes place in either the physical domain or the process domain. During
this process, every DP is systematically checked against each constraint (e.g., DP1 vs. C1, DP1 vs. C2,
DP1 vs. Cn). Whenever a DP or PV violates a constraint the designer can attempt to apply one or
more of Altshullers 40 principles to overcome this problem.
DP1 DP2 DPn
C1
C2
Cn
Figure 0-20 Checking DPs against constraints
18
For increased clarity, some information found in section 4 has been removed from this figure.
73
The last step in a design cycle (before the next level DPs are generated) is the decomposition of FRs. In
order to decompose an FR, the designer must know its corresponding DP. Altshuller has developed
trends of technology evolution that can be applied when decomposing the FRs. This practice would
ensure that designers state FRs to achieve the highest possible performance.
6.4 Example: An engineering change order utilizing axiomatic design and TIPS within the
framework
The switch pictured in Figure 0-21 has been designed to fit within a small volume and manage a current
of 50 A. The design equation for this switch is shown in Equation 0-1.
Open position
Closed position
Allocated volume
Figure 0-21 Schematic drawing of a switch
Equation 0-1
FR1 Conduct Current (50A)
FR2 Break Current (50A)
X O
O X
DP1 Area of plates
DP2 Mechanical movement
'
1
]
1
'