Design Methodologies in Mechanical Engineering

supported by

Artificial Intelligence Techniques

J. Forster, M. Cartmell and P. Fothergill

Aberdeen University
email: forster, cartmel, pat with %cs.abdn@nsfnet .uk.ac

March 1990

Abstract tions, or processes, humans employ to achieve design

as well as attempting to define concise representations
The following is a description of the work at Aberdeen of the entities involved. There have also been at-
University concerning the application of Artificial In- tempts to construct taxonomies of the various stages,
telligence (AI) techniques in engineering design. such as the work by Reuleau [Reu75] or Redtenbacher
This paper is based on the results of research in [Red521 which, although incomplete, provide a conve-
two fields which have traditionally been treated sepa nient starting point for exploring systematic design.
rately: methodologies in engineering design and arti- The motivation for further research into the de-
ficial intelligence techniques, such as knowledge rep- sign aspect is mainly due to the nature of engineer-
resentation and inference. By attempting to integrate ing, e.g. combining pleasing aesthetics with sound
methodological approaches of specification expansion, physical laws, and the tremendous impact the design
solution generation and design assessment within an phase has on numerous aspects of the final product,
artificial intelligence framework, it should be possible e.g. functionality, reliability, cost, etc. This has to be
to enhance significantly the capabilities of designers. emphasized since the scope and nature of engineering
This paper explores the current state of the art in problems has increased to such an extent that the tra-
both domains and attempts to explain how it may be ditional individualistic approach of: “every designer
possible to merge them in a new and useful manner. has his/her own intuitive approach” fails to provide
Also included are a description of the on-going work the necessary information and communication for a
in Aberdeen and a brief outline of the first computer- large project. According to Rodenacker [PB86], 80%-
ized implementation, the Design Specification Expan- 90% of the costs of a product ark created during the
sion system. design. Therefore, this phase deserves a great deal of
attention.
Several mainly European authors have described
Design Methodologies which use systemized methods
1 Introduction to ensure reliability and quality of design solutions.
Other, mainly American and Japanese, have focused
Engineering design has dramatically evolved during on the cognitive processes of design and have at-
the last century, and there have been many attempts tempted to define the operations that huma.ns em-
to improve not only the artifacts designed but also to ploy when designing. For example, the work of Ull-
gain insights into the design process itself. The work man et al. [UDSSS] and Goel et al. [GP89] have dealt
so far has involved examining both the mental opera- with trying to establish a model of the design process.
Consequently, the American engineering literature is
centered around inventiveness.
These two approaches appear to exist in separate
worlds as there are few in either field who are inter-
Permission to copy without fee all or part of this material is granted ested in the results of the others. This is unfortunate
provided that the copies are not made or distributed for direct since one fraction is attempting to model the design
commercial advantage, the ACM copyright notice and the title of the
publication and its date appear, and notice is given that copying is by
permission of the Association for Computing Machinery. To copy other-
wise, or to republish, requires a fee and/or specific permission.
@ 1990 ACM 089791-372-8/90/0007/0819 $1 SO 819
process while the other is concerning with defining In Forster [For891 several European methodologies
what constitutes a good design process. were presented from Rodenacker, Hubka, Leyer, Pahl
The AI community should, and does, take an in- & Beitz, etc. They all identify certain steps in the
terest in engineering design since it involves complex design process and define information and methods
human cognitive processes and other important issues allocated to these stages. The design process is de-
such as representation (“how it is possible to represent scribed by a ‘process model’ with different levels of
engineering entities and processes”); control (“how complexity. (See the comprehensive approach of Pahl
the represented knowledge should be used to obtain a and Beitz [PB88].)
design?“); and met&planning (“how to plan not only
the artifact but the overall plan itself’). 2.2 Design Process
Until the mid-1980’s the main interest was in
electronic chip design, but recently, people such as Several workers describe the design process in terms
Mostow [Mos85] and Dixon [Dix86] have demon- of a step-by-step flowchart. The differences between
strated that the domain of Mechanical Engineering is these approaches lie in the definition of different
much more demanding than that of VLSI design. This stages, the feed-back loops between the different steps,
is mainly due to the intensely problem-specific interre- and the roles of certain elements such as abstraction
lationships among geometry, function, material, aes- or form-design.
thetics, etc. For example, Rodenacker [Rod761 emphasizes the
physical side of the problem: the function or the pur-
pose of the artifact. The design process is seen as a
transformation of information from the abstract to the
2 Engineering Methodologies concrete. According to him, the key is to find an ap-
propriate physical effect. His ideas are therefore useful
2.1 Introduction not only for the development of the artifact, but also
for supporting invention of new machines. He intro-
The approaches to methodological design have a num- duces a function structure to develop the relationship
ber of aims, which can be summarized as follows: between the required effect and the designed artifact.
This utilizes Boolean logic in a persistent attempt to
optimizing over-all performance
suppress irritating parameters.
providing guidance to the inexperienced designer Although Rodenacker restricts himself to examin-
ing the artifact, or rather the generation of its de-
reducing the omission of particular points scription, he still recognizes the importance of these
points:
requiring the designer to make and record design
decisions organization/management of the design office
(very important for re-design)
psychological aspects such as creativity, syner-
facilitating communication within the design gism, etc.
team
issues regarding automation
A great number of authors have made contribu-
tions to the field of Engineering Design Methodolo-
One of his more important claims is that the design
gies. This paper focuses on a few of the more impor-
proce?s is “the reverse of the physical experiment.”
tant works.
([RC76] He gives three basic rules for d’esign: ’
These methods have the common aim of trying to
produce an ‘optimal’ design. To achieve this, one can 1. first qualitative and then quantitative issues
follow two paths: All of these methods share a com- should be considered
mon aim in trying to produce ‘optimal’ design. To
achieve this optimal design, one can follow 2. ‘irritating effects’ should be recognized and at-
tempts made to suppress them
1. to define a procedure which automatically leads
to the optimal design; 3. requirements of the design process should be
adapted from the relations between the system
2. to generate a solution field and then assess the ‘machine’ and the system’s personnel, environ-
derived solutions. ment, budget, etc.

820
 3. determine weak points in every solution and try
But he also sees problems with the introduction of to minimize them
methodologies into engineering design, namely that
one has to: 4. find the solution with the lowest number of weak
points
l ignore personal intuitive work-styles
5. produce documentation to enable practical usage
l apply methodological steps tigorously
He then defines four distinct stages of development.
0 clarify logical interrelations (See table 1).

l carefully investigate the possible physical imple-

mentations

l follow the principle ‘go from the simple to the

complex’

3. Improved Working-Principle
He states that the advantages of employing his
methodology are:

l clear problem specification Table 1: Four development stages after Hansen

0 clear presentation of necessary information Hansen defines the overall process as a ‘develop-
ment’ that can be divided into a conceptual phase
l simplification of difficult tasks and an actual design in terms of work on the drawing
l improved quality of information board, assigning tolerances and exploring geometric
relationships.
In contrast to this, Leyer’s [Ley63] work is more Pahl and Beitz [PB88] define several stages in the
concerned with the ‘form-design’ or embodiment of design steps which progress through several defined
the design. Therefore his work gives first place to stages to the end-product:
‘form-design principles’ and ‘form-design guidance’.
l Clarification of the Task
He defines three phases:
l Conceptual design
1. determining the solution principle by idea or us-
ing known principles l Embodiment design

2. carrying out the actual drawing (kernel design) l Detail design.

3. defining the final layout/embodiment

The clarification expands and establishes the clear
He sees the second step as supporting the final one specification and should provide the next steps with
by means of calculations and starting from a ‘place enough constraints and evaluation criteria.
of e$ect’ where the actual function of the device is For the conceptual phase, Pahl and Beitz empha-
allocated. The final step is supported by guiding the sis the use of function structures (both general and
work on the drawing board with some principles, such boolean) to represent the necessary concept. For the
as the one of constant thickness of walls, homogenous concept development, Pahl et al. looks at meth-
design and a ‘force-flow’ adequate design. ods with intuitive and discursive character. As in-
Another advocate of approaching the problem in a tuitive methods they discuss the established method
series of conceptual stages is Hansens. His approach of Brainstorming, Method 6-3-5 or analogy reasoning
[Ban661 consists of two different parts: First he defines to develop solutions.
five ‘main directives’ l: As discursive methods, Pahl et al. mainly discuss
the systematic study of the underlying physical pro-
1. define the kernel of the problem as the center of cess (parameter-variation) or by a systematic classi-
all possible solutions fication scheme (design catalogs), which are ordered
by criteria, such as types of energy or physical effect.
2. combine parts or principles to achieve the goal
In turn they define criteria to evaluate the concept
‘in German ‘Hauptrichtlinien’ variants, which are generated by the combination of

821
different solution principles: Pahl and Beitz stress the of the inputs and outputs and the function of the de-
necessity for generating a large solution field, in order sign linking these. (see figure 1.)
to be able to make a optimal solution field. This large
solution field has to be assessed in order to find the
most promising sub-functions, not in terms of a good
single or group of sub-functions, but of the overall
solution.
The criteria Pahl et al. define are presented in ta-
ble 2. These criteria are often grouped into solution
charts, to ease and automize the decision process (see
page 113 in [PB88]). The criteria are not equally easy
to determine: the qualitative aspects are preferred be-
cause they are straightforward to assess whereas from
C,D to E,F the assessment becomes more and more
difficult. This is especially true if several of these so-
1
IZfects on enviromnent
lutions pass the ‘negative selection’ stage, but need
to be differentiated as to how well they perform in a I
$ ,,
certain aspect over their rivals. :
TiUOl de Temperature
l . .‘., : >
RPM
j..
qualitative (yer/no)
:,,.: .:..
qurntitrtive

qurntitrtive
Mechanical
>
Energy

Petrol IF
-‘,
Table 2: Evaluation criteria and there characteristics Air fumes

Other evaluation methods they mention are:

k?
l cost analysis / Value analysis (VA)
l use-value-analysis (UVA)
l objective tree
Figure 1: The black box with its Input/Outputs
Regarding the next design step, embodiment de-
sign, they list the basic rules, such as clarity, simplic- The work in function structure has been encouraged
ity, safety, etc., and also define several other special- by a number of workers (Rodenacker [Rod76], Roth
ited principles: [Rot82], VDI-guideline ’ [VDHl] and [VDISS], Pahl
l Principle of force transmission and Beitz [PBSS]).
The separation of a designed artifact into function
l Principle of division of tasks structures is important for the decomposition of the
l Principle of self-help overall problem into sub-problems Eve:n though, de-
signers do not necessarily work this way, a determina-
In turn, they also define more specific guidelines tion of a functional hierarchy is important to enable
for the embodiment phase. These incorporate aspects the methodology to generate a solution field for a par-
which are generally useful for every design, but can ticular aspect of the design. These function structures
be emphasized in one or the other, such as design for do not necessarily remain static, but require a means
corrosion, for standards, ease of assembly, etc. of communicating the specifications a.nd constraints
relevant to them and their descendants.
2.3 Function-Structures These function structures allow the identification of
common sub-problems and allow solutions for a cer-
An important issue in several works is the transforms
tion of the over-all design into a more abstract repre- 2Verein Deutscher Ingeuieure- Association of German
sentation. This ‘black-box’ can be described in terms Engineers

822
tain sub-problem to be shared over a wide range of means that their signal, material and energy inputs
designs. Pahl et al. differentiate between different must connect. The set of all connection solutions is
kinds of Gesigns: in the original design the required the overall solution field. The different sub-solutions
or projected function structure is derived from the are then ranked and compared. This is a necessary
specification whereas in the adaptive design it comes operation in order to prune the number of solutions
from an existing device. early. (The use of design catalogs and fine/detailed
Re-useability is one of the aims of the methodologi- function-structures can generate combinatorial expIo-
cal approach so consequently Roth [Rot821 argues for sions.)
the use of ‘design-catalogs’ (For different types of cat-
alogs see [DL76].)
These catdogs list general functions, materials and 2.5 Is systematic design possible ?
information about machine parts along with their ba- Methodological design has been regularly attacked by
sic equations and properties. Roth emphasizes that those who claim it is impossible to channel inven-
these catalogs must be very complete. He advocates tiveness and creativity, (French [Fre88]); those who
the implementation of such catalogs in computer sys- emphasize the importance of Form Design, (Leyer
tems. [Ley63] and Schwarzkopf et al [SJ84]); and/or those
It has often been said that the effort involved in who assume it is impossible to derive function struc-
systematic design, such as deriving a function struc- tures.
ture, is only worthwhile for mass production, but in However, there have been successful applications
Wiendahl [Wie’?8] examples for the standardization of of Design Methodologies published, e.g. [DHVS83a],
‘function carriers’, (machine elements which implement [DHVS84], [DHVS83b]; along with reports from prac-
basic functions} have been shown which are especially titioners who employ these methodologies in their
important for ‘one-off’ manufacturing. Since the at- daily work. Furthermore, the application of a
tempts to rationalize this type of design by simple methodology can actually stimulate creativity by
classification schemes and pre-printed drawings has leaving the designer free to work on a more concep-
failed, the use of methodological design may provide tual level. This also makes acquired experience more
a solution. transparent and adaptable (Pahl [Pah82].) It has also
Another aspect is the use of function structures to been shown by Ehrlenspiegel et al [EJ87] that design
arrive at a simplified design. In other words, starting methodologies can aid the invention of new entities by
with the designed product, it is difficult, to achieve a providing a personal problem solving method which is
noteworthy simplification of the design. Rodenacker tailored for new approaches to an existing problem.
[RB76] proposes the use of operations of integration The comprehensive work of Pahl et al. unites very
to combine different functions and reduction to re- different approaches and puts them on a common
move aspects of functions in function-structures. An ground e.g. the function-structures of Keller [1<0176]
example of this would be the joining of functions to and Rodenacker [Rod76]: The use of the guided ex-
perform measurement with actuators as combinations pansion of specifications and the compilation of as-
of the measurement of the water and the valve in a
water reservoir.

2.4 Morphological-Structures
Zwicky’s morphological matric [Zwi76] is proposed for
the presentation and evaluation stages. The morpho-
logical matrix represents the field of solutions in terms
of solutions and sub-problems. The overall solution is
presented in terms of the optimized, or legal, combi-
nation of different compatible sub-solutions together
with their technical and economic evaluations. (See
figure 2). Overall solution
This morphological matrix is principally used to vi-
sualize the idea of re-combining the generated sub-
solutions whether they are created by intuitive means,
design catalogs or other methods. To combine them, Figure 2: The Morphological Matrix according to
the different solutions must be compatible. This Zwicky.

823
sessment criteria; morphological matrix and the em- 4 AI and Methodologies
phasis on from design in the embodiment phase.
The positive effects of a methodological proce- Methodologies themselves lead naturally to the use
dure in terms of the quality of both the design and of computers. Examining them, it is true that they
the artifact, and the step-by-step support provided are generally aimed at computers. AI can provide the
by computer-systems makes the effort of introducing tools which are necessary to implement them. This
them into the daily design process worthwhile. means not only means that activities such as geo-
metrical reasoning and constraint management can
be supported by AI, but also that an overall model
of systematic design can be implemented. In general
3 AI and Engineering Design terms, the methodologies identify the knowledge in-
volved, the operations which can be applied to the
3.1 Introduction entities, and the possible order in which t.he ‘design’
can take place. Therefore, methodologies describe a
The AI world began to take an interest in the field of
form of ‘meta-design’.
Design during the 80’s. It is necessary to examine the
different computerized approaches to design in order Because a sizeable amount of research has already
to identify promising approaches for AI to support been done regarding the support of certain design ac-
methodological design. Therefore, a taxonomy has to tivities rather than the design process as a whole,
be developed in order to classify the approaches: methodologies could be implemented in terms of this
‘met&view’ sitting on top of an existing design sup-
l expert-systems which try to automate the design port system. This is, of course, provided the existing
or selection of specific parts system has the necessary internals, both conceptually
and code-wise, to allow a system to examine its inter-
l systems which attempt to represent the human nal entities and either modify or direct them.
design methods and automatically design new de- On a more basic level, it can be seen that the
vices (such as EDISON (Deyer et al. [DFH86]) methodological framework maps onto techniques in
which tries to design by applying analogy reason- the AI world, implementing and enhancing the ideas
ing). of systematic design.
l design supporting systems (DSS - Design Sup-
port Systems like the Edinburgh Designer Sys- The morphological matrix is identified as an ap-
tem - EDS by Smithers et.al. [PSCM86]) which plication of a truth maintenance system (TMS)
are intended to support the engineer with a com- in which each design sub-solution is associated
puterized , integrated tool to cope with design with a set of asserted facts. Successful solutions
(integrated toolbox) of the overall design are presented by combina-
tions of sub-solutions in which all the facts are
The assessment of the current AI tools (DDS) re- compatible. Illegal combinations lead to contra-
veals shortcomings such as the representation of con- dictions which ‘poison’ the system and reduce the
cepts where inheritance and exceptions or persistence combinatorial effects. The determination of de-
of the structure of engineering objects throughout the pendencies in a complex interacting design can
design cycle. The definition of the components of an be supported by a TMS by comparison of the
AI-based design system must be done in terms of the different justifications in designed evolving from
required functionality in order to carry out design. different assumptions.
This involves the information the designer is manipu-
lating and the elements of the design process as have The systematic evaluation of the design specifi-
been outlined by theories of methodological design. cations and the compatibly of the concepts en-
A practical example of how ideas from methodolog- courage both the use of a production system to
ical design can be supported by AI-techniques is the guide the design expansion and the combination
morphological matrix which represents the field of so- of solution variants. The guiding rule-base ‘looks
lution to sub-problems. The overall solutions are the over the designers’ shoulder suggesting, prompt-
best combinations of sub-solutions which necessitate ing and flagging him during the expansion pro-
a truth-maintenance system to record the combina cess. This implements the idea of a meta-tool
tions and their assessed validity, together with any which directs the use of other tools and acts as
justification for apparent failure. conceptual work-bench, rather than just a com-
mon place to exchange data for design.

824
 l The use of function structures can facilitate the tern. This tool should help optimize the design ex-
qualitative assessment and evaluation of concepts pansion, provide the designer with guidance, and en-
on a higher level. This points in the direction able comparisons between different expansions of the
of predicting the qualitative behavior of function same design. The system is implemented in a hybrid
structures and recording their connection to the environment on Sun workstations (Sun-S/Spare). It
overall specification and its linked neighbors. consists of a Hypertext implementation, a language
pre-processing system, (for chunking the specification
into its constituents), an underlying system to record
the expansion, and a control system.
We plan to use the results of the theoretical work to
5 On-going work introduce a production system to ‘look over the shoul-
The current work in Aberdeen is concerned with the der’ of the designer and suggest to him appropriate
expansion of design specifications for the following activities such as elaboration or the introduction of
reasons: specification points.
It is intended that our implementation should even-
l Result of examination of design methodologies tually provide a useful design expansion support tool.
- Specification expansion and assessment is the We shall therefore need to link it to a drawing tool
first step in the design process. It has tremen- for graphical annotations and a spread-sheet for quick
dous impact upon later design steps; consuming calculations. Since useability is a strong point, we
approximately 5% of resources, but determining are concerned with issues such as speed, stability and
90% of the later product cost [BurSO]. Further- portability.
more, all ensuing steps in a systematic design re- The eventual system should provide:
fer to the specification since the assessment cri-
teria are compiled from this. Even though the l a useful support tool for the designer
generation of solutions is possible, without care- l a vehicle to test our ideas for design expan-
ful analysis of the specification, it is difficult to sion, especially the procedural side of following
rank them. a checklist of using a guiding rule-base.
l Examination of different Design Support Systems l a method of performing knowledge acquisition on
(principally the Edinburgh Designer System) - how designers actually perform the ‘design’ ex-
The support Systems recognize no conceptual dif- pansion.
ferences between the various forms of constraints
with which the designer is working. It is only This implementation should not only form a use-
by interpretation that the meaning of the con- ful experimental tool for our investigations of initial
straints becomes clear. However, there are nu- design expansion, but it should also raise some inter-
merous types of constraints. Some are given by esting practical questions. We will need to integrate a
the initial specification; some are derived from natural language system for chunking the input spec-
the specification; and others are introduced by ification with the Hypertext system. This in turn will
the designer during the design process. A design be linked to an expert system which will interact with
support tool should be capable of differentiating the designer - suggesting lines of thought, recording
these for the user. decisions, and keeping track of the different types of
constraints.
Our work is divided into the theoretical and the
practical. The theoretical side examines the design
expansions as they are produced by people using
the expansion schemes described by the methodol- 6 Future Work
ogy work. In this study, we are using approximately
twenty-five initial design specifications from a variety The project will then advance to examining the dif-
of areas. Once the specifications have been expanded, ferent ways in which the evaluation criteria can be
we attempt to determine the links and flow of focus compiled. These two steps will then lead to incor-
between the different aspects. We hope to be able porating systematic design approaches into a DDS
to derive ‘rules’ for expansion paths to guide the de- such as the EDS. We also plan to investigate linking
signer. the ideas of design (conceptual, function-carrier, as-
The theoretical work is closely accompanied by a sembly/manufacturing oriented) and ‘morphological-
practical implementation of a design specification sys- matrix’ into an existing or virtual DSS. We hope we

825
can prove the usefulness of this work by case studies [DL76] G.W. Diekhoerner and F. Lohkamp.
with our industrial collaborators in the oil industry. Objektkataloge-hilfsmittel beim method-
It is our aim to generate software which will not just ischen konstruieren. Konstruktion, pages
validate our ideas, but also provide some degree of 359-364, 1976.
practical usefulness.
[EJ87] K. Ehrlenspiegel and T. John. Inventing
by design methodology, 1987.

Acknowledgments [For 891 J. Forster. Artificial intelligence and de-

sign methodologies in mechanical engi-
neering. Technical report, University of
We would like to thank: Aberdeen, 1989.
SERC (ACME) for financial support
[Fre88] M.J. French. Inventention and evolution,
(grant GR/F/5643.7); GEC Electrical Projects Ltd
Design in Nature and Engineering. Cam-
and Ewbank Preece Ltd for the collaboration in the
bridge University Press, 1988.
project; Andy Robertson of Lucas CAV for his help
with the EDS; Pamela Van Nest for her work on the [GP89] Vinod Goel and Peter Pirolli. Motivat-
language pre-processing system. ing the notion of generic design with in-
formation processing theory: The design
problem space. AI magazine, pages 19-
References 36, 1989.

[BurSO] L,D. Burrow. Personal comments during [Han661 F. Hansen. Konstruktionssystematik.

the dtop public demonstation. Influence VEB Verl. Technik, Berlin, 1966.
of specifications, March 1990.
[Ko176] R. Keller. Konstruktionsmethode juer
[DFH86] M. Deyer, M. Flowers, and J. Hodges. den Machin,ene-, Geraete und Apparate-
Edison, 1986. bau. Springer-Verlag, 1976.

[LeyW A. Leyer. Machinen-Konstruktionslehre.

[DHVS83a] G.W. Diekhoener, H.Birkhofer H.-O. Ho-
Birkhaeser Verlag, 1963.
efller, A. Voeglel, and K.-H. Sachs. Kon-
struktionsmethodik in mittlestaendis- [Mos85] J. Mostow. Towards better models of the
then unternehme: Entwicklung eines design process. AI magazine, pages 44-
antries fuer einen neuartigen behand- 56, 1985.
lungsstuhl. Konstruktion, pages 381-385,
1983. [Pah82] G. Pahl. Foerderung der kreativitaet im
grundlagenfach machinenelemente durch
[DHVS83b] G.W. Diekhoener, H.Birkhofer H.-O. Ho methodisches konstruieren. Konstruk-
effler, A. Voeglel, and K.-H. Sachs. Kon- tion, pages 313-319, 1982.
struktionsmethodik in mittlestaendis-
then unternehmen: Entwicklung eines [PB86] G. Pahl and W. Beitz. Konstruktion-
neuen schliess - und sperrsystems fuer slehre. Springer Verlag, 1986.
schubladenschraenke. Konstruktion,
G. Pahl and W. Beitz. Engineering De-
[PB88]
pages 270-283, 1983. sign (Konstruktionslehre). Springer Ver-
lag, 1977/1988.
[DHVS84] G.W. Diekhoener, H.Birkhofer H.-O. Ho-
effler, A. Voeglel, and K.-H. Sachs. Kon- [PSCM86] R. Popplestone, T. SmithLers, J. Corney,
struktionsmethodik in mittlestaendis- and K. Millington. Engineering design
then unternehmen: Entwicklung eines support systems. Technical report, Uni-
gleichlaufs zwangsmischers. Konstruk- versity of Edinburgh, 1986.
lion, pages 65-69, 1984.
[RB76] W.G. Rodenacker and R. Baumgarth.
[Dix86] J. Dixon. Artificial intelligence and de- Die vereinfachung der geraete beginnt
sign: A mechanical engineering view, mit der functionsstruktur. Konstruktion,
1986. pages 479482, 1976.

826
[RC76] W.G. Rodenacker and U. Clausen.
Regeln des Methodisches Konstruieren.
Krauskopf Verlag, Mainz, 1976.

[Red521 F. Redtenbacher. Printipien der

Mechanik und des Machinenbaus. Basser-
man Verlag, Manheim, 1852.

[I&X75] F. Reuleaux. Theoretische Kinematik.

Vieweg Verlag, Braunschweig, 1875.

[Rod761 W.G. Rodenacker. Methodisches Kon-

struieren. Springer Verlag, 1976.

[R&2] K. Roth. Konslruieren mii Konstruktion-

skatalogen. Springer Verlag, Berlin, 1982.

[SJ84] W. Schwarzkopf and W. Jordan. Die

gestaltung - stiefkiond der konstruktion-
smethodik ? Konstruktion, pages 299-
304, 1984.

[uDSS~] D.G. Ullman, T.G. Dietterich, and L.A.

Stauffer. A model of the mechanical de-
sign proceess based on empirical data,
1988.

[VDI81] VDI. VDI-Guideline 2212, Systematic

search for and optimisation of Engineer-
ing Designs. VDI-Verlag, Duesseldorf,
1981.
[VD185] VDI. VDI-Guideline 2221, Systematic
engineen’ng design of technical systems
and products. VDI-Verlag, Duesseldorf,
1985.

[Wie78] H.-P. Wiendahl. Funktionale standard-

isierung
- ein konzept zum rationalisieren in der
machinenbau-einzelfertigung. Konstruk-
2ion, pages 221-227, 1978.

[Zwi76] F. Zwicky. Entdecken, Erfinden,

Forschen im Morphologischen Weltbild.
Droemer-Knauer, Munich/Zuerich, 1976.

827