Falcon Invoice Discounting: Empowering Your Business Growth
A survey of design philosophies, models, methods and systems
1. 301
Review Paper
A survey of design philosophies, models, methods and
systems
....; F 0 Evbuomwan, * BEng, MSc, PhD, DIC, CEng, MIStructE, S Sivaloganathan, BSc, MSc, PhD, CEng, FIMechE, MIEE,
!IE(Sri Lanka) and A Jebb, BSc, PhD, DIC, CEng, MIMechE, MIEE ·
Engineering Design Centre, City University, London
The study of the design process, design theory and methodology has been a preoccupation of engineers, designers and researchers over
1he fast four to jive decades. As the end of this millenium is approached and with the renewed inrerest around the world in engineering
design. it is fitting to examine the state of the art and currenr status of issues relating to design philosophies, theory and methodology.
(Jeer the last 40 years, many approaches to design have been put forward by various researchers, designers and engineers, both in
, 1cademia and industry, on how design ought to and might be carried out. These proposals on design have tended towards what has
, orne to be regarded as design philosophies, design models and design methods. The thesis of !his paper is to discuss various aspects of
,1eneric research in design, within the above classifications in the light of the work that has been done in the last four decades.
Discussions will focus on various definitions of design, design theory and methodology, the nature and variety of design problems, design
classifications, philosophies, models, methods and systems.
f ev words: design philosophies, design theory, design methodology, design models, design methods, design systems, design process,
rrciduct design, computer aided design
1 INTRODUCTION theory and methodology. Further discussions will be on
The design activity, although performed for many cen- the nature and features of the design process, the nature
turies, did not, however, have any structure or organiz- and stages of thought in design, types of design prob-
;.ttion to it. It was only just after the middle of this lems, product design classification and design goals.
century that efforts began to give some formalism to the
way design was done. What is design? Why is it done?
How is it or can it be done? These questions have been 2.1 Definitions of design
the subject of discussions at various conferences on Several designers, engineers and researchers, from
engineering design and design methodology. In these observation and experience, have expressed their views
conferences, which were held in the United Kingdom on the definition of design or what they consider design
il-5), Europe (~10) and North America (11-18), a to be. Some of these viewpoints are expressed below:
number of ideas were put forward on design method-
ology. These ideas were mostly associated with design Feilden (19): 'Engineering Design is the use of scientific
models, philosophies and methods or techniques as well principles, technical information and imagination in
as applications, and they represented several schools of the definition of a mechanical structure, machine or
thought on design and design methodologies. More system to perform prespecified functions with the
recently, other researchers have started to report on maximum economy and efficiency.'
computer-based design systems . Finkelstein and Finkelstein (20): 'Design is the creative
. The main focus of this paper is to give a detailed elu- process which starts from a requirement and defines a
cidatiOn of design philosophies, models, methods and contrivance or system and the methods of its realis-
systems that have been proposed and developed over ation or implementation, so as to satisfy the require-
the years. Discussions will centre on: definitions of ment. It is a primary human activity and is central to
design and design methodologies, the nature and fea- engineering and the applied arts.'
tures of design problems and the design process, as well Luckman (21): 'Design is a man's first step towards the
~s the stages of thought in design and product classi- mastering of his environment . . . The process of
tJcati?n. The nature and control of design goals will also design is the translation of information in the form of
be discussed, including an extensive review of many requirements, constraints, and experience into poten-
design models, methods and systems. This paper focuses tial solutions which are considered by the designer to
on completely general aspects of engineering design meet required performance characteristics ... some
research, and it should be noted that there is a large creativity or originality must enter into the process
amount of other work in this area. for it to be called design.'
Archer (22): ·. . . design involves a prescription or
2 DESIGN, DESIGN THEORY AND DESIGN model, the intention of embodiment as hardware, and
METHODOLOGY the presence of a creative step.'
The discussions in this section focus on definitions of Caldecote (23): '. . . the basic design function ... to
design, as well as definitions and viewpoints on design design a product which will meet the specification, to
design it so that it will last and be both reliable and
; ~e liS was received on 3 June 1995 and was accepted for publication on easy to maintain, to design it so that it can be eco-
· 'nrember /995.
: Prnenr address: Deparrmenr of Civil Engineering, University of Newcastle. nomically manufactured and will be pleasing to the
·' •{ a.stle upon Tyne.
eye.'
Proc Instn Mech Engrs Vol 210
3. A SURVEY OF DESIGN PHILOSOPHIES, MODELS, METHODS AND SYSTEMS 303
thought, as they move from an abstract problem to a tures are introduced, which still bear some resem-
realizable product. These are the divergent, transform- blance to existing variables or features, and the.
ation and convergent stages of design: decomposition of the problem is known. but the
sub-problems and various alternatives to their
1. Divergence. This is the act of extending the boundary
solution must be synthesized. In other situations,
of a design situation in order to have a large enough
alternative recombination of the sub-problems
solution search space. The divergent search approach may yield new designs. It is also considered that
aims to de-structure the original design brief, while
solving the same problem in different ways, or
identifying the features of the design situation that different problems in the same way (by analogy),
will permit a valuable and feasible degree of change.
would fall under this class.
Divergent search is most productive in the initial
(b) Creative designs. In this case new variables or
stages of design.
features are introduced which bear no similarity
2. Transformation. This is the stage of pattern making, to variables or features in the previous prototype
high-level creativity, flashes of insight, changes of set and the resulting design has very little resem-
and inspired guesswork. The objective here is to blance to existing designs. For creative designs
impose upon the results of the divergent search a no design plan is known, a priori, for the problem
pattern that is precise enough to permit convergence under consideration.
to a single design.
3. Convergence. The main objective of the convergent Sriram et al. (32), in the light of the foregoing,
stage is to progressively reduce secondary uncer- describes the creative-routine spectrum of design as
tainties as fast as possible as well as ruling out alter- follows: 'At the creative end of the spectrum, the design
natives. The end result of this stage should be the process might be nebulous (hazy), spontaneous, chaotic,
reduction of the range of options to a single chosen and imaginative, whereas at the routine end, the design
design as quickly and as cheaply as can be managed is precise, predetermined, systematic, and mathematical.'
and without the need for unforeseen retreats or
recursion.
2.6 Product design classification
The end result of any design process is a product or
2.5 The variety of design problems
system. Such products, depending on the engineering
A design is strongly influenced by the lifestyle, training discipline or domain, vary in one way or the other.
and experience of the designer, and the creativity and Product variation also arises depending on the market
effort a designer puts into a design varies, depending on segment, knowledge available, the design process and
the type of design problem (30). Design problems that manufacturing capabilities. In the light of general con-
confront engineers and designers can be classified under straints, products can be classified as either over-
the following types (30-33): constrained or underconstrained, and depending on the
customer demands and competition in the market, some
1. Routine designs. These are considered to be derived products are considered as static or dynamic. These
from common prototypes with the same set of vari- various forms or classifications are discussed below (34,
ables or features and the structure does not change.
35):
Here a design plan exists, with sub-problem decom-
position, alternatives and prototypical solutions 1. Static product designs. Static products are those
known in advance. whose market share is undiminishing and no changes
2. Redesigns. This involves modifying an existing design are being demanded in the product. The design
to satisfy new requirements or improve its per- concept is already known from existing products,
formance under current requirements. The end result and hence such products are considered to be con-
of redesigns may also exhibit some form of creative, ceptually static (also referred to as dominant design).
innovative or routine design content. Redesigns will 2. Dynamic product designs. Dynamic products have a
be discussed under adaptive designs and variant limited life before the next generation supersedes
designs. them. Here, development is focused on the product,
(a) Adaptive, configurative or transitional designs. and the design process involves the development of
These forms of design involve adapting a known new, radical and alternative designs. In discussing the
system (solution principle remaining the same) to dynamic-static spectrum of products, Clausing (34)
a changed task. They also involve improvements highlights the following types of products sand-
on a basic design by a series of 'detail' refine- wiched between the two extremes: (a) genesis
ments. product, (b) radical product, (c) new product, (d)
(b) Variant, extensional or parametric designs. This clean sheet (generational) product, (e) market-
follows an extrapolative or interpolative pro- segment entry (new) product, (f) market-segment
cedure. The design technique involves using a entry (generational) product, (g) associated product,
proven design as a basis for generating further (h) variant product and (i) customized product.
geometrically similar designs of differing capa- 3. Overconstrained product designs. These products
cities. tend to exist in the high-technology markets. Here,
3. Non-routine designs, original or new designs. These the design process evolves around analysing alterna-
forms of design are also known as original designs tive proposals until the <;orrect (or most acceptable)
and are classified into innovative and creative solution is found. Overconstrained products are
designs. usually subjected to several constraints of function,
(a) Innovative designs. Here new variables or fea- materials, manufacturing processes, some of which
:g IMechE 1996 Proc Instn Mech Engrs Vol 210
5. A SURVEY OF DESIGN PHILOSOPHIES, MODELS, METHODS AND SYSTEMS 305
tory to practical usefulness and that the creativity of good number of these models emphasize the need to
designers can be hampered and may deteriorate if perform more analytical work, prior to the generation
design methodologies are adopted. In this school of solution concepts (39). Models put forward by pro-
emphasis is placed on the significance of case historie~ ponents of prescriptive models are discussed below.
of design, including all necessary knowledge to be learnt
for improving design ability. This school of thought is 4.1.1 Model by J. C. Jones
closely associated with the view that the design ability
cannot be acquired efficiently in a theoretical manner This model by Jones (1) is principally made up of three
but by experience. ' stages, viz. analysis, synthesis and evaluation. At the
analysis stage, the first activity involves producing a
random list of factors related to the problem to be
3.4 Summary solved and/or to its solution. These factors are then
The above schools of thought, although they stand their classified into workable categories and sub-categories,
ground in arguments, are relevant in one way or the after which the interactions between them are investi-
other with respect to design. In today's world, it is gated. The final step then involves rewriting all the
increas!ngly becoming evident that design approaches design requirements into solution neutral performance
belongmg to the syntax (prescriptive models) school of specifications.
thought are more likely to stand the test of time. In the synthesis stage, creative techniques like brain-
Wallace (44) in his article points out that 'the engineer- storming (45, 46) are used to generate ideas and solu-
ing design process cannot be carried out efficiently if it tions to the performance specifications. Limits are then
is left entirely to chance .. .' and ' ... the aim of a sys- set for each partial solution within a range of dimen-
tematic approach is to make the design process more sions, shape and variations in material properties that
visible and comprehensible so that all those providing will satisfy any performance specifications. The next
inputs to the process can appreciate where their contri- step then involves combining compatible partial solu-
butions fit in'. Furthermore, the need to equip and train tions into combined solutions.
young engineers as well as support collaborative design The last stage of this model is the evaluation stage
teams will necessitate the adoption of a structured and which involves mainly two activities. These are (a)
systematic approach to design. methods of evaluation and (b) evaluation for operation,
manufacture and sales. Under methods of evaluation,
Jones advocates the use of evaluation methods to detect
4 DESIGN MODELS errors at the stage when they can be most cheaply cor-
Design models are the representations of philosophies rected. Such methods include evaluation by per-
or strategies proposed to show how design is and may formance specifications and evaluation by use of precise
be done. Often, they are drawn as flow diagrams, judgements.
showing the iterative nature of the design process by a This model emphasizes the need to establish specifi-
feedback link. cations in a solution neutral form as well as investigat-
In the past, design models that arose from various ing interactions between design factors. The synthesis
philosophical viewpoints have tended to belong to two stage does exhibit a bottom-up approach in developing
main classes, namely prescriptive and descriptive the overall design. The idea of evaluating the designs by
models. The prescriptive models are associated with the the pre-operation, pre-production and pre-sales team is
syntactics school of thought and tend to look at the a late occurrence in this model. These teams in a
design process from a global perspective, covering the modern manufacturing industry should be involved
procedural steps (that is suggesting the best way some- right from the start of the design process. In this model,
thing should be done). The descriptive models, on the they should be involved at the analysis stage.
other hand, are concerned with designers' actions and
activities during the design process (that is what is 4.1.2 Model by Asimow
involved in designing and/or how it is done). More
In representing the design activity, Asimow (47) shows
recently, another group of models known here as com-
the process of design in three phases that bear on the
putational models have started to emanate. These com-
solution of the design project, while the part that deals
putational models place emphasis on the use of
with the solution of subordinate problems is repre-
numerical and qualitative computational techniques,
sented as a sequence of operations as every step of the
artificial intelligence techniques, combined with modern
process proceeds. The three phases of design repre-
computing technologies. Each of these design models,
sented are the feasibility study phase, preliminary design
although discussed under one of the above classes, share
phase and detailed design phase:
some characteristics of the other classes.
1. Feasibility study phase. In the feasibility study phase,
the need for the project is established, after which the
4.1 Prescriptive models based on the design process
design problem is explored and the design param-
These models in general tend to prescribe how the eters, constraints and major criteria identified. Plaus-
design process ought to proceed and in some cases ible solutions are generated and then analysed for
appear to suggest how best to carry out design. They their physical realizability, economic worthwhileness
also attempt to encourage designers to adopt improved and financial feasibility.
ways of working. They usually offer a more algorithmic 2. Preliminary design phase. In the preliminary design
and systematic procedure to follow, and are often phase, the best design concept from among the viable
regarded as providing a particular methodology. A solutions is selected. Mathematical models are then
:Q !~hE 1996 Proc lnstn Mech Engrs Vol 210
7. A SURVEY OF DESIGN PHILOSOPHIES, MODELS, METHODS AND SYSTEMS 307
carried out. These studies were used to illustrate design- 1. Programming: establishment of crucial issues and
ing as a sequence of decisions leading from the original proposal of course of action
statement of the requirements to the specification of the 2. Data collection: collection, classification and storing
details of the 'hardware' to be manufactured. The start- of data
ing point in this model is a statement of the main 3. Analysis: identification of sub-problems, preparation
y problem to be solved. This represents the starting node of design specifications, reappraisal of proposed pro-
~ in the 'Marples tree'. gramme and estimation
1 From this node, sub-problems are derived that must 4. Synthesis: preparation of outline design proposals
f be solved before a solution to the main problem is pos- 5. Development: development of prototype design(s),
sible. This involves a cyclic process of analysis of the preparation and execution of validation studies
problem, theorizing solutions, delineating these solu- 6. Communication: preparation of manufacturing
tions and modifying them (49, 50). Figure 2 shows a documents
general representation of a typical sequence of the
The above six stages were further classified and grouped
design process. In this figure, the final solution is the
into three phases, namely analytic, creative and execu-
sum of the solutions a(21211), a(22211), a(22221) and
tive. In describing his model, Archer comments that:
a(232). If, for example, a(2) is preferred to a(l) or a(3), all
' ... the special features of the process of designing is
the sub-problems p(21), p(22) and p(23) must be solved.
that the analytic phase with which it begins requires
Similarly, if a(222) is accepted as a solution to p(22),
objective observation and inductive reasoning, while the
then sub-sub-problems p(2221) and p(2222) must be
creative phase at the heart of it requires involvement,
solved. In the figure, a vertical line denotes a problem,
subjective judgement, and deductive reasoning. Once
while a slanting line denotes a solution. Eder (50)
the crucial decisions are made, the design process con-
further proposes that all precedent solutions to the
tinues with the execution of working drawings, sched-
main problem, such as competitor's models, should also
ules, etc., again in an objective and descriptive mood.
appear on the design tree. This is analogous to competi-
The design process is thus a creative sandwich. The
tive assessment in quality function deployment (51).
bread of objective and systematic analysis may be thick
The model by Marples involves three principal
or thin, but the creative act is always there in the
phases of synthesis, evaluation and decision. At the syn-
thesis phase, two activities are involved, that is the middle.' Figure 3 shows the stages and phases of the
design process as well as their interrelationships.
search for possible solutions and the examination of
proposed solutions. This phase is then followed by the
evaluation of the viable solutions against certain cri-
teria, before a final decision is made in choosing a par- 4.1.8 Model by Krick
ticular solution. Krick (52) in his model describes the design process in
five stages of problem formulation, problem analysis,
search, decision and specification. The first step of
4.1.7 Model by Archer
problem formulation involves defining clearly the design
Archer (22) defines the nature of design methodology in problem to be solved. The next step involves analysing
his model in six stages, viz.: the design problem and arriving at a detailed definition
0 POINT OF PROBLEM FORMULA noN
a331 a332
a321
p3211 p3311 p3321 I
p2111
i p222221
SOLUnON ACCEPnED
SOLUnON REJECTED
PROPOSED SOLUnON - AN AL nERNA TIVE
A SUB-PROBLEM ARISING FROM ANY AL nERNA TIVE
Fig. 2 The design model by Marples
~ I~1echE 1996 Proc Instn Mech Engrs Vol 210
9. A SURVEY OF DESIGN PHILOSOPHIES, MODELS, METHODS AND SYSTEMS 309
4.1.10 Model by Hubka Phase 4. Elaboration
The model by Hubka (53) represents the design process Step 6. Detailing and elaboration
in four phases and six stages or steps. These phases and
steps in a procedural model, as shown in Fig. 5, are:
Phase 1. Elaboration of assigned problem 4.1.11 Model by French
Step 1. Elaborate or clarify assigned specifi- This model, as shown in Fig. 6, is based on the follow-
cation
ing activities of design (54):
Phase 2. Conceptual design
Step 2. Establish functional structures and 1. The analysis of the problem phase involving the
Step 3. Establish concept identification of the need to be satisfied as precisely
Phase 3. Laying out as possible or desirable
Step 4. Establish preliminary layout and 2. The conceptual design phase involving the gener-
Step 5. Establish dimensional layout ation of broad solutions in the form of schemes
"'
a:
wz~
'-' STATE OF MACHINE SYSTEM
tJoe (CIRCLED STATEMENTS)
'-' gs ~
2 8 w ~--------------------------------------------,_--------------------~
~ < 5 ESTABUSHED DESIGN
u.. ...J c.:>
CHARACTERISllCS OF THE DESIGN DOCUMENT
~ §~ MACHINE SYSTEM
"' :::E f-
1. DESIGN SPECIFlCA TlON
2.1 BLACK BOX DIAGRAM
ESTABUSH TE~OI.OGICAL PRINCIPLE
2.2
ESTABUSH SEQUENCE ~ OPERATICN
2.3
2.4 N'PLY liS IN TP AND BOUNDAAJES ~ liS
2.5 E51'ABUSH GROUP1NG ~ FUNCTIONS
5
FUNCTION
2.6 STRUC11JRE Z ( FUNCTlDNAL STRUCTIJRE
SCHEMAllC
2.7 ( OPTIMAL FUNCTlON STRUCTURE
AI~
3.1 INPUTS TO MS. IIOOE ~ ACTION
MORPHOLOGICAL MATRIX
3.2 FAIIIUES ~ FUNCTION-CARRIERS
3.3, 3.4
CDNCEPTIJAL SCHEMA TlC
3.5
CONCEPT SKETCH
4.2 PARTs. ARRANCEIIENT. ROUGH FORM, SOME Dt.£NSIONS
4.3, 4.4 TYPE OF WATER1AL MD MANUFACTURlNG METHODS
4.5
4.6 ( OPTIIr.tAL PREUhiiNARY LAYOUT
AI~
5.2 OEF1N111£ ARRANG04ENT. FORM, ALL DIMENSIONS
5.3, 5.4 MATCR1AL AND MANUFACTURING METHODS,
PARTIAL TOL£RANCES
5.5 DIMENSIONAL DIIIENSIONAL TRUE- TO- SCALE DIMENSIONAL!
LAYOUT 1 LAYOUT 2
LAYOUT
5.6
( OPTIMAL DIIIENSIONAL LAYOUT ) - I
6.1 - MACHINE ELEMENTS
0 RELEASE
I
6.5
5.6. 5.7
~ $§>$ (IdS REPRESENTATION )
DETAIL DRAWING
ASSEMBLY DRAWING
I
Fig. 5 The design model by Hubka
c; l!ect:E 1996 Proc Instn Mech Engrs Vol 210
10. :JlO N F 0 EVBUOMWAN, S SIVALOGANATHAN AND A JEBB
Proposed schemes are critically examined to see if
they satisfy the needs, can be constructed, and how
economic they are both in first cost and in function
throughout their working life. Preliminary structural
analyses are also carried out to check the broad ade-
quacy of schemes.
4. Decision. After successive operations of conception
and appraisal, it then becomes necessary to decide on
a particular design scheme. Criteria for decision
making may include both simplicity and distinction
of the design, as well as constructability. ·
5. Checking and elaboration. This is the stage where
the designer makes sure of the adequacy of what is
proposed and performs elaboration of necessary
details. Here, models can be built and tested. Power-
ful analytical techniques can also be employed in (a)
defining the actions on the structure such as load,
temperature difference, corrosion, etc., (b) analysing
the effects of these actions and (c) comparing these
SELECTED effects with a criterion of adequacy. The end result of
SCHEMES
the design process is the communication of the
detailed design both in the form of drawings and
text.
4.1.13 Total design activity model by Pugh (56)
Pugh regards total design as the systematic activity
necessary from the identification of the market/user
need, to the selling of the successful product to satisfy
that need-an activity that encompasses product,
process, people and organization. The total des~gn
activity model consists principally of a central design
core, which in turn consists of market (user need),
product design specification, conceptual design, detail
design, manufacture and sales. The design process in
this model proceeds, firstly, by identifying a need which,
Fig. 6 The design model by French when satisfied, fits into an existing or a new market.
From the statement of the need, the product design spe-
cification (PDS), representing the specification of the
3. The embodiment of schemes phase involving the
product to be designed, is then formulated. The estab-
development of generated schemes into greater
lished PDS then acts as a mantle that envelops all the
details subsequent stages in the design core, thus acting as the
4. The detailing phase. where the selected scheme is
control for the total design activity. Within this model,
worked into finer details
the design processes flow from market to sales, is an
iterative one and recourse can be made to any of the
4.1.12 Model by Sir Alan Harris (55) earlier stages, as new ideas and information emerge.
This causes interactions between the different stages of
This model is based on proposals regarding the teach- the design core. This model also recognizes the fact that,
ing of design within the civil engineering discipline. The for effective and efficient design to be carried out, it is
model consists of five stages, viz.: appreciation of the necessary to utilize various design techniques, to enable
task, conception, appraisal of concepts, decision, check-
the designer/design team to operate the core activity.
ing and elaboration:
These design techniques or methods include:
1. Appreciation of the task. This means discovering (a) discipline-independent ones which relate directly to
what is needed and ascertaining what resources are
the design core and can be applied to any product
needed and from where. It involves finding out what
or technology, such as tools for performing analysis,
a client wants-regarded as the 'total function'.
synthesis, decision making, modelling, etc.;
2. Conception. In this stage, based on the full digestion
(b) specific discipline-dependent technique~ and tech~o
of the facts generated from the previous stage, ideas logical knowledge such as stress analysis. hydraulics,
of solutions should begin to emanate. Here the thermal analysis, thermodynamic analysis, elec-
designer is putting together what is known of the tronics, etc.
function of the work with tentative ideas of form,
material and method of construction. This model also takes into account, within the overall
3. Appraisal of concepts. This stage is where the search- product development process. the framework of plan-
ing eye based on experience becomes invaluable. ning and organization. thus gaining insight into the way
Part B: Journal of Engineering :!anufacture ~!MechE 1996
11. ,,
A SURVEY OF DESIGN PHILOSOPHIES, MODELS, METHODS AND SYSTEMS 311
if products should be designed within a business structure. 2221, the other models also contained in a more
JW The total design activity model is shown in Fig. 7. detailed form within each of their design phases/stages,
)n the design activities that characterized a majority of the
·a! 4.1.14 The BS 7000 design model (57) other models. The Watts model showed only the two
e- ends of the design phase, that is abstract and concrete,
This model commences with a feasibility study stage
with the interval between represented by a cyclic
and proceeds through conceptual design, embodiment
(iterative, refining and progressive) process.
design, detail design and design for manufacture stages.
The models that were based on design acttv1ttes
It also shows the output of each design stage in the
included those by Jones, Marples, Archer, Krick, Cross
form of design brief, concept drawings, layout drawings,
and Harris. It can also be observed that in all of the
detailed product definition and manufacturing instruc-
models, three key activities were predominant, that is
tions respectively. The model ends with a post-design
·e analysis, synthesis and evaluation. Analysis was mostly
support stage. It can be observed that this model
IS related to analysing the design problem, requirements
derives from other models by Pahl and Beitz (33) and
y and specifications. Synthesis was concerned with gener-
French (54), with design for manufacture included as an
·- ating ideas, proposing solutions to large or small design
additional stage. This model is shown schematically in
l) problems as well as exploring the design soiution space,
Fig. 8.
i, while evaluation involved the appraisal of design solu-
g tions in order to establish whether they satisfied the
4.2 A critical appraisal of prescriptive models
e requirements and specifications and set corporate cri-
,f An in-depth review of the prescriptive models on the teria. The sequence in general also tended to be analysis
design process shows that a majority of them based the first, followed by synthesis and then evaluation. In the
i procedural steps of their models on what can be re- model by Krick, synthesis was replaced by search and
garded as design activities (that is analysis, synthesis, evaluation by decision. The model by Harris represent-
evaluation, decisions, etc.), while others based their pro- ed analysis, synthesis and evaluation by appraisal of the
cedural steps on what can be regarded as the phases/ task, conception and appraisal of concepts respectively.
stages of design (that is conceptual design, embodiment It is not surprising that the three activities of analysis,
design and detailed design). The models that were based synthesis and evaluation were predominant as they rep-
on the phases of the design process include those of resent the core of the design process. If proper analysis
Asimow, Pahl and Beitz, VDI 2221, Watts, Hubka and of the problem or requirements is not carried out, syn-
French. With the exception of that of French and VDI thesizing solutions will be difficult and inappropriate
Technology Technique
TOTAL ACTIVITY
UD~DHS CT
S:PCClfJCAT!(l'-j
SYNTHESIS I
CONCEPTUAL
DESIGN
Equa.t~s to Spec.
J .~ECi-.ANISMS
I DETAIL DESIGN
Equ.:a.tPS to S:pN:.
I
OPTIMISATION I
<C_______~DA~T~A_H~AN~D~L~IN~Gi
1 Df~~~~L~~~~L~w~,
SPECIFICATION
C::.~GANISED
Fig. 7 The total design activity model by Pugh
i;' 1!echE 1996 Proc lnstn Mech Engrs Vol 210
13. A SURVEY OF DESIGN PHILOSOPHIES, MODELS, METHODS AND SYSTEMS 313
and (d) checking the fidelity of the final solution to the To achieve robustness, Taguchi suggests the follow-
original needs. ing. sequence of events in .his design model: (a) systenr·
design, (b) par~meter des~gn and (c) tolerance desi~
43.2 Taguchi's quality loss function model System design IS the physical embodiment of the func-
tional requirements of the product, where special engin-
The recent past has witnessed the shift in focus from eering and scientific knowledge is applied. Parameter
on-line quality control to off-line quality. This has led design is the process of identifying the optimal settings
to increasing focus on the integration of quality into the of various parameters under the control of the designer
early design stage of product development Ensuring to limit variation. Tolerance design involves the control
quality by design thus involves the use of structured off- of the variation in critical parameters when everything
line methods to determine the design configurations else has failed to control the variation of performance
that meet the customer's needs and are robust, where within the required limit .
robustness means that product performance character-
istics are insensitive to variation in the manufacturing
and operating environments (60). One of the main pro- 4.4 Descriptive models
ponents of off-line quality control is the renowned
Japanese, Professor Genichi Taguchi, who introduced Descriptive models emanated both from experience of
the concept of 'quality loss' or 'loss to society'. individual designers and from studies carried out on
Taguchi's methodology is based on the precept that the how designs were created, that is what processes, strat-
lowest cost to society represents the product with the egies and problem-solving methods designers used.
highest quality, which is achieved by reducing variation These models usually emphasize the importance of gen-
in product characteristics. This approach is expressed erating one solution concept early in the process, thus
by what is called the 'loss function'. The loss function is reflecting the 'solution focused' nature of design think-
he ing (39). The original solution goes through a process of
of a mathematical way of qualifying cost as a function of
product variation. This loss function allows a determi- analysis, evaluation, refinement (patching and repair)
nation to be made as to whether further reduction in and development (39, 62, 63).
the variation will continue to reduce costs. The loss In their paper, Finger and Dixon (38) discuss descrip-
function includes production costs as well as costs tive models from a different perspective and have identi-
incurred by the customer during use (61). The simplest fied the research work in this area along two main lines:
ItS
form of the loss function is expressed by a quadratic 1. Research based on techniques from artificial intelli-
relationship obtained from a Taylor series expansion, gence such as protocol analysis, involving systematic
ys and can be approximated by: gathering of data on how designers design.
he L(Y) = k(Y- M) 2 2. Research based on modelling the cognitive process.
fie where The aim of this research is to build computer-based
"R L = loss associated with a particular performance cognitive models, which describe, simulate and
characteristic Y emulate the mental processes and skills used by
M = the performance target value designers while creating a design.
k =loss parameter= LJD6
where
4.4.1 Model by March
Lc = average loss to the customer when the per-
ly formance characteristic is not within the limit The model of the design process proposed by March
m Do (64) draws on the work of the American philosopher
D 0 = customer tolerance limit Peirce on the three modes of reasoning, which are
deduction, induction and abduction (production). In
es The loss function L(Y), which is shown graphically in rephrasing Peirce's remarks, rational designing is con-
ns Fig. 9, can thus be defined as the average of the finan- ceived as having three tasks:
er cial loss due to deviations of the product characteristic
e- Y from the target function M over all customer condi- 1. The creation of a novel composition-accomplished
!d tions up to the time required for the product life. by productive reasoning
J.e ?
LCYl = k(Y-M)~
!d
as
!n
:Js
Je
!d
1e
cit
Lm
_'Lc~, l _
"·~
~
~
-i
I
II
I
;h
e-
/1
ss I
is
Ji .~+Do
:li
n. Fig. 9 Taguchi's quality loss function
~ 1le<:hE 1996 Proc lnstn Mech Engrs Vol 210