design for adaptability - eCommons@USASK
A Thesis Submitted to the College of
Graduate Studies and Research
in Partial Fulfillment of the Requirements
for the Degree of Doctor of Philosophy
in the Department of Mechanical Engineering
University of Saskatchewan
Mehdi Hashemian
© Copyright Mehdi Hashemian, Jun 2005. All rights reserved.
In presenting this thesis in partial fulfillment of the requirements for a Postgraduate
degree from the University of Saskatchewan, I agree that the Libraries of this University
may make it freely available for inspection. I further agree that permission for copying
of this thesis in any manner, in whole or in part, for scholarly purposes may be granted
by the professor or professors who supervised my thesis work or, in their absence, by
the Head of the Department or the Dean of the College in which my thesis work was
done. It is understood that any copying or publication or use of this thesis or parts
thereof for financial gain shall not be allowed without my written permission. It is also
understood that due recognition shall be given to me and to the University of
Saskatchewan in any scholarly use which may be made of any material in my thesis.
Requests for permission to copy or to make other use of material in this thesis in whole
or part should be addressed to:
Head of the Department of Mechanical Engineering
University of Saskatchewan
57 Campus Drive
Saskatoon, Saskatchewan, Canada, S7N 5A9
Manufacturing globalization and sustainable development compel production
enterprises to continuously seek improvements in their products’ performance,
customization, environmental friendliness, cost, and delivery time. The challenges of
this competition cannot be completely addressed through improving production
processes because some issues can only be solved through more innovative ‘design’.
This thesis investigates a new design paradigm called Design for Adaptability or
Adaptable Design (AD) to address some of these challenges.
The purpose of AD is to extend the utility of designs and products. An adaptable
‘design’ allows manufacturers to quickly develop new and upgraded models or
customized products through adapting existing designs with proven quality and costs.
An adaptable ‘product’ can be utilized under varying service requirements thus prevents
premature product replacement. Design adaptability and product adaptability provide
economical and environmental benefits for AD.
To make a product adaptable, its adaptability must be built-in during the design stage.
Methods of design for ‘predetermined’ adaptations are categorized as Specific AD; these
methods design products for versatility, upgrading, variety, and customization. Several
of these methods such as modular/platform design and design for upgrading have been
studied for mechanical design. In the absence of predetermined adaptations, AD aims to
increase the general adaptability of products. General AD involves fundamental
research in design theory and methodology in order to develop practical design methods
and guidelines. This thesis introduces several original concepts and proposes the
subordination of a system to a rational functional structure as an approach for
increasing general adaptability. Such a system would consist of a hierarchical assembly
of autonomous functional modules, emulating the adaptable architecture of a ‘rational
functional structure’. Methods and guidelines are proposed for making the design of
mechanical systems closer to this ideal architecture.
Accordingly, the thesis proposes a methodology for AD in which specific AD is
performed first to take advantage of available ‘forecast’ information, and then general
AD is performed in order to increase adaptability to ‘unforeseen’ changes. Also, a
measure has been defined for the assessment of adaptability. The application of this
methodology has been demonstrated through several conceptual design examples.
I express my most sincere gratitude to my supervisor Professor P. Gu for his guidance
and inspiring ideas during my studies, also for his encouragement, advice, and support
after I returned to the PhD program. This thesis would not be possible without him. I am
grateful to my co-supervisor Professor G. Watson and to the members of the advisory
committee at the University of Saskatchewan. I am also indebted to my friends and
colleagues at the University of Calgary who shared their thoughts about engineering
design research during many discussions and meetings, particularly to Mr. Neil
Schemenauer who also programmed the software for the representation of functions in
this thesis.
Financial support from the Natural Sciences and Engineering Research Council of
Canada (NSERC) and the academic scholarship from the University of Saskatchewan
are acknowledged and appreciated.
Dedicated to my wife Ania
and to our son Hassan.
PERMISSION TO USE ................................................................................................... ii
ABSTRACT .................................................................................................................... iii
ACKNOWLEDGEMENTS ............................................................................................. v
DEDICATION ................................................................................................................ vi
TABLE OF CONTENTS ............................................................................................... vii
LIST OF TABLES ........................................................................................................... x
LIST OF FIGURES......................................................................................................... xi
Chapter 1: Introduction..................................................................................................... 1
1.1. Background............................................................................................................ 2
1.2. Segmentation (Modularization)............................................................................. 8
1.3. Modular Design and Adaptable Design ................................................................ 9
1.4. Thesis Overview.................................................................................................. 11
1.5. Thesis Objectives................................................................................................. 17
1.6. Organization of This Thesis ................................................................................ 17
1.7. Terms and Definitions ......................................................................................... 18
Chapter 2: Literature Review ......................................................................................... 21
2.1. Theoretical Engineering Design Research .......................................................... 21
2.1.1. Descriptive (Cognitive) Models ................................................................... 22
2.1.2. Synthesis and TRIZ ...................................................................................... 24
2.1.3. Axiomatic Design......................................................................................... 31
2.1.4. Systematic Design ........................................................................................ 34
2.1.5. Knowledge-Based Design ............................................................................ 36
2.1.6. Decision-Based Design ................................................................................ 38
2.1.7. The General Design Theory ......................................................................... 40
2.2. Review of Product Configuration Design Research............................................ 42
2.2.1. Functional and Physical Structures .............................................................. 42
2.2.2. Modular Design ............................................................................................ 45
2.2.3. Product Family and Platform Design ........................................................... 51
2.2.4. Life Cycle Objectives of Modularity............................................................ 54
2.3. Discussion............................................................................................................ 59
Chapter 3: Adaptability in Designs and Products........................................................... 61
3.1. Extension of Utility ............................................................................................. 61
3.1.1. Service-Based Economy............................................................................... 62
3.1.2. Extending Utility through Adaptation .......................................................... 64
3.1.3. Discussion..................................................................................................... 66
3.2. Categories of Adaptabilities ................................................................................ 67
3.2.1. Design Adaptability and Product Adaptability............................................. 68
3.2.2. Sequential and Parallel Adaptations............................................................. 71
3.2.3. Specific and General Adaptabilities ............................................................. 73
3.2.4. Summary....................................................................................................... 76
3.3. Design Categories Suitable for AD ..................................................................... 77
3.4. Benefits of Adaptability ...................................................................................... 80
3.4.1. The User: Extended Product Utility ............................................................. 80
3.4.2. The Producer: Extended Design Utility ....................................................... 81
3.4.3. The Environment .......................................................................................... 82
3.5. Summary.............................................................................................................. 83
Chapter 4: Design for Adaptability ................................................................................ 85
4.1. Fundamentals of Design for General Adaptability.............................................. 85
4.1.1. The Design Hierarchy................................................................................... 86
4.1.2. Decomposition and the Design Holon.......................................................... 89
4.1.3. The Rational Functional Structure................................................................ 91
4.1.4. Causality and adaptability ............................................................................ 93
4.1.5. General Adaptability through Subordination ............................................... 93
4.2. The Challenge of Mechanical Design ................................................................. 94
4.3. Measure of Adaptability .................................................................................... 100
4.3.1. The Information Content ............................................................................ 101
4.3.2. General Measure of Adaptability of a Product........................................... 106
4.3.3. Physical States and IC of Adaptation ......................................................... 108
4.3.4. Calculation of Adaptability ........................................................................ 112
4.3.5. Implications of the Adaptability Equation ................................................. 115
4.4. Methods and Guidelines .................................................................................... 116
4.4.1. Specific AD ................................................................................................ 116
4.4.2. General AD................................................................................................. 119
4.5. Adaptable Design Methodology........................................................................ 128
4.6. AD and Other Life-Cycle Design Goals ........................................................... 130
4.7. Summary............................................................................................................ 132
Chapter 5: Examples..................................................................................................... 133
5.1. Specific AD: Versatile Bicycle Accessories ..................................................... 133
5.1.1. The Design Process .................................................................................... 134
5.1.2. Discussion................................................................................................... 139
5.1.3. Other Examples of Specific AD ................................................................. 139
5.2. Examples of Design for General Adaptability .................................................. 140
5.2.1. The Adaptable Design of a Hydraulic Jack................................................ 141
5.2.2. The Adaptable Design of a Vehicle............................................................ 146
5.3. A Comparative Example ................................................................................... 164
5.3.1. A Versatile Home and Garden Tool ........................................................... 164
5.3.2. The General AD of Home and Garden Tools............................................. 171
5.4. Calculation of Adaptability in General AD....................................................... 174
Chapter 6: Summary and Discussions.......................................................................... 175
6.1. Thesis Summary ................................................................................................ 175
6.2. Discussions ........................................................................................................ 178
6.2.1. Function-Based Modularization ................................................................. 179
6.2.2. Information Content ................................................................................... 182
6.2.3. Justification of Design for Adaptability ..................................................... 184
Chapter 7: Conclusion .................................................................................................. 186
7.1. Conclusions of This Research ........................................................................... 186
7.2. Contributions ..................................................................................................... 188
7.3. Future Work....................................................................................................... 189
References .................................................................................................................... 191
Appendix 1: A Function Representation Scheme for Conceptual Mechanical Design 211
A1.1. Function Operands (Physical Entities) ........................................................... 212
A1.2. Actions............................................................................................................ 219
A1.3. The Structure of a Function ............................................................................ 222
A1.4. Examples ........................................................................................................ 225
A1.5. Software Implementation ............................................................................... 228
A1.5.1. Operands.................................................................................................. 228
A1.5.2. Actions..................................................................................................... 230
A1.5.3. Building Larger Functions....................................................................... 232
Appendix 2: Four-Wheel Servo Steering ..................................................................... 236
A2.1. Steering Axis Offset ....................................................................................... 236
A2.2. Unlimited Steering.......................................................................................... 238
A2.3. Adjustable Sensitivity..................................................................................... 239
A2.4. Wheel Alignment............................................................................................ 240
Appendix 3. Task and Information Processing ............................................................ 242
Table 3. 1: The relationships among various aspects of Adaptable Design................... 84
Table 6. 1: Highlights and organization of the thesis................................................... 178
Table A1. 1: The list of operand attributes................................................................... 218
Table A1. 2: The list of action specifiers. .................................................................... 221
Figure 1. 1: Modular Design and Adaptable Design...................................................... 10
Figure 2. 1: The functional structure. ............................................................................. 44
Figure 3. 1: AD within the spectrum of environmental approaches............................... 66
Figure 3. 2: The relation between scarcity of resources and the need for adaptability. . 67
Figure 3. 3: Design Adaptability and Product Adaptability. .......................................... 68
Figure 3. 4: An example of design adaptability (courtesy of SONY). ........................... 70
Figure 3. 5: Adaptable homes (courtesy of the City of Vancouver). ............................. 71
Figure 3. 6: An example of sequential design adaptability. ........................................... 72
Figure 3. 7: Parallel design adaptability, performed by the manufacturer, results in
variety and mass customization (Courtesy of Ford Motor Company). .................. 73
Figure 3. 8: The relations between the availability of specific information and the
justification of initial investment in the four categories of specific AD. ............... 76
Figure 3. 9: Various categories of adaptations. .............................................................. 77
Figure 3. 10: Product adaptability, performed by the user, results in multi-purpose
versatile machines. (Courtesy of Master Lock)...................................................... 81
Figure 3. 11: Adaptation versus post-retirement remedies............................................. 83
Figure 4. 1: The design holon consists of FRs, solutions, and decomposition............... 91
Figure 4. 2: The rational functional structure................................................................. 92
Figure 4. 3: Mechanical systems are generally less adaptable than other engineering
systems. .................................................................................................................. 99
Figure 4. 4: Functional and physical structures may not correspond. .......................... 100
Figure 4. 5: The Ideal and Actual States. ..................................................................... 110
Figure 4. 6: IS2 for the truck example.......................................................................... 111
Figure 4. 7: AS2 for the truck example. ....................................................................... 111
Figure 4. 8: IMIC for the pump example. .................................................................... 112
Figure 4. 9: The replacement of mechanical systems by "soft" electro-mechanical
systems. ................................................................................................................ 122
Figure 4. 10: The adaptable design of a lock................................................................ 125
Figure 4. 11: Variety through the possibility of morphological combination. ............. 126
Figure 4. 12: The applicability of adaptable design and other life cycle design in the
design process....................................................................................................... 131
Figure 5. 1: The functional structures of a carrier rack and a splashguard. ................. 135
Figure 5.2: The functional structure of a U-Lock......................................................... 136
Figure 5.3: The design of an adaptable bicycle rack. ................................................... 138
Figure 5.4: Conceptual designs for a manual force amplifying device........................ 142
Figure 5.5: The design of a double-speed manual hydraulic pump. ............................ 144
Figure 5. 6: An adaptable design and two conventional designs for hydraulic jacks. . 145
Figure 5. 7: A functional structure for a vehicle. ......................................................... 148
Figure 5. 8: The operation of a wheel motor (picture courtesy of WaveCrest Co.)..... 150
Figure 5. 9: Electric wheels designed as independent functional modules. ................. 151
Figure 5. 10: Seats for the function of 'positioning passengers'. .................................. 152
Figure 5. 11: The conventional steering system. .......................................................... 153
Figure 5. 12: By-wire steering. (Picture courtesy of SKF)........................................... 154
Figure 5. 13: The driver steering module in AUTONOMY (Courtesy of GM)........... 154
Figure 5. 14: Increasing the general adaptability of steering systems.......................... 156
Figure 5. 15: Calculating rotation angles for the right and left wheels. ....................... 157
Figure 5. 16: Modular battery cells. ............................................................................. 158
Figure 5. 17: The space frame elements designed for this example............................. 159
Figure 5. 18: A space frame chassis. ............................................................................ 160
Figure 5. 19: Adaptable car configurations. ................................................................. 161
Figure 5. 20: Other types of vehicles that utilize the functional modules. ................... 162
Figure 5. 21: One-to-one correspondence between the functional and physical structures
of the proposed design.......................................................................................... 163
Figure 5.22: Common functions among chainsaws, trimmers, and edgers.................. 165
Figure 5.23: The functional structures of the chainsaw, the hedge trimmer, and the edger.
.............................................................................................................................. 167
Figure 5.24: An adaptable design consisting of a platform, three modules, and proper
interfaces............................................................................................................... 168
Figure 5. 25: An adaptable electric motor designed for the chainsaw. ........................ 172
Figure 5. 26: The chainsaw module. ............................................................................ 172
Figure 5. 27: Adaptable designs for power tools.......................................................... 173
Figure 6. 1: Both quality and probability should be used in the evaluation of a design.
.............................................................................................................................. 184
Figure A1. 1: The hierarchical taxonomy of function operands. ................................. 215
Figure A1.2: The taxonomy of actions......................................................................... 220
Figure A1.3: The constituting elements of function operands and actions. ................. 222
Figure A1.4: Elements of a basic function in the proposed scheme. ........................... 223
Figure A1.5: Construction of functions in the proposed scheme. ................................ 224
Figure A1.6: Representing the function of a thermometer........................................... 226
Figure A1.7: Representing the function of a controller................................................ 227
Figure A1.8: Representing the function of a shaft. ...................................................... 227
Figure A1.9: Specifying the types of function operands.............................................. 229
Figure A1.10: Specifying the relevant attributes of a function operand (solid, material).
.............................................................................................................................. 230
Figure A1.11: Different "types" of actions in TARRAUH. ......................................... 231
Figure A1.12: Quantifying "action specifiers" for the actions of known types. .......... 232
Figure A1.13: Representing networks and trees in a functional structure. .................. 234
Figure A1.14: Representation of the function of a car in TARRAUH......................... 235
Figure A2. 15: The offsetting of the steering axis........................................................ 237
Figure A2. 16: Unrestricted turning of the wheel......................................................... 238
Figure A2. 17: Adjustable sensitivity in servo steering. .............................................. 239
Figure A3. 18: Hierarchy of service providers, the resource for high-level functions. 244
Chapter 1: Introduction
Increasing competition for better product functionality, quality, features, customization,
environmental friendliness, lower cost and shorter delivery time presents unprecedented
challenges for product manufacturing enterprises. These challenges cannot be
completely addressed by utilizing advanced manufacturing technologies and optimizing
production processes. Instead, companies are forced to improve the entire array of
activities related to product development including marketing and problem
identification, design, production, distribution, post-sale services, and environmental
obligations such as recycling.
Of all the activities related to product development, design is considered to be the most
important one. Various studies have concluded that a product’s characteristics are
primarily determined by its design, particularly by the decisions made during the early
stages of the design process ([Ullman 1992], [Boothroyd 1994], [Kushnir 2003],
[Simpson 1998], [Condoor 1999]). Therefore, much research in recent years has been
dedicated to developing a fundamental understanding of the design process, improving
design education, and devising tools and methods for assisting designers.
In this context, this thesis discusses adaptable design as a design paradigm for the
economical success of the producer, the satisfaction of the customer, and the protection
of the environment. Adaptable design, as the name suggests, aims at developing designs
that are adaptable to various circumstances. Adaptability helps the producer reuse the
existing design knowledge and manufacturing infrastructure, which is more cost
effective than creating new designs and production processes. Adaptability allows the
user to utilize the same product under varying circumstances, hence replacing several
products with one. Adaptable design is also beneficial to the environment because it
reduces the total production volume and instead develops products that yield more
service than their conventional counterparts.
In this introductory chapter, the first section is dedicated to providing some background
information. This section first discusses an important premise of this research, which is
the treatment of ‘adaptability’ as a design characteristic. This leads to ‘design for
adaptability’ as a new design paradigm. Next, this section discusses the current state of
research and identifies ‘general adaptability’ as an original research topic. Section 1.2
briefly discusses the principle of segmentation (modularization) which is fundamental
to the approach of this thesis. Section 1.3 is dedicated to clarifying the distinction
between adaptable design and modular design. Section 1.4 provides an overview of the
thesis in the logical sequence of main ideas. This is followed by the thesis organization
and list of terms at the end of this chapter.
1.1. Background
Adaptability as a Design Characteristic
There are practical and economical benefits in the ability to adapt a product to different
service conditions. For example, it would be useful if we could adapt a car to varying
driving needs, adapt a CNC lathe to better technologies that become available, or adapt
a single good design to different sets of requirements and thus produce several different
products. Adaptation becomes particularly beneficial when a product would be put out
of service while it is in good working condition. Such premature retirement of products
might be caused by changes in the needs or expectations of the user, by changes in
operational conditions or government regulations, and increasingly in the modern
engineering market, by the technological obsolescence of components. In such cases,
adaptation creates new service life for products which otherwise would be disposed of.
Despite such obvious advantages of adapting products, adaptation is not always
practically possible. Some adaptations can be performed at reasonable cost and effort,
for example the adaptation of a personal computer to the new technologies such as
faster CPUs and larger memories. Some other adaptations, on the other hand, are too
expensive or difficult to be practical, for instance the adaptation of a car for a different
number of seats or for a different location of the driver in the vehicle.
The difficulty of adapting a product to a new set of service conditions depends on the
differences between the new service and the original service, as well as on certain
attributes of the product that determine how easily the product can be altered from its
current state to the required new state. Examples of these attributes include the way the
product is divided into subsystems, the way its various subsystems are connected, and
the possibility of altering the configurations and functions of various components. The
collective effect of these attributes can be viewed as a product’s ability to be adapted to
new service conditions. This characteristic can be called the product’s “adaptability”
[Gu 2002].
Thus, adaptability can be treated as a characteristic similar to manufacturability,
recyclability, or upgradeability. Similar to these characteristics, the adaptability of a
product depends on many specifications and attributes of the product’s subsystems and
components, hence is easy to describe and understand but is difficult to quantify. Also
similar to these characteristics, the adaptability of a product is primarily determined by
its ‘design’.
Design for Adaptability
A design paradigm is a theoretical framework for designing; it may include rules and
generalized methods, guidelines, specific procedures, software tools, etc. Examples of
engineering design paradigms include concurrent engineering, systematic design, and
decision-based design [Pratt 1993]. In the past few decades, several paradigms have
been developed for the purpose of improving specific characteristics of products during
the design process. These are known as Design for X (DFX) (e.g. “Design For
Assembly”, [Boothroyd 1983]). A DFX paradigm helps designers develop products
which are likely to perform better with respect to characteristic X. For example, design
for manufacturing and design for recycling are established paradigms that help
designers develop products with greater manufacturability and recyclability.
The goal of this thesis is to contribute to the development of a new DFX paradigm, one
which aims at developing products with greater ‘adaptability’. This paradigm can be
called design for adaptability, or adaptable design (AD). While a conventional
mechanical product is designed to serve in its normal operational mode, an adaptable
product is designed to be able to change its operational mode in some circumstances.
Current State of Research
Design for adaptability is not an established paradigm in mechanical engineering design;
therefore there is a shortage of direct literature on the subject. There are, however,
several design methods whose objectives are to increase various types of adaptabilities
in mechanical designs. These methods are presented under different titles in the
One way of locating research pertinent to AD is to categorize various scenarios of
adaptations, then search for the existing design methods which aim at facilitating these
scenarios. This approach, to be discussed in detail in Chapter 3, results in the
identification of four objectives among the design methods which are related to AD:
upgrading, variety, versatility, and customization. Here, “upgrading” refers to
adaptations that occur over the course of time; “variety” and “versatility” refer to the
adaptability of designs and products respectively; and “customization” is a general term
used in the literature to refer to all these scenarios. By this definition of terms, the
existing design methods related to AD can be categorized under the following four
paradigms: design for upgrading, design for variety, design for versatility, and design
for customization.
For instance, methods of design for upgrading are those which aim at facilitating future
adaptations of artifacts; these methods postpone the retirement of products and extend
their service life. Often these methods focus on the technological obsolescence of
components. In such cases the process of upgrading typically involves the replacement
of expired parts. The common method for facilitating this process is to design the
rapidly-expiring subsystems as replaceable modules. A very successful implementation
of this method can be observed in the design of personal computers. Soon after the
initial introduction of PCs to market, it became evident that their premature retirement
was an issue. A typical PC has a relatively long product life because it does not undergo
much wear and damage, yet it has a short service life due to the rapid technological
obsolescence of its components. Therefore, PCs would be disposed of while in good
working condition. The utilization of a modular architecture in the design of PCs avoids
this problem. In this architecture, rapidly expiring parts such as the CPU or the memory
card are designed as separate modules and can be easily replaced with newer ones.
As the PC example shows, the methods of design for upgrading typically assume that
future upgrades are known at the time of design, so that the subsystems which are
bound to be replaced in the future can be designed as detachable modules. The review
of the other design methods related to AD reveals that they also assume that future
adaptations are known in advance thus can be “designed-in” at the beginning of product
planning. For example in ‘design for variety’ a common method is the development of
shared platforms, based on which a family of products can be created through the
addition of differentiating modules. In this procedure, it is assumed that product
variations are foreseen at the time of design, therefore their commonalities can be
developed as shared platforms.
Specific and General Adaptabilities
The above discussion presented an important observation that the existing design
methods, though diverse in their objectives and techniques, have an element in common
which is the assumption of forecast information during designing a product for future
adaptations. Since any of the current methods targets a specific set of adaptability
objectives from the outset, this thesis uses the term “specific adaptability” as an
umbrella term to refer to the aim of the existing design methods. The methods of design
for specific adaptabilities are very helpful, but generally they are only applicable to their
foreseen adaptations. There are certain design characteristics that make one product
generally more adaptable, even to unforeseen changes, than another product with
similar functions. We use the term “general adaptability” to refer to these
Currently, a formal approach towards designing products for general adaptability is not
available in the mechanical engineering design literature. This is primarily due to the
inherent properties of the mechanical design process, to be discussed in Chapter 4. As a
result of these properties a typical mechanical system is designed for a specific
operational mode. In such a system the overall functions are achieved through the
interactions among many subsystems and components which are often useful only in
their exact configuration. Therefore, the structural or functional alterations which are
necessary for an adaptation task are typically very difficult to make.
In design methods for specific adaptabilities, the overall strategy is to provide for the
features which are needed for a ‘predetermined’ set of adaptations. A design method for
general adaptability naturally requires a different strategy because in the absence of
forecast information no particular adaptations can be targeted during the design process.
In order to develop a design method for general adaptability, this thesis first takes a
theoretical approach and presents an ideal architecture for general adaptability. Then it
shows how the new technologies can be utilized to overcome the inherent difficulties of
mechanical design and develop mechanical systems which emulate this ideal adaptable
architecture as closely as possible.
1.2. Segmentation (Modularization)
There are several techniques for enhancing a design with respect to the specific
adaptabilities discussed above. These techniques include modular design, product
family development, and platform design. It can be observed that the underlying
principle in these techniques is the segmentation of a product. In a segmented (modular)
product the alterations made in one place are likely to be confined within one or a few
segments; whereas in a product with a more integral architecture the alterations made in
one place are likely to propagate to the rest of the product. Therefore, a product with a
segmented architecture is generally easier to modify, hence has greater adaptability.
In the segmentation methods for specific adaptabilities, where future changes are known
in advance, the main task is to find a segmentation scenario that yields the best results
with respect to the target objectives. For example, in design for variety the segmentation
criteria are commonality and differentiation. That is, those subsystems which are shared
among a family of products are grouped together as a common platform, and the
differentiating features are developed as add-on modules. As another example, in design
for upgrading the segmentation criterion might be obsolescence. That is, the rapidlyexpiring subsystems are developed as replaceable modules.
Given the effectiveness of segmentation in achieving specific adaptabilities, this thesis
explores the use of the same principle for achieving general adaptability. Therefore, the
main task is to find a segmentation scenario that yields greater general adaptability in a
design. For this purpose, this thesis suggests the use of functions as the criterion for the
segmentation of a design. That is, the physical subsystems of a product are divided in
such a way that every subsystem performs a useful function.
Since the design process begins from the functional domain and proceeds to the
physical domain, the function-based segmentation of a design can be viewed as the
subordination of the physical structure of a product to its functional structure.
Function-based segmentation is the main method of this thesis towards achieving
general adaptability. This method, along with guidelines for implementing it in the
design of mechanical systems, will be discussed in Chapter 4.
1.3. Modular Design and Adaptable Design
In the literature, the term modular design is often used in its broad sense to refer to all
methods of segmentation. For example, platform design might be considered a special
case of modular design in which common subsystems are developed as a shared module.
Modular design, however, is a different concept from adaptable design despite the fact
that modularization is also the main method for increasing adaptability.
Segmentation of a product in modular design might be performed for various objectives,
including those related to adaptability such as upgrading and those unrelated to
adaptability such as material recycling. Adaptable design, on the other hand, may or
may not use the method of segmentation but its objective is invariably related to
adaptability. The following figure illustrates the relation between these two concepts.
Modular Design
• Transporting large objects
• Division of design task
• Material recycling
• Repair & Maintenance
Adaptable Design
Module Replacement:
• Upgrading,
• Variety
• Versatility
• Customization
• Flexibility
• Extra features
• Standardization
• Generic forms
Function-based modularization for
general adaptability
Figure 1. 1: Modular Design and Adaptable Design.
Area 1 on the left includes those objectives of modular design which are not related to
adaptability. These objectives will be discussed in Chapter 2. Area 2 includes methods
of increasing adaptability which are not related to modularization. One such method is
to utilize flexible elements in a design, where flexibility refers to the ability of a system
to be reconfigured without breaking it into its constituent subsystems. Examples of such
integral flexibility can be seen in natural systems such as human skin, and in man-made
objects such as hydraulic hoses, manufacturing adjustments, and flexible shafts.
Another method is the provision of extra features and functionalities in a design for
possible future needs. For example, an output utility shaft is provided in the design of
most farm tractors; thus making the tractor adaptable to various functions. The
applicability of this method is often limited by the higher cost of excess functions.
Other methods in this area include increasing the compatibility among subsystems
through the standardization of systems and their interfaces, and using more regular
surfaces and generic forms which facilitate future alterations and amendments.
Area 3 is the overlap between the two concepts. Methods in this area utilize the
principle of segmentation to localize the structural alterations made for the purpose of
adapting the product. This area includes the conventional methods of modularization for
upgrading, versatility, variety, and customization, as well as the function-based
segmentation method proposed in this thesis for achieving general adaptability.
1.4. Thesis Overview
Given the fact that AD is not an established research topic in mechanical engineering
design, this thesis inevitably presents a few definitions and categorizations. This task
begins by identifying the ‘extension of utility’ as the purpose of adapting an artifact.
This view reveals a logical relationship between the scarcity of resources and the need
for adaptability. This view also helps to identify various categories of adaptations by
considering different modes for extending utility, as mentioned in the background
section. These topics will be discussed in Chapter 3.
The extension of utility might also be the purpose of many tasks which are not
considered adaptation. Therefore, the next step is the adoption of limits for the scope of
tasks related to AD. Three such limits are proposed in this thesis. First, adaptation is the
extension of usage into ‘different’ operational modes. This limit excludes from AD the
methods of prolonging the normal service life of products, such as the methods of
design for durability, repair, and maintenance. Second, adaptation is a process in which
components are utilized as they are. This limit excludes from AD the methods of
recovering artifacts in other forms, such as material recycling. Third, an artifact is
considered ‘adapted’ if a considerable portion of it remains together in the new
operational mode. This limit excludes from AD the methods of design for part reuse. It
can be seen that these limits are relatively subjective, for example the distinction
between part reuse and adaptation might be unclear. However, these limits resolve some
inconsistencies that exist in the usage of the term ‘adaptability’ in the literature, and
help in better determining the scope of AD in this research.
After these definitions are established, the next step is to seek design methods for
increasing various types of adaptabilities in mechanical systems. Current methods are
identified and classified under four categories discussed in the background section.
These are design methods for increasing the upgradeability, customizability, versatility,
and variety of mechanical systems. In any of these methods, a set of future adaptations
is targeted from the outset, and then the product is designed in such a way to facilitate
these foreseen adaptations.
It could be observed that there are no formal design methods for designing mechanical
systems for general adaptability, that is, without a priori aiming at a specific set of
adaptations. Therefore, AD could be divided into specific AD and general AD. Specific
AD refers the design methods discussed above, and general AD refers to a method for
designing adaptability into mechanical systems without targeting particular adaptations.
General AD
As discussed in previous sections, ‘segmentation’ is the principal method for achieving
adaptable architectures. In the methods of specific AD, the segmentation criteria depend
on what adaptations are targeted during the design process. For example ‘average
service life’ and ‘commonality’ are the segmentation criteria in design for upgrading
and in platform design. In general AD, however, different criteria are needed because
adaptations are not predetermined. This thesis proposes the use ‘functions’ as the
segmentation criterion. That is, the subsystems of a product are divided in such a way
that each subsystem independently provides a useful function.
The reason for choosing functions as the segmentation criterion is that a carefully
constructed functional structure can be very adaptable. Chapter 4 presents a model of
design process for this purpose. This scale-independent model follows a decomposition
process similar to the one which has been described by Suh in Axiomatic Design [Suh
1990]. In this process, no function is decomposed unless a solution for it is conceived
first, and any set of sub-functions must be both necessary and sufficient for describing
the requirements of their parent function. Beginning from the initial problem, recursive
decomposition of functions results in a hierarchical structure. In this hierarchy, the
relation between a function and its subordinates is a causal relation. That is, every
function in the hierarchy exists only to assure the proper functioning of the adopted
solution for its parent function. In an ideal functional structure, there are no causal
relations between functions except for those between a parent function and its subfunctions. Therefore, an ideal structure demonstrates two properties when a function is
removed from it. First, this function’s sub-functions and their subordinates become
unnecessary and can be eliminated as well. Second, there is no need to eliminate
anything else from the structure. Due to these properties, when such a functional
structure is modified in some location, these modifications do not propagate to the rest
of the structure. This makes such a structure ideal for adaptation tasks.
Therefore, the approach of this thesis is to imitate the properties of an ideal functional
structure discussed above in the actual design of a product. This ideal functional
structure is called the ‘rational functional structure’ in this thesis because the
decomposition process that generates this structure represents the design rationale. Thus
the proposed method can be equivalently described as ‘function-based segmentation’ or
‘the subordination of the physical structure to a rational functional structure’. In this
method, first the ‘physical functions’ that are expected from the final product are
distinguished from non-physical requirements. Several sets of physical functions can
describe the purpose of the product; the designer should choose the set in which
functions are as useful and recurring as possible. Then, for each function a separate
solution is devised which is self-contained and autonomous in delivering that function.
Such a module should not depend on a particular configuration for its operation, and
should be useable anywhere as long is its proper inputs and outputs are provided. If a
module requires further decomposition and problem solving, the above two steps are
repeated when possible, attempting to develop the subsystems of this module also as
autonomous and functional modules. The result will be a hierarchy of physical modules
that corresponds to the hierarchy of functions in the rational functional structure.
Frame-and-Function Architecture
The functional modules developed in the above procedure, along with product-specific
components if applicable, have to be assembled into the overall product. This assembly
must fulfill the embodiment requirements of the design problem. The overall shape of
the mechanical object can be determined by the spatial positioning of assembled
components, and in some cases, by an actual skeletal structure. If we use the term
‘frame’ for the spatial provisions of the overall embodiment of the product and the term
‘function’ for the functional modules, then the architecture of a product designed by the
above method can be called ‘frame and function’. This architecture is an alternative to
the platform architecture which is commonly used in design for variety and
customization. This will be discussed in detail in Chapter 4.
The development of an artifact through the assembly of functional modules can be seen
in a business model called the virtual enterprise and in object-oriented software
programming. The achievement of this architecture in mechanical systems, however,
might be very difficult or even impossible. Chapter 4 presents several guidelines which
help with implementing this method in the design of mechanical systems. Examples of
these guidelines include:
Subsystems should be functionally autonomous and their functions should be
meaningful and recurring.
The design should begin from the components that interact with the environment
and then proceed to develop the necessary internal mechanisms.
If possible, functions should be achieved by software not by hardware.
Physical dependencies among various assemblies should be minimized (e.g. by
using flexible interfaces and manufacturing adjustments).
Standard components and generic forms should replace product-specific designs
when possible.
Extra features which help with future adaptations and do not add to the cost
should be considered.
The AD Methodology
The overall methodology of ‘design for adaptability’ combines specific AD and general
AD. The procedure first identifies the functional requirements of the original design, as
well as the requirements that are related to future adaptations if such forecast
information is available. Next, a functional structure is developed together with the
conceptual design of the overall product. The product is designed in such a way to
provide for both the functional requirements of the design and the requirements of
targeted adaptabilities. This step assures that the forecast information is utilized and that
specific adaptability is given a higher priority than general adaptability. Then, the
product is further developed using the general guidelines of AD.
Measure of Adaptability
Function-based division of the subsystems of a design might compromise other criteria
such as aesthetics. However, as the increasing scarcity of resources raises the tendency
towards a service-based market, the utility of a mechanical system, which is to perform
a physical function, becomes more important than other criteria. Further, the increasing
prices of natural resources and the increasing affordability of advanced technologies
will justify AD for a greater range of products in the future. These arguments will be
discussed in Chapter 3. At any case, the thesis proposes a measure for adaptability
which can be used in trade-off decisions.
The adaptability of an artifact for a particular adaptation task is calculated based on the
savings that are achieved through adapting an existing product, as opposed to obtaining
a new one. The calculation of saving is based on ‘information content’, which in this
thesis is defined as an indication of ‘total costs’. Since monetary means are the closest
available tool for representing the total costs of activities, the measure of adaptability in
our case studies is based on financial savings. These will be discussed in Chapter 4.
1.5. Thesis Objectives
The objectives of this research can be categorized under three main goals:
Establishing Research Framework
The first aim of this research is to establish a framework with definitions and
categorizations which are needed for exploring “design for adaptability” as a new
design paradigm.
Fundamental and Theoretical Research
The second objective of this thesis is to explore the fundamental principles of design
and establish a theoretical basis for AD. This involves developing a model of the design
process and a method for General AD. This leads to the development of a methodology
which includes Specific AD, General AD, and a method for assessing adaptability.
Methods and Guidelines
The third goal of this research is to develop practical methods and guidelines for the
application of the proposed methodology in mechanical engineering design problems.
1.6. Organization of This Thesis
Chapter 2 reviews the literature on engineering design research. Chapter 3 discusses the
objectives of adaptability in engineering systems and the benefits of AD from the user,
producer, and environmental perspectives. Chapter 4 presents the central arguments and
methods of this thesis; it includes a method for measuring adaptability and the methods
and guidelines for adaptable design. This chapter is largely self-sufficient as some
concepts that have been discussed in Chapter 3 are also briefly repeated in Chapter 4
when required. Chapter 5 provides several examples for both specific AD and general
AD. It is followed by a summary of contributions and conclusions in Chapter 6.
1.7. Terms and Definitions
The following is a list of terms which are used for specific meanings in this thesis.
AD: Adaptable Design, which refers to Design for Adaptability.
Adaptability: the ability of a product to adapt to varying service requirements.
Artifact: the means of accomplishing a human purpose. In engineering design
an artifact is a physical device created for the purpose of performing a function.
Although in some cases the physical object itself is the purpose of creating it; in
a service-based market only the function is the purpose of creating an artifact.
Decomposition: the process of introducing new requirements in a design
problem in order to assure the proper functioning of adopted solutions for the
functional requirements of the problem.
Design: as a noun, design refers to a blueprint or a plan or a recipe for
accomplishing goals. In engineering, design is the set of instructions for the
manufacturing of a physical object.
Environment: a system that includes what is relevant to a goal, hence relevant
to the artifact that achieves that goal.
Frame: the abstract or actual skeleton that determines the overall embodiment.
Frame-and-Function Architecture: a temporary assembly of functional
modules on a spatial frame for a temporary service.
Function: the actual physical effect of a component on material, energy, and
signal. The function of an engineering system is assumed to be its raison d’etre,
thus function is the goal of designing, producing, and using a product.
Functional Structure: the hierarchy of functions, with initial function
requirements in the apex and the decomposed functions in mid and end levels.
This hierarchy is also called the ‘rational functional structure’ because it is
generated through the processes of problem solving, decision making, and
decomposition. This definition is consistent with the definition provided in
Axiomatic Design. It is, however, in some cases not consistent with the more
popular definition provided in Systematic Design. Systematic Design defines the
functional structure as a description of the operation of an existing device, not as
the representation of the device’s purpose or the design rationale.
Information Content (IC), Total Costs: the IC of a design is the total amount
of resources which are needed in order to achieve goals according to that design.
This includes human and intellectual resources, technological expertise and
infrastructure, natural resources, space, materials, energies, etc. The IC of a
design is also equivalently called its Total Costs in this thesis.
Information Processing (IP): the work or effort spent for performing a task.
Any work is equivalently treated as ‘information processing’ or ‘allocation of
resources’ in this thesis.
Information Processing Capacity (IPC): a limited resource that represents the
amount of work or IP which is available for performing a task.
Objective, Goal, Purpose: a preferred state that would not automatically occur
without our intentional intervention, that is, without human design.
Operational Mode: the operational specifications of a product including
working conditions, functions, performance characteristics, user interface,
physical attributes, etc.
Physical Structure: the hierarchy of physical assemblies in a product.
State: the attributes of the environment including its dynamic trends.
Task: the process of accomplishing a set of goals. This definition is utilized
when the concept of information content in this thesis is introduced.
Utility, Service: the purpose of a product, which is to perform a function.
Chapter 2: Literature Review
In this chapter, Section 1 reviews some of the theoretical research in engineering design
and Section 2 reviews the research relevant to product configuration design. Research in
this category is related to design for specific adaptability, as discussed in Chapter 1. A
brief discussion on the current state of research related to AD is provided in Section 2.3.
2.1. Theoretical Engineering Design Research
In recent years, both academia and industry have realized the growing importance of
structured, scientific, and industrially tested theories and methods for design. However,
theoretical research towards establishing the science of design has not produced enough
practical results [Tate 1995]. Design research is expected to yield practical benefits
whereas research on natural sciences is not necessarily relevant to practice ([Reich
1992], [Broadbent 1981], [Cross 1980]). In discussing the current research towards a
scientific theory of engineering design, Dixon states that a scientific theory is the
ultimate goal of design research, and that the design research is in a pre-theory stage
[Dixon 1988]. His view is that cognitive studies involve far too many ill-defined
variables to support a theory and that prescriptive models are premature until they can
be based on a validated theory. He also discusses how pre-theory research in
engineering design can advance towards that goal. Finger and Dixon state:
“One school of thought believes that it is necessary to develop theories of design…
Another school believes that the matter of design theory is an irrelevant distraction…
Some consider that the best path is to concentrate on those portions of the design
process for which we already have useful theories and tools, such as Decision Theory
and Optimization…” [Finger 1989].
Several researchers have contributed to the establishment of scientific principles for
various aspects of the design process. Examples include the works of Yoshikawa and
Tomiyama on a general design theory, the mathematical representation and a general
design governing equation developed by Gu and Zeng, and Suh’s axiomatic design
theory ([Yoshikawa 1981], [Tomiyama 1996], [Suh 1990], [Zeng 1999-a], [Zeng 1999b]). Various other approaches are also of a theoretical nature. Examples include models
of the creative process of design ([Altshuller 1984], [Dorst 2001]); cognitive models
([Kolodner 1983], [Ullman 1987], [Condoor 1992]); Robust Design, House of Quality,
and other techniques of clarifying design objectives and assessing their attainment
([Taguchi 1987], [Clausing 1994], [Ramaswamy 1992], [Belhe 1996], [Chen 2002]);
representation of technical functions ([Cross 1980], [Hashemian 1997-a]); and
systematic design ([Pahl 1988], [Hubka 1996]). This section provides a review of some
of these theoretical approaches towards engineering design.
2.1.1. Descriptive (Cognitive) Models
Research on the nature of the engineering design process may result in the development
of a descriptive model of design. The basic task in developing descriptive models is to
study how humans create designs in order to determine what processes, strategies, and
problem solving methods designers use. One way to develop descriptive models is to
conduct protocol studies [McNeill 1998]. The challenge with such protocol studies is
that much of the design process happens in the mind of the designer and thus the
documentation does not reveal the entire thought process. Further, the process of
documentation may interfere with the designer’s work during the design process.
Despite these shortcomings, protocol studies have revealed some facts about the nature
of design [Finger 1989]. Some of these studies suggested that designers need the help of
different tools depending on their experience with the design task and the stage of
development during the design process. For example, information retrieval tools are
needed when prior experiences with similar designs exist, as in case-based design
[Kolodner 1993]. Another study by Ullman revealed that designers tend to pursue a
single design concept, and that they will patch and repair their original idea rather than
generate new alternatives [Ullman 1987]. If prior experience exists, designers reuse
familiar solutions and will not explore alternatives or innovative ideas unless their new
design fails badly and cannot be salvaged. These observations reveal the impact of bias
or the so-called psychological inertia on the designer’s work. Another interesting
observation is that design is a fractal-like process in which the stages of design repeat
continuously at different times and at different levels of detail [Ostrosi 2003].
Cognitive research has also revealed that during the design process, designers use
general prototypes and adjust them to the particular demands of a given problem. Maher
created a system that represents design knowledge as prototypes [Maher 1987]. The
scope of this method of representing design knowledge, however, is limited to problems
in which synthesis consists of only a morphological combination [Hashemian 1997-b].
In another study, Maher and Tang presented a protocol study of human designers
looking for evidence of co-evolution in design, that is, the simultaneous evolution of
both the design problem and the design solution during the design process [Maher
2003]. Co-evolutionary design is developed as a cognitive model of design in which
designers iteratively search for a design solution and make revisions to the problem
specification. They discuss the similarities and differences between this cognitive model
and a co-evolutionary computational model they have developed. Their computational
model assumes two parallel search spaces, the problem space and the solution space,
both of which evolve during the design process.
To date, descriptive models have mainly focused on generating hypotheses based on
observations of designers without devising experiments to test these hypotheses. These
models, however, have strong advocates. For example, Gero asserts that design systems
must be based on human design processes. He argues:
"Design paradigms based on mathematical models inherit the properties of the
mathematical models on which they are based. Thus, it is possible to prove such
characteristics as feasibility and optimality about a resulting design. However, such a
design paradigm has limitations in two major areas: the processes used to achieve
designs are far removed from the way humans carry out this process; and much of
design can only be represented symbolically but not mathematically" [Gero 1985].
2.1.2. Synthesis and TRIZ
Synthesis is an integral part of any design activity and is recognized as one of the most
important elements of the design process. In this sub-section three approaches to
creative synthesis are discussed: morphological synthesis, brainstorming methods, and
TRIZ. The reason for choosing these three approaches is that each of them represents a
school of thought regarding creativity. Other approaches not discussed here include
literature search, emulation of nature, emulation of technical objects, consulting the
experts, mind mapping, excursion, and checklists ([Pahl 1988], [Higgins 1996],
[Strawbridge 2002]).
Morphological Synthesis
The morphological approach, mainly credited to Zwicky ([Zwicky 1948, 1969]),
represents a philosophy that aims at changing the process of creativity from an intuitive
process to a systematic one. Morphological creativity is a process for creating new ideas
through analyzing the form and structure of the existing ones and changing the
relationships among their components. The process of morphological creativity consists
of the following steps:
Definition: objectives and constraints of the problem are clearly identified.
Abstraction: the related elements of the problem are clustered into groups.
Solution: several solutions for each group are found. This may involve
Morphological Synthesis: the problems and their solutions are put in a
morphological matrix. Many overall solutions are then synthesized through
different combinations of sub-solutions in an unbiased operation.
Pruning: since solutions may be too many, the ones that violate constraints or do
not perform well with respect to the most important criteria are filtered out.
Evaluation: the remaining solutions are evaluated and the best ones are selected.
Though helpful, this method has some drawbacks. The first issue is the division of the
problem into its main sub-problems because usually there are no robust guidelines for
performing this operation and finding the optimal division scenario. The second issue is
that this method assumes that the relationships among sub-problems can be temporarily
suspended, thus sub-problems can be solved in isolation. In practice, especially in
mechanical engineering design problems, the compatibility of these solutions during
synthesis (combination) can be problematic. The third issue is the combinatorial
increase in the number of overall solutions, where most of them are meaningless. The
fourth issue is that the outcome of this method is confined within the limits of ideas
generated for sub-problems. That is, this method does not fully overcome the
psychological inertia ([Allen 1962], [Strawbridge 2002]).
Brainstorming and Lateral Thinking
The brainstorming method was introduced by Osborn in 1957 in the United States,
where the theories of Sigmund Freud were dominant [Osborn 1957]. Freud’s theory
differentiates between the subconscious and conscious minds. The process of ideation,
or ‘enlightenment’, is believed to happen in the subconscious mind where thoughts are
absolutely uncontrolled. The process of analysis, on the other hand, is believed to be a
controlled process that happens in the conscious mind. Osborn asserted that these
processes hinder each other. Therefore, he aimed to develop a mechanism to separate
the above two processes in time. The only problem was to bring the original ideas,
believed to be generated in the subconscious mind, to the forefront, or to the conscious
mind. So he designed and conducted his famous brainstorming meetings, where
everybody was encouraged to present any idea without fear of criticism. The hope was
that in the storm of wild ideas the barrier would be broken and good ideas from the
subconscious would come to the surface. Osborn’s brainstorming method has been
modified many times and several versions of it are currently available ([Hashemian
1995], [Byrne 1993]).
Edward de Bono presented a different creativity method called ‘lateral thinking’ [de
Bono 1992]. He drew an analogy between the process of learning and the way land is
shaped by rain. In his view experiences shape our mind the same way that rainwater
carves grooves in land. As more rain falls, the existing grooves become deeper and the
path that the rainwater follows is no longer arbitrary; instead it traces the existing
patterns of land. Similarly, as a person solves problems, his or her problem-solving
experience turns into the only way to solve a problem. According to de Bono, the
process of thinking (or controlled thinking) follows the routes that are already carved in
the mind. Then he defines creativity as ‘lateral’ thinking, which is analogous to moving
across grooves, or laterally, not along them. This view, however, asserts the same
conclusion as Osborn’s: mind practices can be improved so that a designer might access
innovative ideas that are normally inhibited by the psychological momentum towards
pre-existing solutions. In the basic Trial-and-Error method which will be discussed in
the next sub-section, one task is to increase the number of variants considered to the
greatest extent possible, and the second task is to increase the number of good variants
which are based on innovative and original ideas. Brainstorming and lateral thinking
help with both these tasks.
Theory of the Solution of Inventive Problems (TSIP/TRIZ)
TRIZ, a Russian acronym meaning “the theory of inventive problem solving”, is a
systematic approach to finding innovative solutions for technical problems. With the
thawing of the Cold War climate, TRIZ entered the West over a decade ago when a few
American academics began studying its principles and applying them to real design
situations [Webb 2002]. The application of TRIZ enables designers to create new and
improved products in a way that does not rely on the exhaustive trial of variants or
accidental discoveries. Several field applications of TRIZ have been reported in the
literature ([Domb 1998], [Kourmaev 2003], [Mann 2002]).
The development of TRIZ is credited to Altshuller, a Russian engineer and researcher.
He observed that many processes can be controlled, at least in theory, even if they are as
complicated as nuclear reactions or genetic inheritance ([Altshuller 1984], [Altshuller
1999]). He extended this observation to the process of creativity. The work of Altshuller
and his colleagues started in early 1950's with the goal of developing a new technology
for creativity, one that is both effective and controllable. The result was the Theory of
the Solution of Inventive Problems (TSIP or TRIZ) and its methodical procedure,
Algorithm for the Solution of Inventive Problems (ASIP). He presented his methods of
controllable creativity in the field of technological inventions, as opposed to artistic
creativity. His methods are applicable to both engineering and science.
Altshuller observed that the conventional method of human creativity is the trial-anderror method. In the absence of any knowledge, trials are performed at random. If some
knowledge about the properties of variants is available, trials become more selective
and the process becomes more efficient. The random trial-and-error method is not
effective for complex problems of modern technology. In Altshuller’s schema, the level
of complexity of a problem is decided by three criteria. The first criterion is the number
of trials, which is only a few for simple problems and can be hundreds of thousands for
more complex problems. The second criterion is the scope of required modifications in
the existing state of things; simple problems might be solved by using existing means,
while complex problems might demand a change in the surrounding environment of a
design object in addition to changes to the object itself. The third criterion is the scope
of required expertise, which for simple problems might be confined within the
boundaries of a narrow specialty, but for complex problems might extend into an entire
field of technology, across other fields, or even into abstract levels of science beyond
the operational technology. Noting the inadequacy of trial-and-error methods for
complex problems, Altshuller stated the need for a new technology for invention that
allows complex problems to be solved by fewer trials.
Altshuller and his colleagues began with the postulate that most creative solutions are
generated through analogy, therefore the laws of inventive solutions can be obtained by
observing previous inventions. They studied tens of thousands of patents in order to
generalize conclusions about inventive solutions. They concluded that invention was to
find a good solution in the solution space, and good solutions invariably solved
contradictions. Contradictions happen when an improvement causes deterioration, when
a substance needs to be liquid and solid at the same time, when we need to seal an
object yet need to access it, etc. Contradictions can be eliminated by using various
methods, for example by separating the conflicting elements in space and time. They
noticed that despite the fact that any invention involves the removal of a contradiction,
nobody systematically abstracted a problem into a contradiction in order to find a way
to eliminate it. From studying many innovative designs, they also discovered that
although technical inventions were many, their underlying contradictions were few.
They hypothesized that objective laws must exist that can suggest efficient methods for
the removal of any given contradiction. They stated that these laws could be developed
through studying the methods of contradiction removal embedded in prior inventions,
and that based on these laws the creative process could be systematized.
Based on the above conclusions, they developed a set of laws, rules, principles, and
methods for the development of technical systems such as the ones seen in Altshuller’s
TRIZ Contradiction Matrix analysis and 40 principles [Altshuller 1998]. These
principles have been developed further by several other researchers, improvements and
applications continue to evolve, and new developments are reported in various sources
such as the online TRIZ Journal ( A few examples of these
principles are provided below:
Law of the s-field: All solutions in the form of a technical system need to have a
minimal number of elements in order to become a functional system. An object or
substance (S1) requires interaction with its environment (S2) through a field in order to
deliver its function. The field represents the energies or signals through which the
technical system interacts with the outside world. This minimal system, consisting of
substances (S1, S2) and a field, is called the S-field. The S-Field is both necessary and
sufficient for the minimal description of a technical system.
Law of minimum completeness: A system must fulfill the minimum requirements that
bring the system to life. If any of these requirements fails, the whole system fails.
Law of the ideal solution: The development of a system proceeds towards an ideal
system that provides the function without having a system.
Corollary: A "controllable" system has at least one controllable part.
Method: Consider an ideal solution in which the underlying contradiction is eliminated
without any complexity; the goal is achieved without paying any price; the function is
delivered without having a machine. Then try to find a solution as close to this ideal
solution as possible.
Method: use solution-neutral language to describe a problem.
Principle: Divide an object into independent parts; increase the degree of segmentation;
make the object easy to disassemble.
Principle: Change an object's structure from uniform to non-uniform; make each part of
an object fulfill a different and useful function.
Principle: Make an object perform multiple functions; eliminate other parts.
TRIZ proposes an algorithm for the solution of inventive problems (ASIP). In this
algorithm, a problem is first abstracted into its underlying physical contradiction. Then
prior solutions for this contradiction are selected from a repository of contradiction
removal methods and rules. Next, a solution for removing the contradiction is chosen
and is adapted to fit the characteristics of the problem at hand. This is followed by the
design of the embodiment of the physical system and its evaluation and testing.
2.1.3. Axiomatic Design
Suh [Suh 1990] perceives the design process as a mapping between domains. During
the design process, the problem that is being addressed can be divided into four
domains: customer domain, functional domain, physical domain, and process domain.
In the order listed, the elements associated with each domain are customer needs (CNs),
functional requirements (FRs), design parameters (DPs), and process variables (PVs).
The primary focus of the design process is the mapping from the FRs to the DPs. FRs
are expressed in solution-neutral terms. By definition they are both necessary and
sufficient for the description of the design goal, thus they are the minimum set of
requirements which completely characterize the design objectives for a specific need.
Design parameters are, in effect, solutions to FRs.
The design process progresses from a system level to more detailed levels. That is, it
extends from systems to subsystems to assemblies to parts to part features. This may be
represented in terms of a design hierarchy of decompositions, where each
decomposition in the functional domain is performed only after a solution for a given
FR is found in the physical domain. That is, the designer goes through a process of
zigzagging between domains during the decomposition of a design problem. Although
not as famous as the two axioms, Suh’s statement of the nature of decomposition is in
fact a very important part of the axiomatic design theory. Suh’s view towards
decomposition is adopted in this thesis and is utilized in the development of the design
process model in Chapter 4.
After adopting the above principles for the modeling of the design process, Suh
developed two axioms and several corollaries to guide the design process. His axioms
are quoted below:
The Independence Axiom: Maintain the independence of functional requirements.
The Information Axiom: Minimize the information content (of the design).
As mentioned earlier, in Suh’s approach a design problem is described by a set of FRs
that are both necessary and sufficient for representing the goals. From the necessity of
FRs, it can be logically concluded that redundancy is not allowed, therefore FRs must,
by definition, be independent from each other. Then, the first axiom encourages the
designer to ‘maintain’ this independence, and not ‘couple’ them through the DPs that
are chosen for them. This basically means that for each FR, which by definition must be
independent from other FRs, an independent DP must be found so that the desired value
of the FR can be achieved through adjusting the value if its pertinent DP. The
Information Axiom states that among those designs that maintain the independence of
the functional requirements, the one with the minimum information content is the best
solution. The design with the minimum information content is the one that has the
highest probability of success, where a successful design is the one that delivers the
required FRs. This design is also the one with the least complexity in terms of satisfying
the functional requirements ([Albano 1993], [Suh 2001]).
Suh also introduced the design matrix, which shows the relations between the FRs and
DPs at a given level of the design hierarchy. There are three possibilities for the design
matrix. It can be a matrix populated both above and below the diagonal (coupled
design), a triangular matrix (decoupled design), or a diagonal matrix (uncoupled design).
The Independence Axiom states that the uncoupled design is ideal, and a coupled design
is not acceptable and should be at least converted to an uncoupled design. Axiomatic
design is not a detailed prescription on how to perform every step of the design process.
Instead, it tells us how the artifact should be designed; the design should be uncoupled
and with minimum information. These are general guidelines that help the designer
make better decisions during the design process. Since Suh’s axiomatic design is
relevant to parts of this thesis, further discussions on design axioms are provided in
other chapters.
2.1.4. Systematic Design
The development and advancement of the systematic design methodology is primarily
attributed to the engineering design research performed in Germany ([Hubka 1980,
1988, 1992]). Some works have been translated into English, and the book
“Engineering Design: a Systematic Approach” by Pahl and Beitz [Pahl 1988] is often
considered as representative of the German approach to the systematic design
methodology [Wallace 2000]. Systematic design is a truly prescriptive model of design.
It prescribes a methodical procedure for design, splitting the design procedure into four
main phases: preparatory, conceptual, embodiment, and detail design.
The preparatory phase involves problem definition, or the clarification of the design
task. The designer must understand the customer’s need and represent it as a set of goals
for the design of the artifact in the form of technical requirements, which are called
functions. Also during this stage the technical constraints of the design and the
limitations on time, logistics, and production capabilities, as well as the evaluation
criteria are identified. In the conceptual design phase, the main requirements of the
problem are singled out and abstracted into a solution-neutral representation of physical
functions, which reflect the exchange of material, energy, and signal among objects.
Then these functions are systematically decomposed into a functional structure. The
elements of the function structure are incrementally replaced by solution principles, for
which conceptual physical embodiments, called function carriers, are developed. The
decomposition and replacement of functions continue until all functions have been
replaced by solutions. The systematic design approach to creativity and synthesis is
mainly based on the morphological analysis explained previously. The designer is
encouraged to form a matrix of all functions and their solution alternatives (the
morphological matrix), generate many different combinations of these sub-solutions,
evaluate these combinations, and select one (or more). The elements of a selected
combination are then consolidated into the overall conceptual solution. Then the
embodiment design phase begins; it is a phase that determines the shapes, arrangements
(configuration), and interfaces among components, as well as the shape of the overall
design. Some production and assembly aspects are also considered at this stage. At the
detail design phase, the details of the product are determined and optimized. These
include material types, dimensions, surface finishes and tolerances, etc. The detail
design stage is completed when manufacturing drawings are produced. These design
phases are iterative in nature, both within each phase and between phases.
The systematic design approach is a practice-oriented and detailed methodical
procedure for performing engineering design. It is based on the postulate that benefits
can be obtained if the systematic procedure is followed. Some attempts have been made
to verify this postulate (e.g. [Hykin 1975] [Tebay 1984]). The actual process that
designers follow has been compared with the systematic process; the results indicated
that they are not the same. Also, some designers were taught the systematic process and
were then asked to follow it in designing a case study object. Their performance was
compared with that of a group who possessed similar skills but did not follow the
systematic method; using the systematic approach did not reveal any advantage. These
tests on systematic design are not conclusive, however the fact remains that researchers
have not reported many success stories about the use of the systematic method [Finger
1989]. It seems that systematic design needs to be improved in order to become closer
to what designers would do naturally. Also, the inconclusive experimentation with
systematic design might be attributed to a lack of knowledge and experience with this
process amongst designers; this is a barrier that can be overcome by rigorous training of
designers in the area of the systematic design methodology.
2.1.5. Knowledge-Based Design
Knowledge-based design is a general concept and refers to design theories,
methodologies, and tools that are related to the tasks of capturing, modeling,
representing, comparing, and utilizing various types of design knowledge ([Coyne
1990], [Li and Zhang 1998], [Ullman 1994]). Knowledge-based design systems are
typically computer software systems that operationalize the design knowledge elicited
by knowledge engineers. Different types of knowledge are used during various
activities of the design process, and they are modeled differently from one another. For
example, during synthesis the knowledge of the prior designs (experience) is used,
while during evaluation the knowledge about manufacturing processes and the life cycle
characteristics of a design is utilized.
A fundamental research issue is to determine the appropriate representation frameworks
for different types of design knowledge [Hashemian 1996]. Li and Zhang have
classified the design knowledge into four categories: (a) artifact (product) structures, (b)
artifact behaviors, (c) artifact functions and (d) causalities among structures, behaviors
and functions [Li 1998]. Their classification is consistent with the function-behaviorstate model developed by Tomiyama et al for the computer modeling of functional
design [Tomiyama 1993]. Based on this classification, they developed a hybrid graph
approach to represent design knowledge so that the automatic computer-based
comparison of design knowledge is attained through the comparison of hybrid graphs.
Many researchers assert that ‘prior experience’ is the primary source of creative
knowledge for designers [Kolodner 1992, 1993]. This has resulted in several case-based
design systems for modeling the conceptual design knowledge [Hashemian 1995].
Cased-based design systems directly catalogue prior experiences and do not generalize
them into rules. Several researchers (e.g. [Gero 1989-a, 1989-b]) have suggested that
designers form their individual design experiences into generalized groups of concepts
at many different levels of abstraction. Hashemian and Gu proposed a case-based
design system in which prior design knowledge is divided into three levels of
abstraction, namely solution principles, design prototypes, and parametric designs
[Hashemian 1997-b]. Cases in different levels of abstraction are indexed differently and
are accessed by different retrieval techniques. Their model is intended to represent the
creative knowledge of designers. In some cases the expertise of the knowledgeable
designers in a field can be captured and generalized into heuristic rules. This has
produced numerous rule-based design systems for various tasks of the design process
([Rigo 2003], [Cherian 2001]). These expert systems typically replace human experts,
whose participation is necessary for concurrent engineering. These systems, however,
are applicable to specific problems in a given field. Li and Wu discuss the
generalization of design knowledge [Li 1998]. Generalization can be described as the
process of taking a large number of design examples, then extracting and retaining the
salient properties in the form of conceptual descriptions (concepts) that can assist future
design activities. Generalization includes three aspects: a knowledge representation for
generalized concepts, a description language for design examples, and generalization
operators that extract concepts from examples. Knowledge-based systems can be
classified into three categories: model-based, rule-based, and case-based [Coyne 1987,
2.1.6. Decision-Based Design
Designing involves making decisions at almost every step of the process, generally in
the form of choosing from among several alternatives. The process of design is viewed
as decision making by many researchers (e.g. [Mistree 1990], [Hazelrigg 1998],
[Olewnik 2003], [Wassenaar 2001]). Design activities consist of two processes:
synthesis of solutions, and evaluation of the generated alternatives to find the best one.
The focus of decision-based design (DBD) is the latter. DBD is a methodology that uses
the rules of decision theory and its related sciences in design. Related headings in this
area of science include Utility Theory, Multi-Attribute Decision Theory, Game Theory,
Information Science, Analytical Hierarchy Process, etc. Decision sciences and their
application in design, or DBD, have been extensively discussed in the literature (e.g.
[Luce 1957], [Tribus 1969], [Hazelrigg 1996]).
From an abstract point of view, an important aspect of DBD is that it replaces the
artifact-oriented design process with a decision-oriented process. In design problems
that involve more analysis and calculation than synthesis and decision-making, an
artifact-oriented approach is appropriate. However, in more decision-intensive
situations the artifact can not represent the history of risk and uncertainty associated
with the decisions. Paying direct attention to decisions (instead of the artifact) reduces
the chance of making a wrong decision. Thus, DBD reduces the number of design
iterations, which are invariably unwanted, by paying greater attention to decisions
throughout the entire design process, and by considering all the factors related to
decision-making including life cycle issues. Also, DBD helps designers retain the
design rationale which is captured in the form of decisions and their reasoning. This is
especially useful in redesign tasks, where prior experience is often utilized.
Design decisions are usually made based on multiple criteria. These criteria are often in
conflict with one another (personal preferences, resource requirements, etc.). DBD
offers tools and techniques to deal with conflict resolution and multi-objective
optimization [Chen 2002]. It also offers methods for the inclusion and organization of a
large number of life cycle criteria in the evaluation and optimization of design
alternatives [Yoshimura 2003]. Usually some of the attributes of a solution are
qualitative. DBD suggests methods for the quantitative measurement of qualitative
parameters; an example is the systematic function-based evaluation of design
alternatives [Iyengar 1994]. It also offers guidelines to deal with ambiguity, risk, and
uncertainty that are a part of human decision making and human subjective evaluations.
The decisions may be made individually or collaboratively; DBD also proposes
approaches to distributed decision making [Jeong 2002]. Further, designers rarely know
the entire set of potential alternatives, nor do they know all the evaluation criteria and
their relative importance; in most design problems both the solutions and the evaluation
criteria evolve as work progresses [Maher 2003]. In design problems, often there is not
enough time to obtain high levels of knowledge, thus a final decision has to be made
with incomplete knowledge. Making decisions with incomplete knowledge is another
area of research in DBD [Pender 2001].
2.1.7. The General Design Theory
The General Design Theory (GDT) was advanced by Yoshikawa (the English
presentation [Yoshikawa 1981]). GDT is a formal mathematical theory of design. It
models the design process in the framework of set theory. GDT also provides a
prescription for the development of CAD systems. This sub-section will not discuss the
details of the definitions, axioms, and theorems of GDT. More information about the
theory can be found in [Yoshikawa 1981] and other related papers (e.g. [Tomiyama
1994], [Tomiyama 1996]).
GDT starts with assumptions about objects in the world and uses these assumptions to
prove theorems about the nature of design. Yoshikawa considers design as a “typical”
intellectual activity that humans perform. He considers designing ability intrinsic for
human beings and states that design knowledge and skills are developed unconsciously
by repetitive experiences. He then sets the ultimate aim of his general design theory as
the clarification of the human design ability in a scientific way. This view is essential to
his theory and is briefly explained below.
He models the world as a set of entities (e.g. an apple) and their attributes (e.g. color).
He asserts that humans construct abstract concepts about the classification of objects of
the real world. People have the ability to construct concepts about various
characteristics of entities and classify the natural entities according to these conceptual
characteristics. Each classification is a sub-set of the entities of the world (e.g. all sweet
objects) and includes the elements that share the same characteristics. A characteristic
represents a value, function, etc. (e.g. taste).
Once classification is done, people can identify entities not only by their direct image
(attributes) but also by using their characteristics, that is, by using the classification
(sub-set) they belong to. Many entities belong to several classifications at the same time;
they can be identified through Boolean operations on these classifications. Combining
these classifications can result in concepts that do not have any correspondence to a real
world, but may have a higher value than any existing entity. Yoshikawa asserts that
such “conceptual combination of abstract characteristics is the necessary condition for
Yoshikawa then states that the necessary condition for designing, that is, the act of
creating artificial things which do not exist in the real world, is the formation of
concepts of non-existing entities as the result of performing logical operations on
knowledge about existing things. His view of this logical operation is explained in the
previous paragraph. Designing through the logical operations on the conceptual sets is
the cognitive foundation of his theory. He then presents several definitions and axioms
and proves several theorems based on his assumptions.
Yoshikawa’s approach is mathematical and involves assumptions of an ideal situation
that may differ from the actual design practice. For example it utilizes the concept of
continuity which guarantees that a small change in the design description will result in a
small change in the artifact functionality and vice versa. GDT is a descriptive model,
but it also provides some guidelines (e.g. appropriate data structures) for the
development of CAD systems. Reich states that one concrete conclusion of GDT is
about the way attributes must be presented in CAD systems. According to GDT,
attributes are defined by the entities that have them. This is not as intuitive as the more
common method of defining an entity by the list of its attributes [Reich 1995].
2.2. Review of Product Configuration Design Research
Mechanical design methods which are related to AD can be sought in product
configuration design research, which encompasses such topics as modular design,
product family and platform design, design for mass customization, interface design, etc.
The configuration design stage refers to the determination of a product’s architecture.
Architecture includes the layout of physical subsystems, their functions and
embodiments, and the overall shape of the product [Pahl 1988]. The configuration
design is not treated as a separate stage in the design process by some researchers (e.g.
[Sydenham 2004]). In this case the activities that comprise the configuration design
stage are embedded or implied in other stages of the design process. The reason is that
the embodiments of assemblies and components and their layout in a product are often a
by-product of other design activities, and the product’s configuration is not explicitly
designed. This is despite the fact that many life cycle performances of a product are
influenced by the way its assemblies are organized. In order to materialize the potential
benefits of product configuration design, two elements of a product’s configuration
must be considered: the physical structure and the functional structure.
2.2.1. Functional and Physical Structures
The original functional requirements of a design problem are invariably decomposed
during the design process, resulting in a hierarchy. This hierarchy represents the design
rationale. It shows how functions are decomposed, what decomposition scenario is
selected, what auxiliary or technical functions are introduced at each level, and at what
stage the decomposition of a function ends (Figure 2.1-a).
The end-node functions are directly realized by independent physical parameters
without further decomposition. Functions within the hierarchy are also related to each
other through functional interactions, which are various logical and physical
relationships in the exchange of material, energy, and signal [Pahl 1988]. These
relations can be represented in a graph (Figure 2.1-b). The functional structure is a
representative of both the design rationale (decomposition hierarchy) and functional
interactions (graph) (Figure 2.1-c).
a) decomposition hierarchy
(DF: decomposed function)
b) interaction graph of end-node functions
Overall Function
c) complete functional structure
Figure 2. 1: The functional structure.
The physical structure refers to the hierarchy of physical assemblies, subassemblies,
and components. The physical structure includes the overall layout of assemblies and
components, as well as various relations that exist among them. Similar to the
functional structure, these relationships can be represented by a graph.
Therefore, in conventional mechanical design the configuration design of a product is
often subservient to decisions made during conceptual and detail design stages. That is,
the hierarchy of assemblies and their layout are typically designed in such a way as to
assure the proper functioning of the product. The departure from this practice can be
seen in design methodologies that treat product configuration design as an explicit
design activity. For example modular design develops assemblies and their layout with
the objective of achieving further benefits with respect to the product’s life cycle [Gu
The adaptable design presented in this thesis also extends from conceptual design to
product configuration design; it involves the determination of the overall architecture of
products. The objectives of product configuration design can be categorized as relating
to the physical functions of the product, to aesthetic or ergonomic factors, or to life
cycle factors such as production, serviceability, and recycling [Gu 2004]. The following
sections review some research approaches that aim at the explicit design of the
configuration of mechanical products.
2.2.2. Modular Design
Modular design is a design methodology that aims at developing a product architecture
consisting of distinct sub-systems in order to achieve a set of perceived benefits ([Gu
1999], [Marshall 1999]). Modular products fulfill various overall functions through the
combination of distinct building blocks or modules [Hillstrom 1994]. If the common
practice in a domain is to design integrated products, the application of modular design
requires substantial changes to the existing design procedure. For example,
modularizing a car into separable modules or an engine block into separable cylinders
requires a departure from the current design practice. If, however, products in a domain
naturally consist of distinct sub-systems, modular design is simply considered to be a
good design practice. Examples include the design of detachable speakers for stereo
systems or detachable gearboxes for electro-motors.
A modular product, unlike an integrated product, has its components clustered into
distinct sub-systems (modules) so that these modules can be designed, manufactured
and assembled separately. Modules can be physically detached from the overall product
to be repaired, recycled or upgraded. They may be used in other products that utilize the
same or similar modules, or may be arranged in different configurations to obtain
several functions from the same overall product. The modules of a product, depending
on its size and complexity, may in turn have their components clustered into smaller
modules to form a hierarchical modular structure.
It is difficult to precisely classify products into modular and non-modular because
various levels of modularity may be observed in the construction of products. Ulrich et
al observed that good modular designs usually exhibit some common characteristics
[Ulrich 1991]. First, a modular product is constructed by a set of compatible basic units,
or modules, which can be used to construct a variety of other products. These units can
be general units that are used in various products, or specific units for particular models.
They can be introduced during the initial design, or they can be introduced as new
technologies become available or new demands arise. Second, the interface among units
must allow for simple assembly and disassembly, and must be compatible and
consistent across various relevant units. Third, a modular product is constructed in two
phases. The first phase is the design and production of basic modules. These modules
are general and are not specific to a single product. The second phase is the design and
construction of completed modular products using the basic modules developed in the
first phase, and if required, other parts or procedures specific to the product. The second
phase may be undertaken by the producer of the basic units, or by other parties. This
discussion reveals that the two fundamental issues of modular design are the
development of modules (clustering of components), and the development of interfaces
among them.
Clustering (Segmentation)
Most modular design approaches provide methods for finding the optimal scenario for
the clustering of components of a design. Gu et al have presented a multi-objective
optimization method for the clustering of components for diverse life cycle objectives
([Sosale 1997], [Gu 1997]. In their approach, the relationships among components are
mathematically represented by an interaction matrix. The quantitative values of cells of
this matrix indicate how strongly two components (rows and columns) must be in the
same module based on various life cycle criteria called perspectives. For example: for
the objective of recycling plastic components should be placed in the same module,
while for the objective of repair frequently failing parts should be clustered in the same
module. They use a genetic algorithm method to find the optimal scenario with respect
to several objectives that may be in conflict with each other. The use of genetic
algorithm simplifies the combinatorial optimization problem, as discussed in [Kamrani
2003]. Yu et al have developed a similar approach using genetic algorithms for the
partitioning of a product into an optimal set of modules, and have applied their
clustering method to the modular design of an industrial gas turbine [Yu 2003]. Various
approaches to the clustering of components usually include some or all of the following
general steps of a basic modular design methodology:
Establish the functional structure of the product.
Determine the solution alternatives adopted for each function.
Establish the life cycle objectives and determine their relative importance.
Decide on the level of abstraction in referring to design components.
Build the relationship matrices and quantify their elements (interaction analysis)
Execute algorithms to find the optimal clustering scenario.
Modify shapes and interfaces to comply with the proposed product architecture.
Modify the design data (interaction matrix) based on new relations between
modules and iterate the execution of algorithms.
Evaluate the improvement of the final design with respect to the objectives of
Gu and Slevinsky have discussed the design of interfaces for the assembly of modules
in the overall structure of modular products [Gu 2003]. They emphasize that the
connections between modules must be designed to facilitate the separation and
attachment of modules for parallel manufacturing and assembly, as well as for
performing post product life activities. They present the concept of Mechanical Bus for
facilitating modular and platform product design, and identify the characteristics and
features of mechanical bus interfaces. These include bus functions, locking and release
mechanisms, positioning and locating features, and so on. Sanchez emphasizes that the
standardization of interfaces enables developers to produce the latest components or
devise processes compatible with a company's existing products and processes [Sanchez
2002]. Such standardization of interfaces helps in using the same modules across a
portfolio of products. This results in savings for the producer because members of a
product family not only share design similarities, but also share process plans, parts,
assembly lines and other manufacturing infrastructures [DeLit 2003]. Hillstrom
presented a method that helps the designer clarify how interfaces between modules
influence module functions, and select the best interface location. His method utilizes
Suh's axiomatic design theory together with conventional DFMA tools (Design For
Manufacture and Assembly) [Hillstrom 1994].
Most current modular design approaches aim to develop modules using componentbased clustering. It is assumed that the interactions among various components can be
quantified, and thus can be represented by graphs or matrices. Despite the great benefits
of these techniques, there are also some drawbacks:
First, there is uncertainty about the values or numbers provided in interaction matrices
and they are typically determined by estimation. If several matrices are superimposed
via weighting factors, the numbers in the final matrix can significantly change when
different designers prepare the matrix. The data inside the matrix is fed into a clustering
algorithm. Depending on the sensitivity of the algorithm to the initial data, the final
results may vary substantially, resulting in low repeatability and hence low reliability of
the method. Reliability of the clustering method can be improved by applying stricter
discipline in data collection and refinement, and by the ‘preparation’ of the initial
matrix using data analysis methods such as the Analytical Hierarchy Process, and using
algorithms that are less sensitive to variations and errors in the initial data [Sand 2002].
Second, the clustering problem has combinatorial complexity. The problem of
clustering n components is equivalent to the problem of partitioning an n-element vertex
set. The number of ways the set can be partitioned cannot be calculated by a simple
formula and is a non-polynomial function of n [Stanley 2001]. For every candidate
scenario, a calculation of the objective function based on the interaction matrix is
required. Therefore, an exhaustive method to find the best scenario is computationally
impractical as n increases. Heuristics rules and non-deterministic optimization methods
have been utilized for the clustering problem; examples include genetic algorithms
discussed earlier, and fuzzy cluster identification [Tsai 1999].
Third, the optimal clustering of mechanical components for several life cycle objectives
may result in the development of product-specific modules that do not perform obvious
functions. These modules are very context-sensitive and cannot be used in a variety of
circumstances. This characteristic is particularly important for adaptable design, where
various applications usually require functional reconfiguration more than structural
reconfiguration. Hillstrom addressed this issue by including the functional structure data
to clarify the interface/function interactions, then using these interactions to determine
clustering of components and optimal locations for interfaces [Hillstrom 1994].
Fourth, these approaches often ignore the designer’s freedom to change the values
inside the interaction matrices. These values are based on the specifications and
properties of physical components; for example material type is a property that
determines the interaction value with respect to the recycling objective. The designer
often has freedom to change these properties, or even to change the solution principle
for a particular function at the conceptual level. One approach to addressing the design
freedom is to first find the interaction data within the matrix that are critical to the
clustering algorithm, and then replace the components that affect the data with
alternative solutions. The computer implementation of such a method is possible if the
functional and physical structures can be represented simultaneously (Please see
Appendix 1 and the pilot software TARRAUH). Then, the physical structure determines
the values inside the interaction matrices and the functional structure allows the
designer to choose alternative solutions from a library of design cases, where cases are
indexed according to their functions [Hashemian and Gu 1997].
Despite these shortcomings, modular design has been the most successful approach to
product configuration design. Various field applications of modular design
methodologies have been reported in the literature. For example, Cohen et al applied a
modular design method to the conceptual design of a modular robot in order to achieve
hardware flexibility, as opposed to the conventional methods that rely on software
flexibility [Cohen 1992]. Their strategy is to develop an inventory of basic modular
units, standardized and minimized in size, which will allow the user to configure the
most suitable robot geometry for a given task. Thus, modular robots with hardware
flexibility yield optimal arm geometries. This gives them an advantage over
conventional industrial robots that are used in flexible automation.
2.2.3. Product Family and Platform Design
Industry’s response to rising consumer expectations has been an emerging production
paradigm called mass customization ([Pine 1993], [Tseng 1996], [Rai 2003]). This
paradigm differs from the conventional mass production, which involves the
manufacturing of many identical products using large assembly lines. The approach
towards mass customization adopted most by companies is the development of
platform-oriented multiple product variants. Various design strategies such as design for
variety and product family design have become critical in this respect ([Rai 2003],
[Martin 2002]). A platform-based design strategy is commonly used to create a product
family for variety and mass customization.
Platform design is a specific type of modular design in which the objectives of
modularization relate to multiple products that are considered members of the same
family or portfolio. Platform design is based on the identification of common attributes
within the members of a product family. Many companies are utilizing product families
to increase variety, shorten lead-times, and reduce costs. Simpson states that the key to
a successful product family is the product platform from which it is derived, either by
adding, removing, or substituting one or more modules to the platform or by scaling the
platform in one or more dimensions to target specific market niches [Simpson 2003].
Du et al investigate the fundamental issues underlying product family development.
They introduce the concept of Architecture of Product Family (APF) as a conceptual
structure for the overall logical organization of a family of products. In their work, APF
constructs, which include common bases, differentiation enablers, and configuration
mechanisms, are discussed from both sales and engineering perspectives. Also
discussed are variety generation methods with regard to producing custom products
based on modular product architecture and configure-to-order product development. To
support APF-based product family design, a Generic Product Structure (GPS) is
proposed as the platform for tailoring products to individual customer needs and
generating product variants [Du 2001].
Members of a product family may have similar overall functionality, or they may share
only some of the same functionality. However, to some degree even seemingly
unrelated products can be built on a common platform [Gu 2004]. In the general sense,
a platform is any set of standardized parameters, which are maintained within a group
of products for compatibility ([Farrell 2001], [Simpson 2001]):
Process and Manufacturing Platforms – In this case the platform is not so much
integrated into the design of a product family; it is integrated into the process of
how the products are manufactured.
Component Standard Platforms – Component standard platforms unify
manufacturing issues in multiple products by using common components
whenever possible.
Modular Platforms – This type of platform makes use of modules between
several products so that common parts are used whenever possible [GonzalezZugasti 2000].
The methods and guidelines for the development of platforms and modules involve the
study of similarities among various products within a portfolio ([Gu 2003], [Zha 2002],
[Du 2001], [Tsai 2004]). The optimization of the modularity scenario can also be
performed with respect to other life cycle objectives ([Ishii 1994, 1995, 1998], [Gu
1999], [Yu 2003]). In mass customization, development of customizable features is
guided by the study of customers’ expectations. Tseng et al have suggested design by
customers for mass customization [Tseng 1998]. In their work, the design and
manufacturing capabilities of a company are represented in a product family
architecture, using which customers assert their needs and define variations from base
products. Issues related to product family development are also discussed in [Du 2001],
[Yu 1999]. They present the architecture of the product family as a conceptual structure
and the overall logical organization of constructs: base, differentiation enabler (module),
and configuration mechanisms. Such a modular architecture provides for configure-toorder product development, tailoring products to customer needs and generating product
Simpson et al proposed a model to design a product family based on the concept of a
scalable platform, which can be sized to provide necessary variants [Simpson 2001].
Otto et al presented an optimization-based method for designing product platforms that
takes into consideration both the technical performance requirements and the cost of the
product family. They define a product platform as the set of subsystems shared across
the products that are offered by a firm. They point out that there are two possible ways
to create a product family using a platform-based strategy: integral platform, in which
individually designed assemblies are attached to an integral platform shared among all
products to create different variants of the product family; and modular platform, a
strategy that uses a set of modules as a product platform that is used across the product
family. Then, the common modular platform along with other variant-specific modules
is used to generate the product family variants. They mention the advantages of the
modular platform strategy: reducing the number of module types and instances of each
type lowers the efforts required to design, produce, distribute, operate and maintain the
module instances throughout the life cycle of the product family ([Gonzalez-Zugasti
2000], [Dahmus 2001]).
2.2.4. Life Cycle Objectives of Modularity
Several life cycle objectives or benefits can be sought in the modularization of a design.
The following is a list of most important objectives of modularity, developed by Gu et
al [Gu 1997].
1. Splitting the Task for Parallel Product Development
Sequential development of a complex product may take an unreasonably long time;
therefore the overall task is decomposed into sub-tasks. This allows each sub-task to be
handled independently and performed in parallel with other tasks. Design of complex
products may involve several design teams consisting of several experts across different
fields. The design of a product can be modularized based on such splitting of the task.
As a result, a complex and lengthy project can be reduced to several smaller, focused
and specialized projects that can be performed simultaneously. Eppinger discusses the
use of matrix manipulation techniques for the partitioning of design tasks [Eppinger
1990]. In his approach the rows and columns of the data structure represent design tasks.
The data structure is a binary matrix, where the unity indicates that the task in the row
requires the information from the tasks in the column. The matrix partitioning is done
by rows and column swapping techniques to create a diagonal or a triangular matrix.
Issues related to this objective of modularization include:
Identifying key factors which determine the fate of the design process.
Reducing both perceived and actual complexity of design tasks.
Sharing engineering data at an earlier stage of product development.
Redefining the critical components.
Forcing the designers to organize their decision process.
Coordinating the different tasks.
2. Manufacturing
Higher production volume reduces the per-unit production cost. Through the
development of standard modules that can be shared among various models or among
products within a portfolio or product family, the manufacturer can increase the
production volume for identical units. This results in reduced design and production
effort, reduced inventory, and using the same production line set up and manufactured
component for different products. Further manufacturing benefits of modularity include:
Dividing a product into independent components allows design and production
activities to be specialized and focused.
Component sharing allows higher initial investment in production resources
because these higher costs are distributed across a large number of units.
Acquisition of improved production technology with low unit manufacturing
cost is possible.
The quality and reliability of products can be improved, while the production
lead time is reduced.
3. Assembly
The purpose of modular design for assembly is to modularize a product (only one, not a
variety of products) into sub-assemblies that can be produced and assembled in isolation.
These sub-assemblies are then assembled together to build the overall product. This
type of modularity can be of paramount importance for large and complex designs that
may involve considerable assembly time.
4. Repair and Maintenance
Modularization creates segmentation in the structure of a product. Therefore, when the
product fails, diagnosis can be performed methodically as modules can be examined
individually. Once the faulty module is identified, then its components can be examined
and individual components can be repaired or replaced as necessary. The replacement
of modules and components is simplified as a modular structure generally facilitates
disassembly. Therefore, modularization generally helps with fault detection in real time
and reduces down time. Similarly, modularization can help with the maintenance of a
product through systematic maintenance procedures and easy access to various modules
and components. Modularization may be performed explicitly for the benefit of repair
and maintenance. For repair, frequently-failing components can be grouped in a same
module; in the event of failure this module is replaced with a spare one and the repair
can be performed while the product stays in operation. For maintenance, the
components that require the same frequency of service may be grouped together,
allowing for easy organization of maintenance schedules.
5. Upgrading and Renovation
Today's highly competitive market and high consumer expectations demand
manufacturers to introduce a variety of new models in short periods of time. Often a
new model has to reach the market quickly, and may not permit sufficient time for the
design of the new model from the beginning. In such circumstances only a redesign is
possible. Also, if there is not adequate time for a new manufacturing set-up, the existing
production layouts and even component inventory have to be utilized. Modularization
allows companies to introduce new models to the market quickly and inexpensively as
only a few modules have to be redesigned and reproduced, while many modules,
platforms, and process plans can be reused from an earlier design.
6. Product Variety
Various products frequently share similar functions among their constituent components.
Therefore, it may be possible to develop standard modules for these common functions.
Consequently, a variety of models may be easily developed through the combination
and configuration of these standard modules, and if necessary, any model-specific
modules. This benefit is the most common motivation for modularization and is often
implemented in the form of platform design and add-on modules.
7. Recycling and Reuse
Different components of a product are made from different materials and require
different processes for their recycling. The clustering of components with similar
recycling properties into a single module greatly facilitates recycling as it makes it
possible for a large module to be recycled without any disassembly of its components.
Also, the service life of durable components within a product or the components that do
not undergo much wear and tear can be substantially longer than the service life of the
overall product. Such components can be salvaged and reused when the product is
recycled or disposed of. The clustering of long lasting components in the same module
increases the possibility of reuse and requires less disassembly for part recovery.
8. Customization
Consumer tastes are often highly diverse. This complicates the design and
manufacturing of a product that is satisfactory to all customers. Modularization is a very
efficient technique for product customization. Via the rearrangement, swapping,
replacement, and addition of common modules, as well as using specific modules
developed for specific product demands, the producer is able to develop various product
features and functions for mass customization or make-to-order customization.
2.3. Discussion
Configuration design approaches discussed in this chapter do not make a reference to
design for adaptability. Instead, they are concerned with particular benefits of product
configuration design such as the development of product families for mass
customization or the development of modular products for life cycle objectives. The
idea of adaptable design as a new design paradigm for the development of engineering
products was proposed by Gu [Gu 2002]. The important difference between adaptable
design and the existing product configuration design approaches is that in the AD
paradigm adaptability of a design is treated as a distinct design characteristic.
The next chapter discusses various types of adaptability in detail. These types of
adaptability are applicable to a specific physical product, to a design from which
various products can be developed, or to both. They also include both the adaptations
that happen over the course of time and the adaptations that extend the scope of
application for a product or for its design. Such an extended concept of adaptability
includes the existing approaches towards increasing variety, customization, upgrading,
and even versatility in engineering products. Several of these approaches were reviewed
in this chapter. They all aim at increasing the adaptability of a design for specific
objectives known at the beginning of the design process. This thesis extends adaptable
design to include general adaptability, in which specific objectives are not targeted a
priori. Further, all these approaches are based on the modularization of designs,
whether it is for life cycle performance of a single product or for the reduction of
production cost via sharing product platforms and modules. The general adaptability
proposed in this thesis uses functions as the main criterion in the modularization process,
resulting in a new product architecture in which every product is a temporary assembly
of functional modules.
Chapter 3: Adaptability in Designs and Products
The ultimate goal of AD is to extend the utility of products and their designs. Utility of
a product can be extended over the course of time, or it can be extended in the scope of
applications; adaptable design can facilitate adaptations to foreseen changes or to
unforeseen changes; adaptations may occur for a variety of reasons such as the
upgrading of obsolete components or the customization of functions; and adaptations
can be applied to a physical object or to a design (blueprint) from which a variety of
products can be developed. This section discusses these concepts.
Section 3.1 discusses the environmental aspects of this research and presents a view in
which adaptation is regarded as the extension of the utility of artifacts. Section 3.2
utilizes this view to categorize various types of adaptabilities related to engineering
design. Section 3.3 describes the characteristics which make a design suitable for
adaptability. Section 3.4 discusses the benefits of adaptable design for the user, the
producer, and the environment. Section 3.5 summarizes the chapter.
3.1. Extension of Utility
This section describes the environmental implications of adaptable design. Section 3.1.1
discusses some existing assertions that the environmental problems are critical and
therefore the inevitable trend of market is towards a service-based economy. This term
refers to a marketplace in which the scarcity of resources limits the volume of objects
we can produce, thus the purpose of engineering becomes the utility of objects not the
objects themselves. Section 3.1.2 discusses the extension of utility as a viable response
to the scarcity of resources, and views the adaptation of a product as one way of
extending its utility. Section 3.1.3 summarizes these discussions and concludes a logical
relationship between scarcity of resources and the need for AD.
3.1.1. Service-Based Economy
Recent decades have increasingly witnessed the depletion of natural resources and the
pollution of environment. The seriousness of environmental problems has become more
evident by such phenomena as global warming, which have raised questions about the
sustainability of the current global development ([Meadows 1992], [Mebratu 1998],
[Bhaskar 1995]). Various social, political, economical, agricultural, and industrial
practices contribute to environmental problems. These parameters are often intertwined
and a valid solution for environmental problems inevitably requires changes in almost
all of these practices. The scope of this discussion, however, is limited to the
environmental aspects of engineering production.
Increasing population and the daily dependency of modern societies on various products
have resulted in unprecedented engineering production volumes. Every year billions of
tons of virgin materials are transformed into various products, consuming energy and
producing waste and pollution in the process. These products eventually retire and for
the most part are discarded at landfills. Thus from the environmental point of view
‘engineering production’ is the transformation of natural resources into waste and
pollution. There are concerns that the current transformation rate is beyond Nature’s
rejuvenation capacity ([Meadows 1992], [Bhaskar 1995]).
Concerns about sustainable development have motivated various changes in practices
and processes related to the life cycle of products ([Housechild 1998], [WBCSD 1998],
[Lu 2003], [Tipnis 1998], [Rifera 1999]). The life cycle of a product can be divided into
(disposal/recycling). Reducing the environmental impact of engineering production has
been sought in all three phases of a product’s life cycle. For instance, production
processes have improved in terms of gas emission, energy consumption, and material
waste; products are designed to be more energy efficient and to cause less pollution;
toxic or non-recyclable materials are replaced with more environmentally-friendly
materials; and products are designed for recycling and reuse.
In addition to these life cycle considerations, there is another approach which aims to
“reduce the total production volume” [Hashemian 2004]. Since maintaining the
convenience and quality of life in developed societies requires a certain number of
products to be in service at any given time, lower production volume means longer
service life for existing products. The need for more service with fewer products,
primarily caused by the scarcity of resources, increases the tendency towards a servicebased economy, which is believed to be the market trend of post-industrial societies
([Tomiyama 1995], [Seliger 1997], [Tomiyama 1997], [Seliger 1998], [Tomiyama
1999], [Tomiyama 2000], [Fujimoto 2003]).
The premise of a service-based market is the dematerialization of services, assuming
that services cause less environmental harm than physical products do ([Oksana 2000],
[Brezet 2001]). From a practical standpoint, the dematerialization of services means
obtaining more usage from fewer manufactured products, that is, using products to their
fullest potentials. The increasing of the usage of products is called the ‘extension of
utility’ in this thesis. Also, service-based market represents a utilitarian view towards
engineering. In this view an artifact is created for the purpose of performing a function
and not so much for aesthetics or other properties related to the physical object. This
argument will be utilized in Chapter 4 to propose the ‘frame-and-function’ architecture,
in which a product is constructed as a ‘temporary assembly’ for performing a function.
In this architecture, only the function is important and the product itself will be
dismantled or reconfigured for new usage after its current function is no longer needed.
The next section discusses the extension of utility and its relation to designing products
for adaptability.
3.1.2. Extending Utility through Adaptation
The utility of a mechanical product can be extended in two ways: by prolonging its
service life in its normal operational mode, and by adapting it to new operational modes.
The first method is applicable when the need for the current function remains longer
than a product’s life; the second method is applicable when the product’s current
service becomes unwanted while the product is in good working condition.
Extension of Normal Service: Durability
Products retire when they can no longer satisfactorily deliver their expected functions
due to deterioration, damage, wear, aging, etc. The service life of these products could
be prolonged either by “designing out maintenance” or by “designing for maintenance”
[Markeset 2001]. Design of maintenance-free products is influenced by quality,
reliability, and durability characteristics on one hand and by cost and technological
limitations on the other hand. Design for maintenance requires risk analysis, study of
possible failure scenarios, and making provisions for easy disassembly and repair of
frequently-failing parts [Gu 1999]. Regardless of maintenance issues, in this thesis the
prolonging of normal service is called ‘durability’. Durability extends the utility of
products only in their current operational mode. It becomes irrelevant if the termination
of service is not due to deterioration but due to changes in service requirements.
Extension of Utility to New Services: Adaptability
Many products are discarded while in good working condition because their normal
operation is no longer desired. The most common reason for such premature retirement
is technological obsolescence. Millions of products, especially those containing
electronic parts, are retired annually due to their rapid obsolescence and the
proliferation of new models. Other reasons include changes in the needs or expectations
of the user, changes in operation regulations, and changes in the working environment.
In these cases, the utility of a product can be extended only if it can be adapted to the
new service requirements. Therefore, the process of adaptation can be viewed as the
extension of the utility of an artefact to new services.
Extension of Utility within Environmental Approaches
Figure 3.1 classifies various approaches towards reducing the environmental impact of
engineering production. These approaches are divided into two categories: improving
product life cycles, and reducing production volume. The first category includes the
conventional methods applicable to the three phases of a product’s life cycle. The
second category leads to the extension of utility via durability and adaptability as
discussed earlier in this section.
In the categorization depicted in Figure 3.1, adaptability belongs to the methods of
‘reducing production volume’. However, it is also logically close to the methods of
‘recovery’ among conventional life cycle approaches because both recovery and
adaptability create ‘new usage’ for products. Therefore, recovery methods (recycling
and part reuse) are sometimes called adaptability in the literature [Willems 2003]. In
this thesis, however, adaptation does not refer to material recycling and part reuse;
which do not reuse several components of a product in a new operational mode. This is
based on the three limits which this thesis has adopted for the definition of adaptability
(Chapter 1). The differences between recovery and adaptability will be further clarified
later in this chapter when the environmental benefits of AD are discussed.
Life Cycle
Production (supply chain)
Service (operation)
Proper Disposal
Reduce Environmental
Impacts of Engineering
(New Service)
(Same Service)
Changes in technology
& consumer culture
Figure 3. 1: AD within the spectrum of environmental approaches.
3.1.3. Discussion
There is no general consensus on the seriousness of environmental problems. Therefore,
there is no general agreement that the service-based economy will prevail. This is
evident by numerous products which are designed to be company specific, and by
intense competition for higher productivity and market share instead of increasing
quality and adaptability. These issues have been extensively debated in the literature
([Mebratu 1998], [Hashemian 2004], [Meadows 1992], [Bhaskar 1995]). Regardless of
these debates, it is logically evident that the scarcity of resources makes it unaffordable
to manufacture a large number of products. If such conditions occur, the emphasis of
engineering will inevitably shift from having many objects to obtaining maximum
usage from fewer objects. The extension of usage of products requires them to be
durable for their normal service, or to be adaptable to varying circumstances. Thus this
section established a logical link between the scarcity of resources and the need for
adaptable design. This discussion is summarized in Figure 3.2.
Scarcity of Resources
Service-Based Economy
(More Service, Fewer Objects)
Extension of Utility
(In both time and scope)
The Need for Adaptability
to Varying Requirements
Figure 3. 2: The relation between scarcity of resources and the need for adaptability.
3.2. Categories of Adaptabilities
The adaptation of a product to new services can be viewed as the extension of utility, as
discussed in the previous section. This section utilizes this view to categorize various
types of adaptations. First, adaptations are divided into design adaptability and product
adaptability depending on which item is the subject of adaptation. Then adaptations are
divided into sequential and parallel adaptations depending on whether the extension of
utility occurs over the course of time or it is unrelated to time. Then this section
discusses specific adaptability and general adaptability which were mentioned in
Chapter 1, and explains the four categories of specific adaptability. The section provides
a summary of these categorizations at the end.
3.2.1. Design Adaptability and Product Adaptability
The design process results in the ‘description’ of an entity (part or product) that, when
materialized according to this description, can fulfill a required set of functions. This
description can be in the forms of blueprints, instructions, CAD models, prototypes, etc.
Therefore, the design process and the subsequent production result in the creation of
two entities: a design and a product (Figure 3.3). In ‘design for adaptability’, both of
these two entities can be made adaptable.
The Design Process
Adaptation Task:
Design Adaptability
Design Realization
Adaptation Task:
Product Adaptability
Figure 3. 3: Design Adaptability and Product Adaptability.
Design (Producer) Adaptability
A design is a set of instructions for the creation of a product or many identical products.
The same design, with minor changes, can be used to create different products, usually
in the form of variations of the original product within a product portfolio. Such design
reuse is called design adaptability in this thesis. Design adaptability results in the
creation of a variety of designs based on a common adaptable blueprint, and in the
upgrading of new models through the modification of old designs.
The realization of a design into a physical product is performed by the producer;
therefore design adaptability mainly concerns the producers and can be alternatively
called producer adaptability. Design adaptability is in fact the reuse of design
knowledge including proven concepts, components, methods, and processes. Thus it
yields significant benefits in terms of design cost and time. Further, the adaptation and
reuse of an existing design enables the producer to utilize the existing production
processes and even the existing inventories of manufactured components, hence
yielding further savings in production costs. Although design adaptability may not be of
importance to the user who owns a single product, the design similarity between the
user’s product and other popular models reduces training time for operating different
models, and enables users to swap components for repair and upgrading purposes.
Figure 3.4 shows two models of SONY video cameras that are developed from a single
adaptable design. Design adaptability in these cameras is achieved through a shared
platform which allows the attachment of various modules for different functions.
Although these cameras are produced from an adaptable design, an individual physical
camera is very difficult to modify for a new usage by the user. This example
demonstrates that design adaptability primarily concerns the producer.
Figure 3. 4: An example of design adaptability (courtesy of SONY).
Product (User) Adaptability
The second type of adaptation, which is more directly related to the environmental
benefits discussed earlier, is product adaptability. It refers to the ability of a single
physical product to be used for different service requirements. The adaptation task is
usually performed by the user, so it can be also called user adaptability. User
adaptations include the upgrading and customization of products as well as the
attainment of several functions from a single versatile product. An example of product
adaptability is the development of adaptable homes, where adaptations happen
continually through time as the needs or the lifestyle of the residents change (Figure
3.5). This structure allows for future additions, provides extra space for possible future
use of wheelchairs, and facilitates changes in features and functions of various parts in
and around the house. The adaptable design housing etiquette requires that
modifications be simple to carry out and become cost effective when they are planned
into the initial design of the house. The Adaptable Design Guidelines of the City of
North Vancouver state: [Vancouver 2000]
“Adaptable Design will create livable residences for a wider range of
persons than current housing design permits. Through consideration of how
adaptations could be easily and inexpensively incorporated at a future time,
Adaptable Design will allow for changes which are required by residents
with varying or changing needs, thereby supporting independent living (for
those with moderate disabilities)…. new developments and technology may
result in equivalents that meet the intent of a specified requirement.”
Figure 3. 5: Adaptable homes (courtesy of the City of Vancouver).
3.2.2. Sequential and Parallel Adaptations
The extension of the utility of a product or design may occur over the course of time, or
it might be unrelated to time. These two categories of adaptations are called sequential
and parallel respectively.
Sequential Adaptations
A sequential adaptation occurs over the course of time; it is usually the result of the
emergence of new technologies or the time-related changes in service requirements.
Sequential adaptability extends the ‘service life’ of a product or a design, and it may or
may not be reversible. An example of the sequential adaptation of a product is seen in
adaptable homes (Figure 3.5). A historical example of sequential adaptation of a design
is shown in Figure 3.6. This figure shows the evolution of a design, where the design of
the V2 ballistic missile developed by Wernher Von Braun and others in Germany
during World War II was later adapted to produce the design of the American Redstone
missile, which eventually evolved to the design of the Jupiter and Saturn space rockets.
[Wikipedia Encyclopedia,].
Saturn V Space Rocket
V2 Ballistic Missile
Figure 3. 6: An example of sequential design adaptability.
Parallel Adaptations
A parallel adaptation extends the usage of a design or a product into various
applications. Parallel adaptability is unrelated to time; it extends the ‘service scope’ of a
product or a design and the adaptation is usually reversible.
Parallel adaptability of a ‘product’ means that the same product can be set up in various
ways to perform different functions. The adaptation of a product to various usages is
performed by the user in a usually reversible and simple procedure. This typically
results in the development of versatile products which are capable of performing several
functions. In parallel adaptation of a ‘design’, the same design is adapted by the
manufacturer to produce a variety of products for different customers, or to produce
customized products at almost the same price as the standard models. The
diversification of a product portfolio through the parallel adaptation of a ‘design’ is
known as mass customization in the literature, as discussed in Chapter 2. Figure 3.7
shows three Ford vehicles that share not only the same general design, but also various
parts and assemblies such as the chassis, headlights, engines, and accessories.
Ford Explorer Sport
Ford Explorer Spot Trac
Ford F150 Pick-Up
Figure 3. 7: Parallel design adaptability, performed by the manufacturer, results in variety and
mass customization (Courtesy of Ford Motor Company).
3.2.3. Specific and General Adaptabilities
Unlike the conventional design process in which a product is designed for a nominal set
of functions, AD develops products that can also be adapted to different or additional
functions beyond their normal operational mode. Therefore, it seems that the designer
must have some ideas as to what these additional requirements are, and design the
product accordingly. In many cases such forecast information exists and is utilized, for
example in the design of video cameras shown in Figure 3.4. This is called “specific
adaptability” because the provisions in the design are made for specific adaptations
which are known in advance. However, it is also possible to design products in such a
way that they are generally more adaptable than conventional designs even if no
forecast information is available. This is called “general adaptability”. Specific and
general adaptabilities were discussed in Chapter 1. This section further discusses the
four categories of specific adaptability: versatility, upgrading, variety, and
customization. It should be mentioned that these categories are not mutually exclusive
and there is an overlap between them.
Design for Versatility
If the additional functions which are expected from a product are definitely known
during the design process, the product can be designed to deliver multiple functions
including the original FRs and the additional FRs. Since in design for product versatility
the additional FRs are treated the same way as the original FRs, it could be treated as a
conventional design process. In this thesis, however, design for versatility is considered
as a category of AD in which the maximum amount of forecast information is available.
A product is designed for versatility if adaptations from one function to another occur
frequently; therefore the product is designed so that these adaptations do not require
significant alteration of the product and often involve a simple procedure which can be
performed by the user.
Design for Upgrading
Upgrading is the adaptation of existing designs and products to new needs or
technologies as they become applicable. An example of design for upgrading is the
design of computer systems. An individual computer is upgradeable by the user because
its rapidly-expiring components are designed as easily replaceable units. A computer is
also upgradeable in its ‘design’ as the architecture of a computer is designed to allow
the manufacture to incorporate new technologies.
Design for Variety
Variety refers to parallel ‘design adaptability’ thus it concerns the producer as discussed
earlier in this section. In design for variety a single design (blueprint) is used to produce
a variety of products. Design for variety is also called ‘design for mass customization’
or ‘product family/portfolio development’ in the literature. An example of design for
variety was shown in Figure 3.7, where several vehicles in the company’s portfolio
shared similar designs.
Design for Customization
Customization is a general term which refers to the adaptation of a product or design to
specific preferences. For example, various features and functionalities of SONY video
cameras such as the image stabilizer, digital zoom, and the side screen are designed in
the form of optional units (Figure 3.4). Then many models can be easily developed by
the morphological combination of various features in response to diverse customer
Figure 3.8 shows the role of forecast information in these four categories. It can be seen
that the availability of forecast information justifies more initial investment, as in
versatile products, while in the absence of such information less initial investment is
made because there is no certainty that adaptations will be required after the design is
Initial Investment in Adaptability
- Probability and frequency of adaptations
- Specificity of adaptation procedures
- Ease of adaptation tasks
Specific Information
Figure 3. 8: The relations between the availability of specific information and the justification of
initial investment in the four categories of specific AD.
3.2.4. Summary
According to the categorizations of this section adaptations can be parallel or sequential,
and they can be applied to a design or to a product. These two divisions create four
categories. The chart in Figure 3.9 shows these categories within Specific AD:
upgrading is the sequential adaptation of both designs and products, variety is the
parallel adaptation of a design, versatility is the parallel adaptation of a physical product,
and customization is a general term for all these categories. It can be seen that such
classifications are not made for general AD because general adaptability does not
predetermine any type of adaptation.
Unrelated to
Time (Parallel)
(New Models)
Adaptation of
a Product
(physical product)
Adaptation of
a Design
Adaptation of
a Product
Adaptation of
a Design
Applicable to all categories without
targeting specific adaptabilities
Figure 3. 9: Various categories of adaptations.
3.3. Design Categories Suitable for AD
Notwithstanding its benefits, design for adaptability also has some disadvantages. For
instance designing a product for adaptability might result in additional costs and the loss
of optimality in weight and performance. Therefore, it is important to decide if and to
what extent AD is applicable to a design problem. The final decisions require trade-off
analysis which takes into consideration the need for adaptability, the cost of adaptable
design, the cost or effort of performing an adaptation task, the frequency of adaptation
tasks, marketing, etc. A preliminary assessment, however, can be made based on certain
characteristics of the design problem at hand. This section discusses four design
characteristics which increase the applicability of AD.
Superfluousness in this thesis refers to the ‘unused potential’ of a product and is the
most important criterion in choosing a product for adaptable design. Superfluousness is
a result of a product’s idle time during its operational life, and more importantly, a
result of its premature retirement. Products without superfluousness are those which are
mainly ‘consumed’ by their usage. These products, such as drill bits, cutting tools, and
brake pads, should be designed for durability not for adaptability.
Product Variations
An important characteristic of a design is the features it shares in common with other
designs which are considered as belonging to the same family or portfolio. Such
commonalities make it possible to adapt a design from one product to a similar one,
thus justifying design for adaptability. Variations of a product can occur over the course
of time (new models, upgrades, and customization), or they can occur in parallel as a
company’s product variety and diversity. Products which have a larger number of
variants are generally more suitable for adaptable design. The common method of
designing such products for adaptability is the use of shared platforms and
differentiating modules as discussed in Chapter 2.
Environmental Impact
Some products contain materials that are hazardous or are difficult to recycle, reuse or
even dispose of. Products with larger environmental impact in their life cycle yield
better environmental benefits if they can be adapted.
Financial Significance
The financial significance of a design project depends on the production volume, unit
price, capital investment (infrastructure), and development time. High production
volume justifies more initial investment in design, including the use of adaptable design
techniques to improve the fate of thousands or millions of objects after they retire. Unit
price has a similar effect and it becomes increasingly important for massive one-of-akind projects such as a nuclear power plant. In such projects, both the physical product
and the design knowledge generated should be made adaptable for future use (product
adaptability and design adaptability). Capital investment and development time are
important when variety is involved. That is, by replacing multiple designs with one
adaptable design large capital investments and long development times will not recur
for each individual design.
An Exemplary Category
Military design and production projects often have all the above characteristics. Military
products typically have a long product life because they are designed for quality,
durability, and reliability; yet their actual usage time is often much shorter than their
potential service life. Therefore these products are superfluous. Military products also
have many variations for different deployment conditions, and they have variations
which evolve in the course of time. Also, some military products have high
environmental impacts because they utilize hazardous materials in their construction or
in their production processes. Moreover, military projects are often of great financial
significance due to large investments in design, development, tests, and refinements.
These characteristics make military products a suitable category for adaptable design.
Some of these products could be designed for adaptation to civil usage during their idle
times or for permanent adaptation to civil usage after their obsolescence. They might
also be designed for upgradeability to new technologies and new demands, so that their
retirement is postponed. This is particularly beneficial when adaptation results in
avoiding the disposal of environmentally hazardous materials.
3.4. Benefits of Adaptability
This section discusses the benefits of AD for the user, the producer, and the
3.4.1. The User: Extended Product Utility
As discussed in Section 3.2.1, the user is mainly concerned with product adaptability.
In this category, an adaptable product should be designed in such a way that adaptation
tasks can be easily performed by the user. The user-related benefits of adaptability
result from the fact that an adaptable product replaces several products in its service life,
thus saving money, storage space, maintenance, installation costs, etc. It also provides
the user with the possibility for customizations which are not available in the market
(personalization). Figure 3.10 shows an example of a bicycle U-lock that is also a
carrier rack. This versatile design enables the user to replace two products with one.
Figure 3. 10: Product adaptability, performed by the user, results in multi-purpose versatile
machines. (Courtesy of Master Lock).
3.4.2. The Producer: Extended Design Utility
As discussed in Section 3.2.1, the producer is mainly concerned with design
adaptability. The producer may use the same design, its associated process plans,
manufacturing set-ups, and even existing parts and assemblies to produce different
products for different clients [DeLit 2003]. Unlike ‘user adaptability’ in which the
adaptation is performed by the customer, in this category the adaptation tasks are
performed by the manufacturer in the factory, where the required tools and expertise are
readily available. Therefore, the primary concern is the long-term benefits not the ease
of adaptation tasks. The producer’s benefits include the reuse of design knowledge, the
reduction of production time and cost via intra-company standardization, the reduction
in the cost of post-sale services, and gaining marketing advantage through user and
environmental benefits.
3.4.3. The Environment
The existing environmental remedies for the problem of product retirement are the
recovery techniques, which are based on redirecting the flow of used products from
disposal in landfill back into the production supply chain [Hashemian 2004]. Adaptation
of a product has a similar effect because it also redirects the retired product back into
new service. Figure 3.11 compares adaptability and other recovery methods. In this
figure, along the time axis the processes of the production supply chain occur from left
to right until the product is delivered to the user. Then there is a usage phase, followed
by the end point of service life (retirement) at the right end of the time axis. At this
point, various recovery methods return the product to different points on the time axis.
The figure shows the cumulative environmental impact (EI) on the vertical axis. For
simplicity, EI is assumed to be a linear function of production stages (represented by
time in this figure). Therefore, from any point on the time axis to the end of the
production chain, EI accumulates proportionally. It can be seen that the closer the
returning point of a recovery method is to the final product, the less EI is created to
finish the production (the area of triangles). Therefore, the priority of the recovery
methods can be listed in the following order: first, durability, repair, and maintenance
that extend the normal operation; second, adaptation that extends the service life in a
new operational mode; third, part salvage that reuses parts as they are; fourth, material
salvage that uses existing material for manufacturing new parts; fifth, material recycling
that involves shredding and reprocessing raw materials.
Part Salvage
Raw Mat. Processed Mat.
Usage by User
Production Supply Chain
Figure 3. 11: Adaptation versus post-retirement remedies.
3.5. Summary
The discussions of this chapter are summarized in Table 3.1. The left side of the table
lists various characteristics, types, and benefits of adaptability. The right side of the
table describes these characteristics and benefits for design adaptability and product
adaptability in two separate columns.
Adaptable Design
Characteristics of
Design Adaptability
Product Adaptability
The one who performs the
adaptation task
Producer, rarely user, by both
in large projects
User, rarely producer, by
both in large projects
Results of adaptation
Population of products from
same design
One product, several or
extended usages
Category for
the extension
of Utility
(in time)
Upgrading, new models,
Extended service life,
upgrading, customization
(in scope)
Product variety and mass
Versatility of products
Variety of choices at lower
costs, product familiarity
- One product replaces a
few, various savings
- Upgrading and
- Adapting to changing
needs of the user
- Lower cost and time of
- Better market share due
to user benefits
- Quick Market response and
technology update
- Better market acceptance
due to environmental
- Market share due to variety
and mass customization
Better use of resources during
production, part salvage due
to shared modules
More service with fewer
Less waste and pollution
Table 3. 1: The relationships among various aspects of Adaptable Design.
Chapter 4: Design for Adaptability
This chapter presents this thesis’s method of designing products for adaptability.
Section 4.1 discusses the theoretical arguments of the function-based segmentation
approach for general AD. Section 4.2 discusses the difficulties of applying this
approach to the design of mechanical systems. Section 4.3 presents a ‘measure’ for
adaptability based on the concept of information content. Section 4.4 presents methods
and guidelines which help with the adaptable design of mechanical systems. Section 4.5
describes the overall methodology of design for adaptability including both specific AD
and general AD. Section 4.6 discusses the extension of this methodology for the
inclusion of life cycle design. Section 4.7 provides a brief summary of this chapter.
4.1. Fundamentals of Design for General Adaptability
This thesis divided ‘design for adaptability’ into specific AD and general AD. Specific
AD is an umbrella term which refers to the methods of designing a product for a
predetermined set of adaptations. General AD, on the other hand, refers to designing
products for general adaptability without targeting a set of predetermined adaptations.
Methods of specific AD are straightforward because their adaptability requirements are
determined from the outset and these predetermined adaptations guide the design
process. For example, in design for upgrading the rapidly expiring components of a
product might be designed as replaceable modules, and in design for variety a family of
products might be developed from a shared platform and several differentiating
modules. These methods were discussed in Chapter 3.
In design for general adaptability, however, predetermined adaptations cannot be used
to guide the design process because such forecast information is unavailable. Therefore
general AD requires a different approach. Chapter 1 briefly mentioned that the approach
of this thesis is to emulate the adaptability of a rational functional structure in the
physical structure of a design. This method, which does not depend on forecast
information on future adaptations, could be called ‘function-based segmentation’, or
‘subordination of the physical structure to the rational functional structure’. This section
discusses the theoretical reasoning for this approach. Although some arguments of this
section are applicable to any type of design, it should be assumed that the scope of
discussions is limited to mechanical engineering design.
4.1.1. The Design Hierarchy
A system may be defined by describing its functions, and without detailed specification
of its mechanisms. Complex systems might be expected to be constructed in a hierarchy
of levels. Then, similar to the whole system, a subsystem may be defined by describing
the functions of that subsystem, and without detailed specification of its submechanisms. The subsystems of an artifact at lower levels nest themselves according to
a hierarchical schema in a semi-independent way, each performing the function they
were designed for at the level they are nested in the hierarchy. The design of each
component can then be carried out with some degree of independence from the design
of others, since each component will affect the others largely through its function and
independently of the details of the mechanisms that accomplish the function. This
hierarchical ordering is the “shape of design” [Simon 1969].
A hierarchical system is not an aggregation of elementary parts, and in its functional
aspects is not a network of elementary units of behavior. It is a system that is composed
of interrelated semi-autonomous subsystems, each of the latter being, in turn, hierarchic
in structure, until we reach some lowest level of elementary subsystems. A hierarchical
structure can be represented by a multi-leveled organization of nodes, where each node
branches into sub-nodes to form a tree. Each node within the hierarchic tree has two
properties: it is a whole relative to its own constituent parts, and at the same time a part
of the larger whole above it in the hierarchy. This dual characteristic has been called the
‘Janus effect’, named after Janus the two-faced Roman god with faces looking in
opposite directions [Koestler 1967]. Any member of a hierarchy has a ‘whole’ face
looking down towards subordinate levels, and a ‘part’ face looking up towards the apex.
Not being satisfied with words such as sub-whole or sub-system, Koestler coined the
term ‘holon’, from the Greek holos (whole) and on (part), to designate a node in the
hierarchic tree.
Scale Independence
An important property that can be deduced from the above discussion is the scale
independence of a hierarchic structure. That is, every holon in a hierarchy is bound by
the Janus effect regardless of the level it is positioned in the hierarchy; this includes the
apex, the middle holons, and the end-node holons. The apex of a given hierarchy, itself
a holon, is a node of a larger hierarchy from which it was dissected; the end-node of a
hierarchy, also a holon, can be decomposed further into its constituent elements.
The scale independency of hierarchical structures raises questions as to which holons
are the apex and the end-nodes of a given hierarchy; that is, where a hierarchy begins
and where it ends. For a design hierarchy, these determinations depend on the
circumstances of a design task as explained below.
Apex and End Nodes
The apex of a design hierarchy is the system which is chosen for the task at hand; its
parent systems are considered as the working environment of the chosen system. For
instance the design of a vehicle might be the apex in a design problem, that is, ‘vehicle
design’ initiates the task. The vehicle itself is a part of a larger system for urban
transportation, involving the design of lane widths and bridge capacities. To the
designer, however, the system of interest is the vehicle and road characteristics are
considered as the constraints or attributes of the vehicle’s working environment. As a
contrasting example, the apex of another design problem might be the design of a
gearbox for a vehicle; in this example the vehicle is considered as the working
environment for the gearbox system.
The end-nodes are decided by a relatively arbitrary decision as to where to stop the
decomposition process. In engineering design the end nodes are the unambiguous
description of an artifact. These descriptions are then communicated to another entity,
for whom these descriptions form the apex of a new problem. This issue is further
discussed in the next section, which explains the internal mechanism by which a
hierarchic tree is formed in the design process.
4.1.2. Decomposition and the Design Holon
The Design Holon
The previous section discussed three properties of a design hierarchy: the nodes in the
tree structure are ‘holons’ bound by the Janus effect; the hierarchy is a self-similar and
scale independent structure; and the apex and end holons are decided arbitrarily. The
self-similarity of a design hierarchy means that it is constructed from a recurring pattern.
This pattern, which is a general template for a single node or holon, is called the design
holon in this thesis. From the Janus effect it can be deduced that a design holon should
have a mechanism for producing its subordinate holons, and its own existence should be
attributable to a higher holon from which it has been decomposed.
The design holon consists of three elements: a problem, a solution which is synthesized
for this problem, and the decomposition of the problem into sub-problems based on the
chosen solution. The reason for considering the solution as an ingredient a design holon
is that, in the decomposition schema adopted in this thesis, decomposition is performed
only after a solution for the problem is found. Suh has discussed the relation between
choosing a solution and decomposing a problem [Suh 1990]. Describing the goals of a
design task by a set of functional requirements (FRs), he states: “FRs at the ith level
cannot be decomposed into the next level of the FR hierarchy without first going over to
the physical domain and developing a solution that satisfies the ith level FRs.”
The goal of a design problem is described by a set of functional requirements (FRs). In
order to achieve the overall goal, a solution must be devised for each FR. This process
includes synthesis, which involves creativity and experience, and evaluation, which
results in choosing a solution from several alternatives. Then, the realization of the
chosen solution imposes its own requirements, which are the conditions that have to be
met to assure the proper functioning of the chosen solution. Since these are the
requirements for the functioning of the solution, they are called the functional
requirements. These sub-FRs form new sub-holons, and this is the process of
decomposition in design. This discussion also reveals that the initial goals are also
called FRs because they too are the requirements of an adopted solution for a higher
system. This higher system, however, is beyond the scope of the design problem at hand
as discussed in the previous section.
Decomposition Rules
The decomposition of FRs into sub-FRs proceeds through “zigzagging” between the
problem space and the solution space, or between FRs and their solutions [Suh 1990].
The fact that a problem (FR) needs to be solved before it can be decomposed is an
important rule of decomposition adopted in this thesis.
The new FRs that are produced as the result of decomposing their parent FR should be
both necessary and sufficient for the attainment of the goal of their parent FR. The
sufficiency of decomposition means that when the solution for FRi generates n new
requirements, the fulfillment of these n requirements at the (i+1) level should guarantee
the proper functioning of the adopted solution for FRi. The necessity of FRs means that
every FR within the set created by the decomposition process needs to be explicitly
resolved by the designer in order for the chosen solution to function properly.
The process of decomposition in design changes the representation of the problem from
FRi to FR(i+1); which means from more uncertain and abstract FRs to more deterministic
and concrete ones. The decomposition process and the subsequent functional structure
are not unique. They depend on the formulation of FRs for any given node, and on the
solutions chosen for each FR. Figure 4.1 depicts the elements of a design holon. It can
be readily observed that this self-repeating pattern results in a hierarchy.
Figure 4. 1: The design holon consists of FRs, solutions, and decomposition.
Beginning from the apex, the decomposition generates new FRs in a hierarchical
fashion. The question is: where does the design process end? The answer lies in
reducing the uncertainty or increasing the specificity of FRs to the level of ‘available’
resources. In mechanical engineering, a design is decomposed to a level of specificity
where FRs can be achieved by available technological resources. This is the end holon,
and the designer need not be concerned with how such specific goals are fulfilled. The
end holons will be further discussed when the concepts of ‘task’ and ‘information
content’ are presented in Section 4.4.1.
4.1.3. The Rational Functional Structure
Using the concept of design holon, a hierarchical model of the design process can be
developed. In this model, the initial design problem is represented by a set of FRs. FRs
are assumed to be both necessary and sufficient for the representation of the initial goals.
For each FR, solutions are synthesized and a set of solutions for FRs is chosen. The
functional independence among the solutions within the set should be maintained, as
suggested by Suh’s independence axiom. Then, these solutions are decomposed into
their own functional requirements. The new FRs that emanate from a decomposed FR,
similar to the initial FRs of the design problem, have to be both necessary and sufficient
for achieving their goal. The goal of such a set of FRs is the proper functioning of the
chosen solution for their parent FR. Decomposition of FRs through this zigzagging
process continues, and the process ends when the blueprint for the creation of the
artifact is specified detailed enough to be sent to the manufacturer. This process results
in a hierarchical structure illustrated in Figure 4.2.
Initial Design Problem
Abstract FRs
Specific FRs
Non-Physical FR
Physical FR
Intermediate solution
physical solution
End-node solutions
Figure 4. 2: The rational functional structure.
In the above model, the initial FRs of the design problem are divided into physical and
non-physical FRs, denoted by (F) and (NF) respectively. Based on the nature of the
chosen solution, a FR is decomposed into other FRs which in turn can be F or NF. The
distinction between physical and non-physical FRs is important for AD because the
physical structure is subordinated to the hierarchy of physical functions only.
This hierarchy represents the design rationale, and is called a ‘rational functional
structure’ in this thesis. This hierarchy is distinguished from a more common definition
of functional structure which is a ‘descriptive’ representation of functions [Pahl 1988].
A rational functional structure has certain properties which make it adaptable. These
properties are discussed in the next section.
4.1.4. Causality and adaptability
In a rational functional structure described above, the relation between every node and
its subordinates is a ‘causal’ relation. That is, every subordinate node is a new FR
which is ‘caused’ by the chosen solution of its parent function, and there is no other
reason for this node to exist. This causality implies two properties when a node is
removed from the hierarchy: first, all its subordinates become unnecessary and can be
removed without affecting the rest of the structure; second, there is no need to eliminate
anything else. These properties make such a structure suitable for adaptability as
discussed in Chapter 1.
4.1.5. General Adaptability through Subordination
The approach of this thesis towards achieving general adaptability is to subordinate the
physical structure of a product to a rational functional structure described above. This
subordination results in the product having the same adaptability as its rational
functional structure.
In order to achieve this subordination, the architecture of a product should be designed
by the same rules which generate a rational functional structure. First, each physical
function should be fulfilled by a distinct subsystem. Second, each subsystem must
contain within it all the elements it needs for its proper functioning. Third, a subsystem
should not have any duties other than its nominal function. In this fashion, the
elimination of a subsystem does not affect the rest of the product.
The architecture of a subordinated system will be an assembly of autonomous
functional modules. Each module, in turn, is a system which can be also developed as
an assembly of independent modules. The nesting of this structure results in a hierarchy
which corresponds to a rational functional structure. It might not be practical, however,
to follow this division to detailed levels because small subsystems are often needed as a
whole, and there is no need to develop their insignificant components as independent
modules. The next section discusses the challenges of incorporating this architecture in
the design of mechanical systems.
4.2. The Challenge of Mechanical Design
Few mechanical devices can be adapted to varying service requirements without
considerable effort; therefore most mechanical devices stay in their normal operational
mode until their retirement. The lack of adaptability in mechanical systems can be
attributed to the two broad properties explained in this section.
The Nature of Mechanical Components (Structural Connectivity)
Adaptation is the modification of the internal mechanisms of a system in response to
outside variations. In engineering systems, adaptation is a response to the new
requirements for service or operation and invariably involves the modification of the
internal structure of the artifact. For example, the addition of a room to a house is the
internal modification that adapts the house to the new spatial requirements. Therefore,
the adaptability of a system is a reflection of its flexibility in allowing the necessary
internal modifications.
Performing structural modifications is particularly difficult in mechanical systems. The
reason is that the functions of mechanical components are achieved via their forms and
the geometrical or spatial orders among interacting components. Any modification in
the structure of a component, difficult and costly by itself, may also disturb the spatial
and geometrical relationships, and thus affect the function of several other components.
These components, which may be functionally independent from a logical point of view,
are often connected via various constraints such as size, shape, alignment, adjacency,
attachment, closure, motion, direction, etc. As a result, any modification is often
propagated throughout the product, hence making the adaptation process costly. This
property can be called structural connectivity.
Since structural connectivity is an inherent property of mechanical systems, the
available remedies for reducing propagation of changes are limited to two categories:
use of alternative technologies, and segmentation of the structure. The first category
basically avoids the use of solid components and their associated spatial constraints, and
replaces them with hydraulic, electronic and software systems. The second category is
based on the premise that in a modular structure modifications are likely to be confined
within segments and are less prone to propagating into other segments. Modular design,
platform design, interface and bus system design, and manufacturing adjustment design
are examples of design approaches for the segmentation of mechanical systems ([Otto
1994], [Lee 2003], [Chen 1994], [Yu 2003], [Gu 1997], [Gu 2003]).
The Nature of the Mechanical Design Process (Functional Ambiguity)
An engineering product is expected to deliver some physical functions related to
materials, energies, and signals; the design process involves identifying the required
functions and finding solutions for them ([Pahl 1988], [Suh 1990]). The relation
between a function and its solution is a causal relationship, which means that a solution
would not exist if its function was not required. Ideally, each function is performed by
its corresponding solution independent from other components in the product.
If the solution for a function is a complex system, this solution may impose its own
functional requirements for which the designer must find solutions as well. This is the
process of decomposition by which the design proceeds from abstract and complex
functions to more specific and simpler functions. At every level of decomposition, the
relation between functions and their solutions remain causal, as described above.
Section 4.1 discussed an ideal rational functional structure, which is a hierarchy of
causal relationships where every function in the hierarchy is only related to its parent
function and to its subfunctions. This section also discussed the adaptability of such a
structure: if a function is no longer required, it can be eliminated together with all its
subfunctions and their corresponding solutions, and there is no need to eliminate
anything else.
Unfortunately, the conventional mechanical design process does not result in an ideal
scenario, and the hierarchy of physical assemblies does not reflect the rational
functional structure for two reasons.
First, a mechanical design typically follows the ideal hierarchy for only one level of
decomposition at a time. If at any point of the design process the number of required
functions is n, the designer needs to find n solutions for these functions. There is a
causal relationship at this level and every solution exists because of its corresponding
function. Then, some of these solutions may impose their own requirements. The new
requirements for a solution can be caused by both the functional elements of the chosen
solution, and by the constraints that relate this solution to other solutions. The latter
may disturb the hierarchical tree of functions because a newly emerged function can be
listed under several higher functions in the hierarchy not just under its parent function.
As the design proceeds, various constraints become increasingly important in driving
the design process, and it becomes unclear how some new subfunctions are related to
the original design problem [Whitney 1993]. Therefore, if an initial design function is
eliminated or the solution for it changes, as is the case in adaptation, it is not always
obvious which components ought to be removed. For example, the grill in the front of a
car, a part of exterior design, is functionally related to the internal combustion engine. If
the engine is replaced by an electric or hybrid engine, there will be no functional need
for the grill.
Second, due to omnipresent constraints on weight, size, and cost, component sharing is
an integral part of the mechanical design process. Therefore, several functions at
various levels of the design hierarchy can be achieved via several design features
(design parameter or DPs according to Suh) of a single component or assembly. This
component sharing is often necessary because part redundancy in mechanical systems is
usually unaffordable. For example, the main function of an engine of a vehicle is to
provide power to the transmission system; but it also drives the alternator and water
pump, provides crash safety for passengers, and provides heat for the cabin in winter.
For these reasons there is often some ambiguity as to the function of mechanical
components. A component may contribute to several functions, yet it may not be
autonomous in fulfilling a single obvious function. This makes mechanical components
product-specific, as opposed to function-specific. This is the reason mechanical parts are
difficult to use in any other product than the original product they were designed for.
Current modular design techniques provide little help in this regard because their
clustering algorithms are based on physical parts not based on functions [Gu 1999].
Mechanical Design and Other Engineering Designs
Figure 4.3 shows the difficulties of adapting mechanical systems in comparison with
some other engineering systems. Software systems are the most adaptable category of
engineering systems. The connectivity among various parts of a software system is
limited to the exchange of input and output variables, and the modularization of
software is always based on functions. Electrical systems involve physical embodiments
and constraints; thus they tend to have some structural connectivity and may be
product-specific. Buildings and other structures related to civil engineering have more
structural connectivity than electrical systems, but their functions are generally clear,
and a civil engineering system can often be used in different designs and configurations.
Mechanical systems are the least adaptable, because mechanical design imposes both
structural connectivity and functional ambiguity.
Product-Specific Components
(Functional Ambiguity)
Figure 4. 3: Mechanical systems are generally less adaptable than other engineering systems.
A subsystem within a mechanical product may contribute to several functions while it
might not be sufficient for a single useful function because this function is obtained
from the work of several subsystems. As a result, the function of a mechanical
subsystem might be ambiguous and this subsystem might be usable only the context of
its original product. The physical attributes of a subsystem, as well as the design
requirements of its mating parts and interfaces with other assemblies, often make the
subsystem even more product-specific. Also, due to the presence of geometrical and
other physical constraints, the functional independence between two design parameters
does not mean that they are structurally independent or that they are even separate parts.
The example of a can/bottle opener ([Suh 1990], p51) satisfies the independence axiom,
but the two FRs (open can, open bottle) are obtained from two independent design
parameters that are structurally implemented on a single part. Further, the
decomposition of an FR into new sub-FRs in mechanical design is driven by both
functions and constraints; after a few levels of decomposition there will be many FRs in
the functional structure that have no obvious relation with the original design problem.
Due to these properties, the physical structure of a mechanical system typically does not
correspond to a rational functional structure (Figure 4.4). In such a case as the one
depicted in the figure, when a function is removed from the functional structure on the
left, it is difficult to identify and implement the necessary changes in the physical
structure on the right. This chapter will present methods and guidelines which help in
subordinating the physical structure of a mechanical system to the rational functional
structure as much as possible thus enhancing the general adaptability of designs.
Figure 4. 4: Functional and physical structures may not correspond.
4.3. Measure of Adaptability
The term ‘adaptability’ for a product refers to the ability of the physical product to be
adapted to a new operational mode; for a design (blueprint) it refers to the ability of the
design to be adapted to produce a new design whose physical manifestation can serve in
a new operational mode. In the first case, the subject of adaptation is a physical product,
and in the second case, it is the design of a product. These are called ‘product
adaptability’ and ‘design adaptability’ respectively, as explained in Chapter 3. This
section discusses the quantification of adaptability for products only; and the
discussions are not repeated for design adaptability. Unless the distinction between the
design and product adaptabilities is explicitly mentioned, the statements can be
generalized to include both.
An ‘adaptation task’ is the actual process of modifying a product from its current state
to a new state that enables the product to function in the new operational mode. A
quantitative measure of adaptability should be an indication of how easily an adaptation
task can be carried out. This depends on both the design of the product and the
adaptation task in hand. Therefore, in measuring the adaptability of a product, the
adaptation tasks that are the target of the measurement have to be specified. This
schema for measuring adaptability requires a criterion for quantifying the ease or
difficulty of performing an individual task. This section first develops this criterion
called the ‘information content’; then it presents the formula for measuring the
adaptability of a product for a target set of adaptation tasks. The implications of this
formula lead to the development of useful guidelines for AD, which are discussed at the
end of this section.
4.3.1. The Information Content
The fundamental element of design is the presence of goals. There are always limits on
the ways a goal can be achieved. Therefore, the attainment of goals requires a plausible
plan, which is called a design. A design is plausible if its execution delivers the goals.
The execution of a plan involves the consumption of time and natural/artificial
resources. The resources that are consumed in order to achieve goals may also be called
costs. In this section, the information content of a design will be defined as the total
costs of achieving goals by that design.
While natural sciences study the existing laws and phenomena of nature, design
sciences are related to human intent and the attainment of goals. The latter has been
called “the sciences of the artificial” by Herbert Simon [Simon 1969]. Simon states that
the term artificial refers to man-made artifacts, and that the distinguishing feature of
man-made artifacts is that they are created to serve human purpose.
Goals may be described by a desired state of things. That is, in an artificial activity the
goal is to change an existing or naturally-occurring state of the environment to a
preferred state. The alteration of the existing state requires an artifact, without which the
preferred state would not occur. The artifact requires a design. Therefore, the goal of a
design task is to create an artifact that achieves the preferred state. If the desired state
could be achieved without an artifact, there could be no talk of design. In engineering,
the design task is to generate the blueprint or instructions for the creation of an artifact
that performs a set of predetermined functions, thus achieving the desired state in the
working environment of the artifact.
Limits and Constraints
The working environment of an artifact includes both natural and artificial objects. The
behavior of the environment is bound at all times by laws and properties that govern its
dynamic states. For example: physical laws govern the trajectory of a projectile and the
flow of thermal energy between objects; government regulations may affect the price
and availability of certain materials at a given time; market trends may affect the service
life of a design; and limitations in human physical abilities may constrain the operation
of a device.
Further, the limitedness of resources of all kinds may impose various constraints on a
plausible solution. For instance, the limits on the steady-state pollution absorption
capacity of a river might constrain the acceptable designs of a sewage processing plant;
the limits on the amount of available solar energy might constrain the design of solar
vehicles; the limits on time might constrain a project’s plan; and so on.
As a general rule, these laws, properties, and limitations exist as facts and cannot be
controlled by the designer. In a design problem, these limitations are usually interpreted
as ‘constraints’. Some constraints are explicitly mentioned in a design task, and some
are considered as general knowledge. Due to these constraints, changing an existing
situation to the target situation requires work or effort. In engineering design, these
constraints are responsible for the costs and complexities of creating an artifact that
satisfies the FRs.
Task and Information Processing
In this section, a task refers to a process that begins with a set of goals and ends when
these goals are achieved. Setting the goals is not within the scope of a task itself. Goals
represent the desired state, which does not exist or occur automatically through the
existing trend of the environment. The goal of an ‘engineering task’ is to create an
artifact with the desired functions and specifications. Achieving this goal involves
design and production, and requires work. This thesis uses the term information
processing (IP) to refer to the work or effort spent to accomplish a task. This term
indicates that every task ultimately consists of making decisions and allocating limited
resources (Appendix 3).
There is no single common means to measure, compare and trade various types of IP
capacities. For instance, a CNC machine simplifies a task and reduces the amount of IP
required. However, this machine utilizes more complex technology, and requires a
skilled operator. These are scarce resources and require more IP to obtain than a simple
machine with a less skilled operator. Although it is difficult to effectively quantify IP,
monetary value often proves to be an efficient method to measure IP and trade it
between different service providers.
The Information Content of a Design
A task is always accomplished according to a plan or a course of action, which can be
called its ‘design’. Simon stated: “Everyone designs who devises courses of action
aimed at changing existing situations into preferred ones” [Simon 1969]. A design
determines the path from the statement of goals to the fulfillment of those goals. The
amount of work needed to finish a task depends on this path. The information content
(IC) of a design is the total IP needed to accomplish the task (attain goals) by following
that design.
In engineering, a task is the creation of an artifact that is capable of delivering the
required function. A design is finished when the blueprint for the creation of the artifact
is specified to a level that can be communicated and contracted to available service
providers, with reasonable certainty that they will successfully achieve their task. This
is the end node in the hierarchical decomposition process. The IC of a finished design is
measured by the total costs of realizing it. The total costs of a design represent the
amount of IP required to attain the goal according to that design, which is the total cost
of manifesting the design in the form of an artifact which successfully delivers the
required functions.
The IC of an incomplete design is more difficult to determine. Incomplete designs exist
at every middle node in the design hierarchy. These designs are the different
alternatives that the designer has to choose from during the design process. It is difficult
to determine if a particular choice will result in a successful design at the end of the
design process. If a design proves to be unsuccessful, it should be regarded as iteration
in the design process. The IC of an unsuccessful design can be considered ‘infinity’
because such a design does not lead to the accomplishment of goals and one can view it
a requiring infinite amount of work to succeed. If an incomplete design yields a
successful design when it is finalized, its IC is the sum of the IP of finishing the design
task and the IC of the final design. In other words, the IC of an incomplete design is the
cost of finishing the design task plus the cost of materializing the finished design.
Summary and Conclusions
The Sciences of the Artificial are about achieving goals; a goal is a preferred
state; changing the existing state to a preferred state requires an artifact; and the
creation of an artifact requires a design.
Due to the presence of constraints, any alteration in the state of environment
requires work, which can be translated into information processing. The amount
of this IP depends on the design of the artifact which is intended to achieve the
desired state, and is called the information content of that design.
The IC of a design can never be zero. Zero IC means no change of states, hence
no design.
The IC of an ideal design approaches zero. This is consistent with Altshuller’s
law discussed in Section 2.1.2.
The best design is the closest one to the ideal design.
The rest of this section develops a formula for measuring the adaptability of products
based on the notion of IC developed above. The following symbols are used:
A single product
Set of relevant adaptation tasks for P
A single adaptation task for product P
Pr (Spi)
Probability or frequency of Spi
Inf (Spi)
The information content of Spi
A (P)
Adaptability of product P for the set Sp
Actual Final State
Ideal Final State
Ideal Minimum Information Content
Normalized Adaptability Factor
4.3.2. General Measure of Adaptability of a Product
Assume a product P that has to be adapted to fulfill some new requirements that are not
delivered by its current operation. This task is an “adaptation task” for P. The set of
“relevant” or “target” adaptation tasks for P is denoted by Sp. Sp is always a sub-set of
an indefinite set, which cannot be exhaustively known, of all plausible or potential
adaptations for P in its lifetime. Sp is usually chosen by higher management decisions
based on the circumstances that determine which adaptations are relevant to assessing
the adaptability of P, especially in comparison with other products.
A single adaptation task for P is a member of Sp and is denoted by Spi. Accomplishing
Spi requires a certain amount of work, which represents the level of complexity of
achieving the task amidst constraints and with limited resources. As discussed earlier,
this is called the information content of Spi, denoted by Inf (Spi), and represents the
total costs associated with performing Spi including time, resources, financial costs, etc.
The members of Sp may not be of equal importance; some adaptation tasks are more
likely to occur or are more frequently needed than others. The probability of the
occurrence of a single adaptation Spi is denoted by Pr (Spi). The higher the Pr (Spi), the
more important it is to make the product adaptable for Spi. For example, the
replacement of drill bits is an adaptability that is certainly needed in a drilling machine;
therefore all drilling machines have a mechanism to facilitate this process.
In this thesis, Adaptability is a measure of the suitability a product for adaptation to
varying service requirements. Adaptability for product P is denoted by A (P) and is
calculated over Sp, which is the set of relevant adaptations. Adaptability is adversely
proportional to the total cost of performing adaptation tasks within Sp. For a given P
and Sp, A(P) and can be written as:
A( P) ∝
∑ Pr( Spi ) Inf (Spi )
4.3.3. Physical States and IC of Adaptation
Inf (Spi) depends on how much the design of product P facilitates the task Spi. For
example assume that P is an electrical hand drill to be adapted for a grinding task. The
IC of this task is very small because the machine is already designed to perform this
adaptation by simply replacing the drill bit with a grinding wheel through an easily
detachable interface such as a chuck. As another example, assume that P is a car, to be
adapted to carry large loads. This adaptation requires major modification of the car to
transform its passenger cabin into cargo space. The IC to carry out this adaptation is
considerable unless the car is designed for this adaptation. For example, the car may be
designed so that the passenger cabin is mounted on the platform of a cargo space and
can be easily removed. It can be seen that the IC of an adaptation task depends on how
much the product has to be modified in order to deliver the new requirements.
Assume that the current physical state of product is S1. If at this current state the
product can fulfill the new requirements demanded by Spi, the information content for
the adaptation task is zero because no modification is required. This situation typically
happens only if the product is designed for versatility and can perform several functions,
often requiring little or zero work to switch from one function to another. Generally
speaking, however, the fulfillment of the new requirements demands a change in the
state of the product to a new state, or S2. Various changes in the physical states of P can
be chosen to make the product capable of delivering the new requirements. S2
represents the option which is chosen, assumed to be a logical choice that involves the
minimum information processing required with the starting point being S1. It is
assumed that once the product is put in the final state S2, it can successfully deliver the
new requirements of the adaptation task Spi with no further information processing.
Thus, the information content of Spi depends on the work needed to change the current
state of the product (S1) to the required state (S2):
Inf ( Sp i ) = F ( S1, S 2)
Actual and Ideal States
Spi demands the product fulfill a set of requirements different from the requirements
that it is currently fulfilling. If a new machine was designed, the state of this machine
would involve only the necessary features and functions. We call this the “Ideal State”,
in the sense that it is the physical state which is the “minimal” embodiment needed to
satisfy the new requirements. The ideal state is similar to the TRIZ concept of “SField” discussed in Chapter 2. During an adaptation task, the creation of the ideal state
may interfere with the existing state of the product and require extra work. The final
state of the product, after all the necessary changes have been made and the product is
capable of delivering the new requirements, is called the “Actual State”. This state has
to provide for both the ideal state (minimal required embodiment) and all the necessary
changes in the state that are required to assure the machine operates properly (Figure
4.5). The actual final state after the adaptation process is denoted by AS2. Thus
equation (2) is modified as follows:
Inf ( Sp i ) = F ( S1, AS 2)
Figure 4. 5: The Ideal and Actual States.
The “Ideal Minimum” Information Content of Adaptation
If we denote the information processing required to change the product from one state
to another by
Inf ( Si → Sj )
we can write:
Inf ( S 1→ AS 2 ) ≥ Inf ( S 1→ IS 2)
The right side of the equation is the “Ideal Minimum Information Content” (IMIC) for
the adaptation of P to the new requirements imposed by Spi. IMIC is the cost of
creating the minimal physical state that P needs in order to deliver the new requirements.
This involves only adding the missing elements and components to the current state of P.
The actual IC of adaptation is higher, however, due to propagation of modifications
beyond the minimal required physical state. The work required to accommodate the
propagated changes is the difference between the ideal and actual costs of adaptation.
The sum of such extra costs and IMIC is the actual IC of adaptation.
The distinction between IMIC and the extra work is important because the former is
related to what is available in P, while the latter depends on how P is designed and
constructed. The IMIC can be lower if the product, before adaptation, possesses some of
the required features of the final state IS2. This can be provided for in the initial
creation of P if Spi is foreseen; this is design for versatility discussed in Chapter 3. The
extra work included in the actual IC of Spi can be reduced if the product’s design does
not propagate changes, for example if it has a modular architecture that confines
modifications within one segment.
As an example consider a car which is being adapted to perform the function of a pickup truck. The IS2 in this case needs the missing cargo box, which consists of a platform,
enclosure, and an access gate (Figure 4.6). This structure is not available in the product,
the car in this case, and needs to be added. Other functions of the vehicle such as
mobility and control are already available.
Figure 4. 6: IS2 for the truck example.
Extra work is needed to implement the manifestation of IS2, such as removal of seats,
reduction of cabin space, and replacement of struts (Figure 4.7). Such changes are
necessary because of the way the car is designed; they do not directly help with the
achievement of the function, which is achieved by the cargo box.
Figure 4. 7: AS2 for the truck example.
As another example, assume a design requirement that is to transfer water from a river
to a reservoir. The chosen solution is a pump, and a car is to be adapted for this task
because it possesses the mechanical energy needed to drive the pump (Figure 4.8).
P: A car
Spi: Transfer water
IS2: A rotary pump
Figure 4. 8: IMIC for the pump example.
By looking at the functional structure of a car, it is easily seen that the rotary
mechanical energy is available. Therefore all the prerequisites for it (such as engine)
already exist in the physical state of a car. Therefore, the IMIC only requires the
inclusion of the pump. Any further work required to do this is the difference between
Inf (S1, AS2) and Inf (S1, IS2). This difference indicates how well the original design
of the car suits this adaptation task. For example if the vehicle is equipped with a utility
output shaft, a pump can be easily attached to it.
4.3.4. Calculation of Adaptability
Equation (1) indicated that general adaptability of P is inversely proportional to the IC
of the members of the set Sp. However, the information content of adaptation cannot
directly appear in the equation in a linear proportion. That is, if the IC of a task doubles,
it does not mean that adaptability is half. In fact, the element that is directly
proportional to the general adaptability of a product is the amount of ‘saving’ that is
achieved by adapting the product as opposed to manufacturing a new product. This
measure is also more intuitive because if this saving is zero, adaptability is zero and if
this saving doubles, we can say that the product is doubly adaptable.
The amount of savings achieved in an adaptation task can be represented by
Inf ( ZERO→ IS 2 ) − Inf ( S 1→ IS 2)
). This is basically the total cost, or IC, of making a new
machine minus the cost of adaptation. If this value is negative or zero, then there will be
no saving by adaptation and the procurement of a new machine may have to be
considered. This saving can be divided by the cost of building the new machine in order
to “normalize” this parameter. Thus, for a given adaptation task Spi we define a
parameter called the “Adaptability Factor” (AF). The adaptability factor for a single
adaptation task Spi is an indication of normalized saving achieved by Spi.
AF ( Spi ) =
Inf ( ZERO → IS 2 ) − Inf ( S 1→ AS 2)
Inf ( ZERO → IS 2)
= 1−
Inf ( S 1→ AS 2 )
Inf ( ZERO → IS 2)
We can readily see that:
0 ≤ AF ( Spi) ≤ 1
Equation 6 indicates the boundary values, 0 and 1, for the adaptability factor. If AF for
a task is zero or negative, adaptation does not apply. Also, A.F. cannot be more than
one because AF=1 means that the adaptation task has zero information content. The AF
of one occurs in versatile machines which, from the beginning, are designed to be
capable of delivering the extra service.
The goal of AD is to design-in adaptability for a product during the initial conceptual
design of that product. In design for ‘specific adaptability’, target adaptation tasks are
known in advance. Therefore, an adaptable design should provide solutions for some
additional FRs, which are needed for future adaptation tasks, in addition to solutions for
the initial FRs of the conventional design. Therefore, the total amount of savings is the
difference between the IC of multiple products and the IC of a single adaptable product.
Therefore, a general measure of the adaptability of a product can be obtained through
the combination of Equations (1) and (6):
A( P) = ∑ Pr( Spi ) AF ( Spi )
i =1
And by substituting the adaptability factor AF(Spi) from Equation 5 we have:
Inf ( S 1→ AS 2)
i =1
Inf ( ZERO → IS 2 )
A( P ) = ∑ Pr( Spi )(1 −
Because every machine performs at least the initial set of requirements for which it was
designed, no adaptation for that task is required and both probability and AF are one.
Therefore, in case a machine does not adapt to any function other than its intended
function, its adaptability is one. Thus we can write:
A( P) ≥ 1
Depending on the value of n in Equation 8, the adaptability of products can range from
one to five or ten or more. The adaptability of a machine, measured by the proposed
scheme, gives an estimate of the number of machines that can be replaced by a single
machine. For example if a machine performs three separate functions, its adaptability
would be three.
4.3.5. Implications of the Adaptability Equation
Inf ( ZERO → S 1) ≥ ∑ Inf ( ZERO→ IS 2i )
For a given product P and given adaptation task Spi, the amount of IMIC is
i =1
Inf ( S 1→ AS 2 ) ≥ Inf ( Zero→ IS 2 )
then this adaptable design is not justified.
then the adaptation task Spi is not justified.
fixed (with the legitimate assumption that IS2 is decided). IMIC can only be
reduced at the initial design and construction stages of P. An initial design that
has features of IS2 or includes add-on modules, however, adds to the initial
costs and can be justified only if Pr(Spi) is high.
The IC of extra work, caused by the implementation of AS2, can be reduced if
modifications are not propagated throughout the product. A modular architecture
with flexible and standard interfaces can confine modifications.
Access to both utilized and potential functions of P must be increased through
the subordination of physical assemblies to meaningful and recurring functions
and the design of assemblies that have accessible and generic input and output
connections for energy, material, and signal. The development of assemblies as
functional modules is likely to increase the number of useful adaptations, which
is ‘n’ in the equation.
The cost of adaptability is the sum of the IC of developing an adaptable design
and the IC of the adaptation task. This cost should be less than the IC of creating
two separate products in order to justify AD. Adaptability thus is more
applicable when resources are scarce and a reduction in their consumption
justifies the higher costs of AD.
4.4. Methods and Guidelines
As discussed in Section 4.1, the design hierarchy is developed by a recurring pattern of
nested divisions. The careful planning of the hierarchy, which is the key to achieving
adaptability, can be for the most part guided by the forecast information about possible
or expected adaptations if such information is available at the time of design. In the
absence of such information the general adaptability of a design to unforeseen
circumstances can be improved if certain etiquettes are observed.
This section first describes the criteria for the development of functional structures for
specific AD. Then it presents guidelines for the development of generally adaptable
mechanical products through the subordination of a design to a rational functional
structure as discussed in Section 4.1. The division of guidelines for specific and general
adaptabilities is not critical as all these guidelines should be considered in the overall
methodology of design for adaptability, which will be discussed in Section 4.5.
4.4.1. Specific AD
The development of the functional structure begins with determining the initial FRs.
This process is not unique; the designer may establish the FRs in various ways. Also,
the decomposition of FRs is not unique and depends on the designer’s choices of
solutions for every FR, and the way he/she decomposes every FR according to its
adopted solution. Therefore, especially early in the design process, there is some degree
of freedom in determining the functional structure, which subsequently influences the
corresponding physical structure. Careful development of the functional structure
results in the development of physical assemblies that perform meaningful or recurring
functions. Physical assemblies developed in this way can be used in different products
or configurations, hence facilitating adaptations. In specific AD the available
information on future adaptations can be utilized to develop a functional structure which
is suitable for foreseen adaptations. Specific adaptations in Chapter 3 were divided into
overlapping categories: versatility, variety, upgrading, and customization.
Product versatility, primarily relevant to the user, was defined as the ability of a single
product to be reconfigured for various functions. A versatile product may replace
multiple products, and its creation is justified when all these multiple functions are
needed in the normal service of the product. The additional functional requirements
(AFR) can be incorporated into the initial FRs. In this case the functional structure for
various FRs is developed in such a way as to take maximum advantage of the existing
functions and features of the product.
Product variety, primarily relevant to the producer, was defined as using the same basic
design and production infrastructure for the development of various models. In this case
the functional structure for every model consists of a base platform and model-specific
differentiating modules. The methods of detecting commonalities and optimizing the
compromise between differentiation and commonalities were discussed in Chapter 2.
Upgradeability is similar to variety with the added factor of time. Upgrading means the
utilization of new technologies or capabilities to enhance existing products for the user,
or to enhance existing designs for the new production by the producer. The main
technique, similar to that of variety, is modularization where frequently-upgraded parts
are designed as easily replaceable modules. Customizability is a broader category. It
applies to both product (user) and design (producer) adaptability. It can also be
unrelated to time similar to variety, or sequential in time similar to upgrading. The basic
technique in this category is also modularization.
From this discussion it is clear that depending on the objectives of specific AD,
different methods are used for the development of the functional structure of a product:
versatility involves multiple functions, variety involves component sharing, and
upgrading involves component replacement. These methods have been discussed in the
literature and were reviewed in Chapter 2. The following describes some guidelines
related to specific AD.
Define the primary FRs and the additional FRs (AFRs) for versatile product
design. Find low-cost solutions for achieving AFRs in the original design.
Provide extra features and functionalities in a design for possible future needs.
For example, an output utility shaft can be (and usually is) provided in the
design of farm tractors; thus making the tractor adaptable to various functions.
Utilize the existing features and components to achieve extra functionalities.
This guideline is against the rules of general AD which encourage the
fulfillment of functions via separate subsystems. For example, an axe handle can
be designed to have mounting contours on ‘both’ ends, thus enabling the user to
extend the usage when one end breaks.
Identify a group of products that can be developed from a shared adaptable
design. Identify common or recurring elements, either functional or structural,
among products within the portfolio. Design these common elements as a shared
Identify the differentiating features among products within a portfolio and
design them as add-on modules. Develop a parametric design for these modules
so that they can be custom-made for various specifications.
Design the interface between platforms and modules for easy attachments and
detachments, such as self-aligning and lock-and-release mechanisms.
Facilitate the replacement of components which are likely to require upgrading,
such as the components that undergo rapid technological obsolescence.
Identify customizable features, often found at the output ports where the
functions of a product are delivered to the outer environment. Then design a
product for the easy alteration, replacement, or addition of these features.
4.4.2. General AD
Chapter 1 and Section 4.l of this chapter explained the approach of this thesis for
achieving general adaptability in mechanical systems. This approach is based on the
principle of segmentation discussed in Chapter 1, and it was suggested that in the
absence of forecast information functions can be used as the segmentation criterion,
hence calling the approach function-based segmentation. Section 4.1 discussed the
adaptability properties of a rational functional structure and discussed the mechanisms
by which this structure can be created. Therefore it was suggested that the development
of the actual architecture of the product should emulate the same adaptability properties,
hence calling the approach the subordination of the physical structure to a rational
functional structure. Since a functional structure is in fact the nested segmentation of
functions, the two ways of calling the proposed approach are conceptually identical.
The design etiquette which is encouraged in this approach has the following three
principles. These characteristics are sought at the first few levels of decomposition of
functions as discussed in Section 4.1.
First, an adaptable product has a modular architecture, so that the required
modifications of an adaptation task do not propagate throughout the product.
The nesting of modularization generates a hierarchy of subsystems.
Second, subsystems are functional modules which are designed to perform
unambiguous and useful functions. The operation of a functional module is
relatively independent from the product it serves, and its functional interaction
with other assemblies in any product is the exchange of inputs and outputs.
Third, subsystems are autonomous and self-contained so that they can perform
their function independently from their working environment.
Section 4.2 discussed the challenges of applying this approach to the design of
mechanical systems. This section describes the methods and guidelines which help in
overcoming these difficulties and thus achieving a generally adaptable architecture in
the design of mechanical systems.
Utilizing alternative technologies (software instead of hardware)
The function of many mechanical systems can be described by the output they generate
from a given input [Pahl 1988]. The relation between the input and the output can be
very complicated. For example, turning the steering wheel of a car causes the wheels to
turn, at different angles for inner and outer wheels, and at different sensitivity ratios for
different vehicle speeds. These functions can be achieved by mechanical means: a rack
and pinion system turns the wheels, the geometry of the steering knuckle achieves the
difference in angles of the two front wheels, and a governor adjusts sensitivity
according to vehicle speed. These functions, which are the relation between inputs and
outputs, could be performed in software if inputs were turned into computer data, and
outputs were generated by electric motors (actuators) at computer command.
A technological enabler for the development of generally adaptable products is the
digital and software control systems, which are becoming increasingly affordable. The
phrase by-wire technology refers to various systems that have replaced conventional
mechanical, hydraulic, pneumatic, and electro-mechanical systems with digital control
systems. This technology creates a soft link between inputs and outputs of a system,
where functions are performed via software and not by means of mechanical
components. Inputs are processed by a computer that generates the appropriate signals
to activate the electric-powered actuators which generate the appropriate output. The
following figure shows the schematic diagram of the general design of such systems.
End Effector
(output function)
Transducers and sensory
devices (if closed-loop)
Communication line
Device ports
ports (input)
Overall program
Device programs
Input programs
Operation command
(user interface)
Command data
Figure 4. 9: The replacement of mechanical systems by "soft" electro-mechanical systems.
Autonomy of modules
Develop subsystems as autonomous modules. These modules do not depend on a
particular configuration or on other components within a product for their proper
functioning. They should contain their required elements within them.
Mounting small subsystems
From the above rule it can be deduced that a subsystem should be mounted on the
system it belongs to “functionally”. For example a car’s radiator, which is mounted on
the car’s body due to spatial considerations, should be mounted on the engine so that the
engine becomes autonomous in its function.
Functions meaningful and recurring
There are several ways of decomposing a given FR as discussed in Section 4.1. Among
several scenarios, one should be chosen whose sub-FRs are more meaningful, widely
usable, and recurring. The “meaning” of a function will be discussed in Chapter 6.
First develop input/output functions, and then develop internal mechanisms
This guideline can be deduced from the previous guideline. Since the purpose of a
device is understood from its input/output functions, these functions are more likely to
be meaningful. For instance, in the design of a vehicle an important function is
accomplished by the ‘seat’, whose function is easily understood by the user. Internal
components such as transmission or rack-and-pinion, though complex in their designs,
are of no direct relevance to the user.
Smaller Sizes
Assemblies, while capable of delivering the required function, should be developed
with the smallest size that can be reasonably achieved. This may require replacing
mechanical systems with alternative technologies or reducing the size of mechanical
parts when possible. Utilization of smaller sizes reduces the active spatial constraints
and thus facilitates modifications and rearrangements during adaptation tasks.
Integral Flexibility
Use flexible elements whose flexibility is not related to modularity: hydraulic hoses,
manufacturing adjustments, flexible shafts, wires, universal joints, etc.
Increase the compatibility among subsystems through the standardization of systems
and their interfaces. For instance in a subsystem which provides mechanical power, a
shaft is a more standard and compatible form of output than a gear which requires its
exact mating and adjustments. The interfaces among interacting mechanical
components must be designed to provide for high levels of compatibility and flexibility.
The proper design for adaptability and exchangeability of mechanical interfaces is
discussed in the design of mechanical bus systems ([Gu 2002], [Gu 2003]).
Manufacturing Adjustments
Adaptability and flexibility can be increased by the use of manufacturing adjustments.
These are tuning design features that provide for high tolerance in dimensions or
positioning of parts, or they allow variations in other parameters such as voltage. This
tolerance increases adaptability of the product, both in its design and its construction, to
environmental variations (i.e. noise) and to varying service requirements. ([Otto 1994],
[Lee 2003]).
Regular and Generic Forms
Use regular surfaces and generic forms which facilitate future alterations and
amendments. Flat or cylindrical surfaces are preferred over sculpture surfaces;
rectangular boxes are preferred over complex-shaped objects; and so on.
Physical Independence
Functional modules, which are functionally independent by definition, should also be
made physically independent as much as possible. This guideline emphasizes the
difference between general AD and the Independence Axiom proposed in [Suh 1990].
An Example: Adaptable Design of a Lock
A lock consists of two input/output functions. One is the intake of the user’s signal
indicating which state of the device is desired: locked or unlocked. The second is the
actual locking action, which depends on the application of the lock. For example, in a
lock that is used on a hinged door of a building, the locking action might be to drive a
deadbolt between the door and its frame; and in a lock used in a revolving door, the
locking action might be to prevent the main shaft from spinning.
If the lock is designed by the rules of general AD, its two functions should be
performed by two independent and autonomous modules, and the connection between
these modules must be made as flexible as possible. Figure 4.10 illustrates such a design,
which utilizes the general electro mechanical system discussed earlier in this section.
The user module receives the open/close command, which could be in the form of
turning a key, swiping a card, punching a code, using remote control, etc., and
transforms it into an electric signal through the appropriate transducer. This signal is
sent to the action module, where through a relay it activates an electric motor for the
locking action.
Input Signal
Electric Pulse
Figure 4. 10: The adaptable design of a lock.
Since the two functional modules in this design are connected by electric wires or
wireless transmission, the lock system can be easily adapted to various configurations
and installations. Therefore this design provides ‘product adaptability’. It also provides
‘design adaptability’ (variety) because modules are autonomous in their functions, thus
various lock systems can be developed through the morphological combination of
alternative designs for the two modules (Chapter 2). This is shown in Figure 4.11.
User Signal
Turning a key
Swiping a card
Entering password
Proximity ID
Remote control
Locking Action
Compatible for
morphological combinations
Driving deadbolt
Locking a hinge
Powered entrances
Handicapped access
Figure 4. 11: Variety through the possibility of morphological combination.
The Frame and Function Architecture
In a service-based, as opposed to object-based economy, a product is a temporary
assembly of functional elements in a configuration that is capable of delivering a
temporary service [Ayres 1998]. There is no need for the product to exist as an object
when its service is no longer required. Thus selling a product is in fact selling use
([Seliger 1997], [Lindahl 2001], [Seliger 2000]). This concept can be best observed in
the new corporate model known as virtual enterprise. It is a company that is comprised
of various functionally independent units that can provide their services regardless of
the company they serve ([Goldman 1995], [Tolle 2003], [Gou 2003], [Bechler 1997]).
A virtual enterprise is a temporary assembly of such autonomous units, often globally
distributed and connected via communication networks, formed to exploit fast-changing
opportunities such as making one particular type of product or delivering one particular
type of service, then dissolving after the project is finished to allow the partners to find
new partners.
Function-based segmentation of physical assemblies is the key method for the
development of products that emulate the flexibility and adaptability of a virtual
enterprise. A subsystem within a product is a relatively self-sufficient and autonomous
module that corresponds to an obvious function in the functional hierarchy. These
functions are generally understandable and describable, and more importantly, they
recur in different designs. Such assemblies can be used in many overall products with
different configurations and functions, as long as their interfaces have the appropriate
inputs and outputs.
Therefore, in a service-based view towards engineering, a product is designed for a
temporary use. It consists of a frame, as well as various functional assemblies. For any
required overall function or service, a configuration for the product is designed, in
which the frame determines the overall layout or embodiment of the product. Therefore
a frame is designed to suit the embodiment requirements of the intended usage. A frame
may be an abstract spatial order among physical assemblies, an actual physical structure
such as a chassis or truss, or a combination of both. Then, various functional assemblies
are added to the frame in order to achieve the desired functions. These modules might
be mounted on a physical frame or on each other in the appropriate spatial order.
Special components and product-specific assemblies, if required, can also be installed to
make the product functional.
With this architecture, for any new application or operational mode the product can be
adapted by the appropriate configuration of the frame and functions. An adaptation task
involves the reconfiguration, replacement, and redesign of functional modules and the
appropriate modification of the frame and special parts. The frame-and-function
architecture possesses the properties of general adaptability discussed in this chapter.
4.5. Adaptable Design Methodology
If forecast information on future adaptations is available, it should be utilized. Thus in
the overall AD methodology ‘design for specific adaptabilities’ has a higher priority
than ‘design for general adaptability’. This section describes the proposed methodology
in this sequence.
Phase 1. Specific AD
Define the original design problem (FRs).
Identify the set of target adaptation tasks (Sp). This process utilizes forecast
information on versatility, upgrading, customization, and variety.
Develop a functional structure that includes both original FRs and the
requirements of future adaptations (AFRs).
Design the physical structure of the product according to the applicable methods
and guidelines of specific AD. Depending on the methods used, subsystems
might be developed as shared platforms, replaceable or interchangeable modules,
optional add-on features, etc.
Phase 2. General AD
In the absence of sufficient forecast information, assemblies should be designed
as functional modules, beginning with modules that interact with the
environment (input/output functions), then developing the internal mechanisms.
The suggestions of specific AD overrule the suggestions of general AD.
Functional modules should be developed so that they are as autonomous and
self-sufficient as possible.
Phase 3. Development
Maintain the functional and physical independence between subsystems,
especially between functional modules developed by the guidelines of general
AD. Connect subsystems by means of soft interfaces such as wires and
manufacturing adjustments.
Whenever possible achieve functions by software, use alternative designs and
technologies if necessary.
Reduce sizes and utilize generic forms and regular surfaces in order to decrease
spatial constraints. Also, utilize standard interfaces in order to increase
Develop a spatial frame for the overall embodiment. Add the required
subsystems to this frame.
Phase 4. Hierarchy
Apply the above methodology to the development of individual assemblies.
Depending on the complexity of a design, the segmentation process can be
nested for a few levels, dividing the overall design into assemblies and dividing
the larger assemblies into smaller assemblies. This results in a hierarchical
structure of physical parts that conforms to the hierarchy of the rational
functional structure.
4.6. AD and Other Life-Cycle Design Goals
The environmental benefit of AD stems from reducing production volume by having
fewer products for the same amount of service. This can be translated to using fewer
natural resources in preparing finished goods. In addition to adaptability, other
environmental benefits might be sought when designing an adaptable product. These
include: the use of non-toxic and environmentally friendly materials; design for
recycling and reuse when applicable; design for manufacturing processes with low
environmental impact on nonrenewable resources; and designing products that consume
less energy and causes less pollution in their operation. Most of these environmental
characteristics of a product are relatively independent from its overall architecture,
which is determined by AD. They are determined by other design decisions such as the
choices of materials and manufacturing processes, or the choice of solution principles
for the operation of the artifact.
In addition to the reduction of environmental impacts, several other life cycle objectives
may be sought in the design process. These include: quality issues such as performance,
reliability, durability, and safety; design for manual or automated assembly; design for
repair and maintenance; design for low cost manufacturing; and design for rapid
product development. Most of these characteristics are also to a large extent determined
by design specifications other than the product’s overall architecture.
AD is applicable from the early stage of the design process until it determines the
overall architecture and embodiment of the product at the conceptual level. As the
design proceeds to more detailed stages, there is less freedom to make function-related
changes in design and AD becomes inapplicable. On the other hand, other life cycle
design issues mentioned above are less relevant at the beginning of the design process
because little is known about parts, material types, repair frequency, etc. As design
proceeds to further levels of detail, more detailed information is available and these life
cycle issues become more relevant.
Therefore, adaptable design ought to be performed at earlier stages of design than those
life cycle design issues which depend on detailed design specifications. This statement
suggests a sequence in procedures and eliminates the conflict, which is fortunate. That
is, adaptable design must be performed first, and then when further detailed information
is available, other life cycle aspects can be considered. Iterations, as an inherent part of
any design process, might be inevitable.
Figure 4.12 illustrates this sequence. The dotted curves indicate that the driving
objective of modularization at the beginning of the design process is adaptability, which
is based on the segmentation of functions. Towards the end of the process other life
cycle goals related to detail specifications become more applicable.
Applicability of methods
Driving Objectives of
Life cycle objectives
related to detail
design specifications
Adaptable Design
Design Process
Design Problem
Final Design
Figure 4. 12: The applicability of adaptable design and other life cycle design in the design process.
4.7. Summary
This chapter described the design hierarchy and the decomposition mechanism by
which this hierarchy is generated. In an ideal scenario, this process results in a rational
functional structure with a causal relation among its elements. This relation makes such
a structure suitable for adaptability. Section 4.1 concluded that the imitation of this
structure in the actual architecture of a product is a logical approach towards achieving
general adaptability in designs. Section 4.2 discussed the inherent properties of
mechanical design that hinder this imitation. These properties stem from the structural
connectivity among solid parts, and more importantly, from the functional ambiguity of
subsystems which are created through the conventional mechanical design process.
Section 4.3 introduced a measure for adaptability based on the amount of ‘saving’
which is achieved through adaptation. These savings were measured by the information
content, which is asserted to represent ‘total costs’ more accurately than monetary value
does. Section 4.4 discussed the methods and guidelines for both specific AD and
general AD. Section 4.5 presented the overall methodology of design for adaptability in
which specific AD was given a higher priority than general AD. Section 4.6 showed
how this methodology can be extended to accommodate life cycle design.
The next chapter will present the conceptual design of a few mechanical systems which
are designed according to the overall methodology and the guidelines of this chapter.
Chapter 5: Examples
This chapter discusses the adaptable design of a few mechanical systems. These
examples are conceptual designs; and for simplicity they are not designed for life cycle
objectives. Section 5.1 discusses an example of specific AD in which a product is
designed for predetermined adaptations. Section 5.2 presents two mechanical systems
which are designed for general adaptability without targeting particular adaptations.
Section 5.3 first describes an example of design for specific adaptability from the design
literature, and then designs the same system for general adaptability in order to
demonstrate the differences between specific AD and general AD.
5.1. Specific AD: Versatile Bicycle Accessories
Among the four categories of specific AD discussed in Chapter 3, design for versatility
is the most specific category. Design for versatility has the most forecast information,
and specifically designs a product for a predetermined set of adaptations (Chapter 3,
Figure 3.8). A versatile design has been chosen as the example of specific AD in this
section. This example is the versatile design of bicycle accessories similar to the Master
Lock design shown in Figure 3.10. This is a case of product (user) adaptability; and
adaptations from one function to another occur frequently and reversibly. Therefore the
final design should allow these adaptations to be performed easily by the user.
5.1.1. The Design Process
The design process follows the AD methodology discussed in Chapter 4.
Step 1: Original FRs and Additional FRs
Various accessories are available for bicycles: locks, carrier racks, fenders, bottle
holders, tools, storage compartments, etc. These accessories can be attached to a bicycle
in several ways. There is a level of redundancy if these accessories are individually
installed, hence suggesting the possibility of a versatile design which can perform the
function of multiple accessories.
After comparing various functions of bike accessories, the carrier rack and the
splashguard have been chosen for adaptable design. The reason for this choice is the
similarity between the functional structures of these two devices. Their initial functional
requirements are:
Carrier rack:
FR1: Attach firmly to the bicycle.
FR2: Hold a load of up to 10kg.
FR3: Do not interfere with the normal operation of the bike.
FR1: Attach to the bike.
FR2: Protect the rider from water thrown by the rear wheel.
FR3: Do not interfere with the normal operation of the bike.
Figure 5.1 shows the functional structures as well as solutions for both the carrier rack
and the splashguard. Physical functions, which require physical components as their
solutions, are denoted by F; and non-physical functions are denoted by NF. The
comparison of these structures reveals a common physical function, which is the
attachment of an accessory to the same location on the bicycle. Therefore, one module
could perform this common function in both applications.
Carrier Rack
Functional structure
of the Carrier Rack
F: attach to bike
Solutions for
physical functions
F: hold load
Bracket under seat
Strong Metal Frame
NF: do not interfere
Position above
the rear wheel
Conceptual design
Functional structure
of the Splashguard
Solutions for physical
F: attach to bike
F: block splash
NF: do not interfere
Bracket under seat
Cover top of wheel
Position above
the rear wheel
Conceptual design
Figure 5. 1: The functional structures of a carrier rack and a splashguard.
Also, the comparison of the physical embodiments of bicycle accessories reveals that
there is a similarity between the metal frame of the carrier rack and a U-lock. This
suggests the possibility of including the U-lock in this versatile design. Therefore, its
FRs must be added to the FRs of the previous two devices in the list of AFRs
(Additional FRs) for adaptable design. The FRs of the U-lock are as follows:
FR1: Attach to the bicycle when not in use.
FR2: Lock the bicycle.
FR3: Provide security against theft.
FR4: Do not interfere with the normal operation of the bike.
The schematic diagram of the functional structure and a conceptual solution for this
product are shown in Figure 5.2.
F: attach to bike
F: Lock/Unlock
Bracket on frame
U-frame locked on
a cross bar
F: Security
Strong Metal Frame
Figure 5.2: The functional structure of a U-Lock.
NF: do not interfere
Position inside
the frame
Step 2: Three Individual Designs
This step develops individual solutions or conceptual designs for individual applications.
In this example these are known from existing designs. The conceptual designs for the
splashguard, the carrier rack, and the u-lock are shown in Figures 5.1 and 5.2. These
designs represent the Ideal Initial States for individual designs, as explained in Section 2
of Chapter 4. The ideal initial states are needed in order to assess the adaptability of
final design using Equation (8) of Chapter 4.
Step 3: The Conceptual Design of the Adaptable Product
Figure 5.3 shows an adaptable design for this example. The assembly of the overall
product is shown on top, and the solid models for major components are shown on the
bottom. The U-shaped frame is designed for quick attachment and is made strong to
provide both lock security and rack strength. The deadbolt is designed asymmetrically,
so that it slides into the bracket only when its orientation for the insertion of the Uframe into holes is correct. The deadbolt also includes a square hole for the attachment
of wrench bits, and a patterned hole at the end for the attachment of screwdriver bits.
These extra features can be created at negligible cost. The bracket is designed
symmetrically so that the deadbolt can be inserted from either side. Its design includes
tapering ribs that transfer the load of the rack to the seat bar, thus achieving strength
without adding to weight. The top of the bracket is designed as a simple flat surface in
accordance with the AD guidelines. This flat surface is likely to facilitate the addition of
extra features to the design in the future. By attaching a fender to the same bracket, this
design can perform all three sets of FRs.
Figure 5.3: The design of an adaptable bicycle rack.
Step 4: Adaptability
In the measurement of adaptability for this example three simplifying assumptions are
made. First, it is assumed that the probability of occurrence for all adaptation tasks is 1,
because the product will be used as a rack, a lock, and a splashguard during any given
service period. Second, it is assumed that adapting the configuration of the product from
one usage to another, for example from lock to rack, does not require noticeable effort.
Therefore the information content of performing any adaptation task is zero. The
information contents for various tasks involved in Equation (8) are representative of
their complexities and costs. The above adaptable design is more complex and costly to
materialize than any of the three individual products. However, the third assumption is
that these higher initial costs are negligible. The effects of such higher costs will be
discussed in another example in this chapter (Section 5.3.1). With these three
assumptions, Equation (8) of Chapter 4 can be used to measure the adaptability of this
product as follows:
Adaptability= (1-0) + (1-0) + (1-0) = 3
This means that the adaptable product designed above will replace three products which
would be needed for the same service.
5.1.2. Discussion
The final design in this example is similar to an existing product. This is not unexpected
as specific AD represents the existing design methods. This example demonstrates the
fundamental element of specific AD which is to build predetermined adaptabilities into
the product during the design process. It can be seen that performing predetermined
adaptations in the final design is very easy. This design, however, is difficult to adapt to
unforeseen changes. For instance, this design cannot be adapted to a new type of lock
such as a chain lock.
5.1.3. Other Examples of Specific AD
Several other examples of specific adaptability were presented in Chapter 3. The Sony
video cameras and Ford automobiles are examples of product variety (Figures 3.4 and
3.8). Both of these designs, unlike the bicycle accessory designed in this section, are
instances of design (producer) adaptability, in which the adaptation of the same basic
design for the production of different models is performed by the producer.
An example of specific AD for upgrading is the design of personal computer systems.
Various parts such as the memories and processors undergo rapid technological
obsolescence. These components are designed as easily-detachable modules. Therefore,
the product can be upgraded by the user who can have these parts replaced easily, hence
extending the product’s life. Upgradeable computers may also be considered an instance
of specific adaptability for the producers, who can efficiently upgrade their products
and constantly utilize the state-of-the-art technology in their models. Some other
instances of specific AD such as the design of modular robots were discussed in
Chapter 2.
5.2. Examples of Design for General Adaptability
General AD aims to subordinate the physical structure of a product to its rational
functional structure. The subsystems of such a product are autonomous functional
modules; that is, each module independently performs a meaningful function
corresponding to the functional structure. Since such ideal architecture may not be
practically feasible, the guidelines of AD help the designer develop products which are
closer to this architecture. These guidelines suggest that the development of modules
begin from the input/output functions, which determine the purpose of a device, and
that other functions be achieved by software instead of mechanical components where
possible. The guidelines also encourage the use of flexible connections, standard
interfaces, and generic forms as discussed in Chapter 4.
Thus, the main process of design for general adaptability is to develop a rational
functional structure and then develop independent modules for each FR in this structure
with the help of the guidelines. There are many ways to develop a functional structure
for a design problem, and the choice reflects the perception on the designer and
determines the final outcome. It can be seen that this process does not depend on
forecast information regarding future adaptations. Therefore, unlike the previous section,
the examples in this section do not target predetermined adaptabilities.
This section presents two examples of general AD. They are conceptual designs for the
purpose of demonstrating the basic procedure of design for general adaptability. The
details of these systems are not discussed and they are not designed for financial or
aesthetics criteria. When applicable, the technological feasibility of a design concept is
shown using similar industrial applications.
5.2.1. The Adaptable Design of a Hydraulic Jack
In solution-neutral terms, this example is the design of a mechanical device that
amplifies the user’s muscular force (input) to the required level (output). In general AD,
these input and output functions should be designed as separate modules, and the
connections between them must be made as flexible as possible.
Figure 5.4 shows a few conceptual designs for manual force amplification. The first
four designs utilize solid mechanical connections such as gears or chain-and-sprocket.
These designs impose fixed spatial relations between the input and the output. In the
hydraulic systems shown on the bottom, however, the input and output functions are
performed by independent modules (pump and jack). The connection between these two
modules is a hydraulic hose, which is more flexible than solid parts. Thus, these
hydraulic systems are more adaptable than the other designs in the figure.
Handwheel, worm-G, chain, rack-pinion
Handwheel, spur-G, worm-G, crank-slider
Handwheel, spur-G, chain, power-screw
lever, spur-G, worm-G, power-screw
Handwheel, crank-slider, pump, jack
lever, crank-slider, pump, jack
Figure 5.4: Conceptual designs for a manual force amplifying device.
Figure 5.5 illustrates a versatile design for a manual hydraulic pump. In order to be able
to utilize this pump in various applications, it is designed to operate at two different
speeds for different loads. The top of the figure shows the main pump body with its
hydraulic conduits, and the double-diameter plunger assembly. The bottom of the figure
shows the assembly of the double speed manual pump and its operation mechanism.
When the ‘selector lever’ shuts the bypass, the pump works at high speed and low
pressure. The lifting of the plunger sucks the fluid into both cylinders, and pressing it
down delivers the liquid from both cylinders through the output hose. When the
‘selector lever’ opens the bypass, the bigger cylinder becomes idle because it is
connected to the fluid tank during both upward and downward strokes. In this case, only
the small plunger is active, resulting in higher force but lower speed.
Pump body
Plunger assembly
Plunger Assembly
Pump Body
(from tank)
High Volume,
Low pressure
Low Volume,
High pressure
(to jack)
Figure 5.5: The design of a double-speed manual hydraulic pump.
Figure 5.6 illustrates the adaptable design of the hydraulic jack assembly. The pump is
mounted inside a fluid tank. This tank is a rectangular box and supports the pump
handle on top. A hose connects the pump and tank assembly to a hydraulic jack, which
is mounted on a small trolley so that it can be moved to the desired location under a
vehicle. The bottom of this figure shows two conventional hydraulic jacks.
Adaptable Hydraulic Jack
Hydraulic Bottle Jack
Hydraulic Garage Jack
Figure 5. 6: An adaptable design and two conventional designs for hydraulic jacks.
In this example the general adaptability of a system is increased through the
development of independent functional modules, and without specifically preparing the
product for predetermined adaptations. In the proposed design, the hydraulic pump is an
independent module which can perform its function (delivery of hydraulic energy) in
many circumstances. For example, it can be used to operate an automobile jack, a press,
or a metal cutting tool such as the Jaws of Life used in emergency rescue operations.
Similarly, the jack is designed as an independent module that can take its hydraulic
input from the above pump or other sources of hydraulic pressure such as an electric
pump. Therefore, this system and its modules can be adapted to various circumstances.
In contrast, the conventional designs shown in the Figure 5.6 are usable only in their
normal operational mode.
5.2.2. The Adaptable Design of a Vehicle
The function of a modern vehicle goes well beyond the simple provision of mobility. In
the design of vehicles many complex performance requirements are considered such as
ergonomics, steering and road handling, suspension and dynamic stability, passenger
comfort, collision safety, fuel efficiency, pollution control, aesthetics, and so on. The
example in this section does not consider these requirements, and only designs a vehicle
for its primary function of transportation. The conceptual designs discussed in this
section are developed by the guidelines of design for general adaptability, and are
intended to demonstrate the application of the frame-and-function architecture.
In this section, first the functional structure of a vehicle is developed; then for each
function in this structure an independent physical module is designed. For an
operational mode a vehicle is assembled from these modules in a configuration (frame)
that suits a temporary service. In this section several vehicles for different operational
modes are conceptually designed through the assembly of functional modules.
The Functional Structure
As suggested in the Phase 2 of the methodology in Chapter 4, the development of the
functional structure begins by identifying the intended input/output functions. These
functions represent the purpose of a vehicle and they are thus the main design objectives.
In this example the following functions are considered for a vehicle:
Mobility on land (Traction)
Positioning passengers
Driving and control
Source of power
There are many other input/output functions that are not considered such as protecting
occupants from the outside environment, carrying loads, attaching optional devices, and
so on. A vehicle’s design should also satisfy many requirements which are not physical
functions, such as safety, cost, aesthetics, and simplicity. These requirements are not
discussed in this example. Figure 5.7 shows the simplified functional structure of a
vehicle based on the above four functions.
Non-physical FRs
- Safety
- Cost
- Aesthetics
- Simplicity
- ….
Auxiliary Functions
Wheel Speed
Energy Source
Steering system
- Cabin
- Heat / AC
- Cargo
- Lights
- ….
Figure 5. 7: A functional structure for a vehicle.
Functional Modules
The next step is the development of independent modules for performing each of the
four functions in the functional structure.
Function One, Traction; Solution, Motorized Wheels
For the function of land traction the chosen solution in this example is a motorized
wheel. This module has two sub-functions, the first is to provide rotational mechanical
power to the wheel and the second is to provide traction on terrain. Thus it consists of
two subsystems: an electric motor and a rim/tire assembly.
The electric motor is designed to have its stator and electronic control systems mounted
on a stationary inner shaft, while its outside housing is the rotor. The advantages of
having the rotor on the outside are: the possibility of mounting the motor inside the
wheel, access to motor from both sides, and using the body of the motor as the hub of
the wheel thus saving in material. Figure 5.8 shows the operation of the motor within
the vehicle’s overall system. The figure also shows the system boundary for the motor
module. As the figure shows, the motor technology with dynamic reconfiguration for
performance optimization is commercially available. The function of this module is to
receive electric power and operation command signals, and generate mechanical power
with characteristics that are demanded by the command signal. Therefore the motor has
two input ports, which can be designed as sockets, and one output, which is the rotary
outside hub.
Driver’s Operation
Digital Signals
(Computer Input Data)
Performance Signals (DSP Input):
- Speed & Direction
- Torque $ Braking
Digital Signal
Processor (DSP)
Input ports of the
motor (Sockets)
Switch Signals
Sensor Data
(Feedback Loop)
Permanent Magnet Motor with a control
system that dynamically reconfigures the
motor for peak efficiency (WaveCrest Co.).
Motor’s System Boundary
Mechanical Power (Torque) with
desired performance characteristics
Figure 5. 8: The operation of a wheel motor (picture courtesy of WaveCrest Co.).
Figure 5.9 shows an embodiment design for this motor. This design has a splined
outside hub for mounting different rims and tires. It also has mechanical, electrical, and
electronic connection ports for the coupling of two or more motors. The figure also
shows three types of rim/tire assemblies. Any of these can be mounted on the motor for
different applications as shown in the bottom of the figure. The wheel on the right is
wider and includes two electric motors in its hub. An assembled wheel is an
independent unit for the traction function, and it is connected to the rest of vehicle
through two sockets, one for operation signals and the other for power.
Electric Motor
Connection sockets
and coupling ports
Double Motor
Figure 5. 9: Electric wheels designed as independent functional modules.
Function Two, Positioning Passengers; Solution, Seats
The seat should be design as an independent module that can be used for its function in
any configuration. There are many such designs available, such as the two seats shown
in Figure 5.10.
Figure 5. 10: Seats for the function of 'positioning passengers'.
Function Three, Control; Solution, Electronic Driver Interface
The three functions of controlling a vehicle are: increasing speed, reducing speed
(braking), and steering. The acceleration function in a motorized wheel is readily
available because the torque and speed of the motor can be directly controlled by the
driver. The braking function, including regenerative braking, can be implemented
through the “by-wire” braking technology. In automotive industry the term “by-wire”
symbolizes the tendency towards utilizing electro-mechanically driven devices instead
of mechanical, hydraulic or pneumatic components [Doriben 2003]. The brake-by-wire
design is available in the literature and is not discussed here. Therefore, this section
only discusses the steering function and the design of an innovative steering system.
The functional structure of a steering mechanism consists of two input/output functions:
the driver’s steering interface (input), and the actual turning of the wheels (output). This
section first briefly describes the conventional design of steering systems as well a more
adaptable design recently introduced in the industry. Then it discusses the conceptual
design of a steering system which is designed for general adaptability.
The Conventional Steering System
In a conventional steering system the driver’s interface is a steering wheel, which is
connected to a rack-and pinion via a shaft (Figure 5.11). The linear motion of the rack is
transformed into the pivotal motion of wheels by the steering knuckles. Thus in this
system the input and output functions are connected via mechanical links, which impose
various spatial and structural constraints in the design of a vehicle. For instance the
location of the steering wheel is coupled with interior design, and the designs of the
rack-and-pinion assembly and steering knuckles are linked with the design of chassis.
Therefore, modifying and adapting this system is difficult. A conventional steering
system is typically used in its nominal operation mode only.
Figure 5. 11: The conventional steering system.
By-Wire Steering
The separation of the driver input function from the rest of the steering system can be
seen in by-wire steering (Figure 5.12). In this design the input function is accomplished
by an independent module which is connected to the rest of the system by wires. This
module consists of a driver interface (wheel, handgrips, etc.), position sensors, and the
electronic/software systems that generate the appropriate signals for the pinion’s servo
motor. The rest of the steering system includes various feedback sensors for closed-loop
control; a servo motor for driving the pinion, and mechanical components such as the
rack and steering knuckles similar to the mechanisms of a conventional design.
Figure 5. 12: By-wire steering. (Picture courtesy of SKF)
Figure 5.13 shows the driver interface module designed by GM for the by-wire steering
of their 2002 concept car called AUTONOMY. It can be seen that the steering handle
can be mounted in different locations. Therefore, the position of the driver is not
constrained, resulting in flexibility and adaptability in the design of an automobile.
Figure 5. 13: The driver steering module in AUTONOMY (Courtesy of GM).
A Generally-Adaptable Design: Servo Steering
The by-wire steering system is more adaptable than the conventional design because its
driver interface is designed as an independent module. The rest of the system, however,
is as non adaptable as the conventional design. Figure 5.14 shows how the adaptability
of the design can be further increased. Diagram (a) represents the conventional design.
Diagram (b) shows the by-wire steering system, in which the steering shaft and the
booster are replaced by a servo motor that turns the pinion, while the rack-and pinion
and steering knuckles remain intact. Diagram (c) shows our proposed design in which
the wheels are independently turned by two separate servo motors. This design, called
servo steering in this thesis, eliminates the need for the rack-and-pinion, hence
removing the structural dependency between the right and left wheels.
Steering Wheel
(a) Conventional Steering
Rack and Pinion
Driver Interface
(b) By-wire Steering
Servo Motor
(b) Servo Steering
Servo Motors
Figure 5. 14: Increasing the general adaptability of steering systems.
The mechanical separation of two steered (front) wheels requires the synchronization of
the turning angles for the two wheels, which can be easily achieved by software. This is
illustrated in Figure 5.15. When the driver steers, the right and left front wheels turn at
different angles so that their axes intersect on a point located on the rear axis. This point
is the “instantaneous center of rotation” for the vehicle. The real-time calculation and
implementation of these angles is an easy task. Such technology has been utilized in
CNC machines for decades. The servo steering system can also be applied to any
number of wheels, which creates new possibilities for steering a vehicle (Appendix 2).
Front steering at a small angle
Front steering at a large angle
Figure 5. 15: Calculating rotation angles for the right and left wheels.
Therefore, in the servo steering design both the driver’s interface (input) and the turning
mechanism for each individual wheel (output) are developed as independent functional
modules. These modules perform a meaningful function according to the functional
structure of a vehicle, and the connections among them are flexible wires. Further, all
mechanical links between the input and output functions are replaced by software. It can
be easily seen that this design imposes few constraints on the vehicle and thus facilitates
modifications and adaptations.
Function Four, Energy; Solution, Batteries
There are several solutions for an onboard source of power: internal combustion engine,
gas turbine, fuel cell, rechargeable batteries, solar energy, and so on. All these solutions
are valid and currently used in various vehicles. For example diesel generators or gas
turbine generators are used in train locomotives, solar panels are used in recreational
light vehicles, and batteries or fuel cells are used in electric cars. In this section the
battery is chosen as the solution. Figure 5.16 shows the concept of a modular battery for
this example. The batteries have regular shape and flat surfaces and can be stacked up in
various geometries to obtain the required voltage or power.
Figure 5. 16: Modular battery cells.
Frame (Chassis)
The chassis in this example is a spatial frame for the assembly of modules in an
appropriate configuration. This section discusses the adaptable design of a chassis for a
slow-moving vehicle such as a lunar vehicle or an AGV (Automated Guided Vehicle).
In this design, adaptability is achieved through the structural modularization of the
chassis, which can be designed as a space frame. A space frame is a three dimensional
truss consisting of links and joints. Figure 5.17 shows the space frame elements
designed for the chassis. The links are metal tubes of several standard lengths with a
bolt assembly on each end. The tube lengths are designed to allow the construction of
various geometries through primitive triangular patterns. The tubes in this design also
have holes for passing wires and installing sockets. The bolt assemblies are designed to
allow a bolt to turn without having to turn the tubes. The joints are forged meal balls
that have several threaded holes at frequently-needed angles for the attachment of bolts.
A simple frame in the bottom of the figure illustrates how the links are connected
through joints to obtain triangular patterns.
Bolt Assembly
Link Assembly
Figure 5. 17: The space frame elements designed for this example.
In this design, links and joints are independent functional modules that perform their
functions regardless of the configuration they belong to. With an inventory of a few
types of joints and a few sizes of links, a large number of frames with different sizes
and configurations can be developed. A space frame structure is thus an adaptable
design; it can be easily dismantled and reconstructed for new requirements. A space
frame chassis can be easily modified in width, length, height, location of wheels, and
the load-bearing configuration. Figure 5.18 shows a sample chassis designed by space
frame elements. This design is a two-dimensional frame for better illustration; an actual
chassis is a three dimensional frame so that it can carry vertical loads. The corner
brackets are for the installation of wheels. They are designed as moment-bearing solid
pieces so that they can convert the twists and forces of wheels into linear forces at the
brackets’ contact joints with the space frame.
Figure 5. 18: A space frame chassis.
With the assembly of functional modules on various frames, various vehicles for
various applications can be created. Figure 5.19 shows two sample configurations for
cars. The batteries, seat, driver control and other modules can be mounted anywhere on
these chasses and connected to the rest of the system by wires. The tubes in the space
frame are equipped with holes for passing wires and sockets for plugging various
control ports. The servo motors for the servo steering system can be mounted on any
bracket; thus any wheel can be steered if required.
Servomotor mounts
Independent motorized wheels
Space frame truss elements
Double-motor wheel
Single-motor wheel
Figure 5. 19: Adaptable car configurations.
Figure 5.20 shows a few other examples of vehicles which can be constructed using the
functional modules discussed above. The vehicles in this figure have applicationspecific frames which are not constructed from the space frame components. Many
other vehicles with different configurations are also possible, such as golf carts, tandem
bicycles, trucks, etc.
Seat Mount
Figure 5. 20: Other types of vehicles that utilize the functional modules.
Although the designs of this section are simple and only consider the basic functions of
a vehicle, they demonstrate the ideas and strategies of design for adaptability. Figure
5.21 shows the rational functional structure of a vehicle in which functions are
represented by their chosen solutions not in solution-neutral terms. The figure also
shows a schematic diagram of the vehicle design. This figure illustrates the one-to-one
correspondence between the functional and physical structures in this design.
By-Wire braking
Servo Motor
Servo Steering
Chassis (frame) and
connection ports
Figure 5. 21: One-to-one correspondence between the functional and physical structures of the
proposed design.
This vehicle, unlike the conventional designs, consists of modules whose functions are
directly relevant to driving, and whose purposes are obvious and understandable by the
average user. These modules are: wheels for traction, seats for the occupants, control
interfaces for driving, batteries for power, servo motors for steering, and brakes. Any of
these independent modules can be utilized in other designs and configurations. For
instance the same seat and batteries can be utilized in a car or in a wheelchair. Further,
each module can be replaced by a different design for the same function, and without
affecting the rest of the vehicle. For instance batteries can be replaced by fuel cells or
by an engine/generator assembly; or the driving interface can be replaced by remote
control. Various vehicles can thus be generated through the morphological combination
of different solutions for modules as discussed in Chapter 4.
5.3. A Comparative Example
This section discusses two different designs for electric home and garden tools. The first
design, which is chosen from the literature, demonstrates the process of designing a
product for predetermined adaptations. The second design is developed for general
adaptability without aiming at predetermined adaptations. The comparison of these two
designs demonstrates the differences between specific AD and general AD.
5.3.1. A Versatile Home and Garden Tool
This example is chosen from [Gu 2003]. It is the design of a versatile home and garden
tool that replaces a chainsaw, a hedge trimmer, and a grass edger. Figure 5.22 illustrates
some commercially available models of these three products. This figure also highlights
the functional similarities between them. These similarities indicate redundancy in
having three individual products, hence suggesting the possibility of replacing these
products with a single versatile design. The process of designing a versatile product for
these functions is similar to the procedure described in Example 5.1. However, in this
example the versatility of the product is achieved through a shared platform and several
add-on modules.
Figure 5.22: Common functions among chainsaws, trimmers, and edgers.
Step 1: Original FRs and Additional FRs
The functional requirements (FRs) for the individual products are:
FR1: be held and controlled by hand.
FR2: use electric motor to provide mechanical power.
FR3: provide the rolling motion of the chainsaw.
FR4: provide cutting action (the chainsaw).
Hedge Trimmer:
FR1: be held and controlled by hand.
FR2: use electric motor to provide mechanical power.
FR3: provide the reciprocal motion of the trimmer.
FR4: provide trimming action (the clippers).
FR1: be held and controlled by hand.
FR2: use electric motor to provide mechanical power.
FR3: provide the rotary motion of the edging blades (or strings).
FR4: provide grass-cutting action (spinning blades or strings).
Figure 5.23 shows the functional structures of these three products and the conceptual
solutions for these functions. The common elements among these designs are a handle
and a switch for holding and controlling the device, and an electric motor for
mechanical power. These common elements, shown on the left side of the figure, can be
developed as a shared platform. The differentiating features shown on the right can be
developed as add-on modules for individual applications.
Rolling Motion
Hand grip/Control
Mechanical Power/ Speed
Chain and Sprocket
Cutting Action
Electric Motor/ Gearbox
Specific Module (Chainsaw Attachment)
Common Features (Platform)
Hedge Trimmer
Reciprocal Motion
Hand grip/Control
Mechanical Power/ Speed
Crank and Rocker
Cutting Action
Electric Motor/ Gearbox
Common Features (Platform)
Specific Module (Hedge Trimmer Attachment)
Rotary Motion
Hand grip/Control
Mechanical Power/ Speed
Long Shaft
Cutting Action
Blades or Wires
Electric Motor/ Gearbox
Specific Module (Edger Attachment)
Common Features (Platform)
Figure 5.23: The functional structures of the chainsaw, the hedge trimmer, and the edger.
Step 2: Individual Designs
The conceptual designs for individual products are shown in Figure 5.23.
Step 3: The Conceptual Design of the Adaptable Product
Figure 5.24 illustrates the solid model of the overall adaptable design, developed by Gu
and Slevinsky [Gu 2003]. The shared platform is shown on the left. It includes a motor,
two handles, and a switch. This figure also displays the interface between the platform
and modules. This interface is designed based on the guidelines of mechanical bus
systems that incorporate lock-and-release and self-adjustment features [Gu 2003]. Three
add-on modules for the three functions are shown in center of the figure, and the
assembled chainsaw is shown on right.
Figure 5.24: An adaptable design consisting of a platform, three modules, and proper interfaces.
Step 4: Adaptability
Since all three functions of this product are commonly used during its service life, the
probability of every adaptation task is 1. In order to calculate the remaining parameters
in the Equation (8) of Chapter 4, the information content of every task involved,
including design, manufacturing, and adaptation, is measured by monetary value
(Chapter 4, Section 4.1). The cost of every task is thus assumed to represent its ‘total
costs’ in terms of the consumption of natural and artificial resources as discussed in
Chapter 4. For instance it is assumed that the consumption of nature’s absorption
capacity is included in the price of artefacts. With these assumptions, the adaptability of
this design can be calculated based on the costs of components and activities. The
following costs are assumed here:
The cost of an individual chainsaw: $80
The cost of an individual hedge trimmer: $100
The cost of an individual edger: $60
The cost of the shared platform: $50
The cost of the add-on module for the chainsaw: $50
The cost of the add-on module for the hedge trimmer: $70
The cost of the add-on module for the edger: $30
The cost of swapping available modules: $0
The calculation of adaptability in Equation (8) is based on the saving which is achieved
by adapting a product instead of replacing it. Therefore this calculation assumes a
nominal operational mode (initial state S1) from which the product is adapted to
another mode (actual final state AS2). For this example, the normal operational mode
of this design is assumed to be the chainsaw. Also, this versatile product has been
designed for effortless adaptation from one application to another. Therefore the cost of
adaptation is in fact the initial extra investment for obtaining the add-on modules. The
‘Adaptability Factor’ (AF) for the three applications of this design can now be
calculated using Equation (5) of Chapter 4:
AF ( Spi ) =
Inf ( ZERO → IS 2 ) − Inf ( S 1→ AS 2)
Inf ( ZERO → IS 2)
= 1−
Inf ( S 1→ AS 2 )
Inf ( ZERO → IS 2)
Chainsaw: The adaptable chainsaw consists of a platform and a chainsaw module.
Therefore its cost is $50+$50 = $100. It is $20 more than the cost of an individual
chainsaw ($80). This is the cost of adaptability for the nominal operational mode.
Therefore the adaptability factor for the chainsaw is:
AF = 1 – (20/80) = 0.75
Hedge Trimmer: The cost of changing S1 to S2 is $70, which is the price of the add-on
module for the hedge trimmer. The Inf
(ZERO to IS2)
is $100, which is the price of an
individual hedge trimmer. Therefore:
AF = 1 - (70/100) = 0.3
Edger: The cost of changing S1 to AS2 is $30, which is the cost of the add-on module.
The Inf (ZERO to IS2) is $60, which is the cost of an individual edger. Therefore:
AF = 1- (30/60) = 0.5
The total adaptability can be calculated as the sum of adaptability factors:
A( P) = ∑ Pr( Spi ) AF ( Spi )
i =1
Adaptability= 0.75 + 0.3 + 0.5 = 1.55
Since this number is more than 1, this adaptable design is justified.
5.3.2. The General AD of Home and Garden Tools
This section discusses the design of an electric chainsaw for general adaptability. In the
previous example three applications for the versatile product were targeted during the
design process; here the chainsaw is developed as an assembly of functional modules
without targeting specific adaptations. The functional structure of a chainsaw (Figure
5.23) consists of an electric motor, handles and switches, and the chainsaw itself. For
general adaptability, these functions should be accomplished by independent modules.
Figure 5.25 shows the design of the motor module. There are certain features in this
design that make it more adaptable. First, the housing has regular shapes and surfaces,
and also includes several mounting holes for possible future applications. Second, the
two inputs of the motor, which are electric power and operation signals, are not
designed as integral electric cords and integral control switches. Instead, they are
designed as separate components because their requirements vary from one application
to another. Therefore the housing has sockets so that the electric power and control
switches can be connected in any configuration that suits a particular application. Third,
the output is a rotating shaft which is more universally usable than other types of
mechanical output. In particular, this output is more compatible with unforeseen
applications than the gear output in the previous example. Fourth, the motor is a
versatile type which can deliver different performance profiles for different applications.
The electronic control systems and relay switches for this purpose are located inside the
housing, and receive their operational signals through the electronic port (switch socket)
on the motor.
Inside: rotor/stator;
electronic system for the
control and optimization
of motor performance;
relay switches.
Output shaft
Switch socket
Power socket
Figure 5. 25: An adaptable electric motor designed for the chainsaw.
Similar to the motor, the chainsaw should also be designed as an independent module.
The input of this module is rotary mechanical power, and its output is the chainsaw
action. Figure 5.26 illustrates the conceptual design of the chainsaw module. The input
connection is made via the insertion of a shaft into the hub of a sprocket inside, which
creates the rolling motion of the chain as the output. This module is also equipped with
a handle, a protective shield, and switches for ON/OFF and speed control. These
switches are connected to the motor through the connection socket, thus allowing this
module to utilize different motors. It can be seen that this module does not depend on a
specific electric motor for its operation, and can take its input from any source of
mechanical power even from a gas engine. The flat surface on the side of the sprocket
housing facilitates the installation of various sources of mechanical power.
Handle and switches
Sprocket inside
Shaft input
Motor connection socket
Figure 5. 26: The chainsaw module.
Figure 5.27 shows the chainsaw assembly which consists of the motor module, the
chainsaw module, and a handle. This figure also shows several other power tools which
utilize some of these functional modules. These applications were not known during the
design of the chainsaw. Thus this example demonstrates the general adaptability of the
proposed design to unforeseen adaptations.
Electric Cord
Socket connection wire
(a) Chainsaw
(b) Grinder
(c) Hedger
(d) Sander
Figure 5. 27: Adaptable designs for power tools.
This section discussed two different designs for power tools. The first design was
prepared for predetermined adaptations, while the second design was developed for
general adaptability. The first design performs well for its intended adaptations, but its
platform and other modules are difficult to use in other applications. For instance the
output of the platform is a gear, which requires its exact mating interface. The second
design is not optimized for any particular adaptations, thus any adaptation task requires
some effort. However, it can be adapted to unforeseen conditions with reasonable effort.
5.4. Calculation of Adaptability in General AD
In this chapter adaptability was only calculated for the examples of specific adaptability
and not for the examples of design for general adaptability. Chapter 4 discussed that Sp,
the set of target adaptation tasks considered in calculating the adaptability of product P,
is a sub-set of the large set of all potential adaptations for P in its lifetime. If Sp is
known at the time of design, the methodology of specific AD should be utilized and the
adaptability can be calculated as the sum of adaptability factors. General AD, however,
aims at developing a certain architecture for products that makes them more adaptable
without targeting a specific set of adaptations. In this case it is not possible to evaluate
the results according to the adaptability formula at the time of design. Instead, the
adaptability of the final product for various adaptation tasks can be evaluated after
adaptation tasks are decided. Once Sp is decided, the calculation of adaptability is
similar to the procedure discussed in the examples of specific AD.
Chapter 6: Summary and Discussions
6.1. Thesis Summary
The goal of this thesis was to contribute to the development of a new design paradigm
called design for adaptability, in which the adaptability of a product to varying service
conditions is perceived as a design characteristic.
The extension of utility was identified as the underlying reason for adapting an artifact.
This view revealed a logical relationship between the scarcity of resources and the need
for AD. It also helped with the categorization of adaptation scenarios through
considering different modes for extending utility. The study of literature within these
categories revealed that current methods assume a predetermined set of adaptations, and
then design a product for specific adaptabilities. This research thus focused on
developing a design method for increasing the general adaptability of mechanical
designs without targeting particular adaptations.
Modularization was recognized as a principal method for achieving adaptability. While
predetermined adaptations related to versatility, variety, upgrading, and customization
make up the segmentation criteria for the existing modular design methods,
modularization for general adaptability required a different criterion. The study of the
mechanical design process resulted in choosing functions as the segmentation criterion.
This choice stemmed from the adaptability properties of a rational functional structure,
and the proposition was that the subordination of the actual architecture of a product to
a rational functional structure would yield similar adaptability in the design.
The architecture of such a product would be a hierarchy of autonomous functional
modules, along with spatial frames and product-specific components when required.
This architecture was called frame-and-function; the contrasts between this architecture
and the popular platform architecture highlight the differences between specific AD and
general AD. Three design etiquettes were proposed for achieving this architecture in the
design of mechanical systems. First, designs should be modularized and relations
between modules should be made as flexible as possible. Second, modules should
perform meaningful functions which occur frequently in mechanical designs. This can
be achieved through the proper choice of sub-FRs during the decomposition of
problems. Third, modules should be developed as self-sufficient and independent
systems. These autonomous modules can then be used in different configurations and in
different systems without depending on a system’s layout and other specifications.
The function-based division of subsystems is practiced in the design of virtual
enterprises, software, electronic systems, and other areas of design in which systemlevel design and component-level design can be carried out independently. The inherent
properties of mechanical design, however, hinder the application of this method in the
design of mechanical systems. Two such properties were discussed in this thesis: the
structural connectivity of mechanical components caused by spatial constraints; and the
functional ambiguity of mechanical subsystems, which are typically designed for a
specific function within their parent system and are not usable elsewhere. Although
such inherent characteristics of mechanical design are somewhat inevitable, the thesis
proposed several mitigating guidelines which help with making the design of
mechanical systems closer to the ideal frame-and-function architecture.
The thesis also proposed a formula for assessing the adaptability of a design based on
the amount of ‘savings’ which are achieved via adaptation. This assessment does not
help with measuring the adaptability of a design based on its architecture. It is an afterthe-fact method and requires information on costs and difficulties of performing
adaptation tasks and procuring new products. These costs and difficulties are measured
by information content in the proposed formula.
Then the thesis proposed a methodology for AD in which specific AD is performed first
in order to take advantage of forecast information, and then general AD is performed in
order to increase the adaptability of the design to unforeseen changes. The application
of this methodology was demonstrated through several conceptual design examples.
The highlights of the thesis are summarized in Table 6.1.
Justification of
AD research
Identify the
purpose of
Extension of
Scarcity Æ service-based market Æ
adaptation instead of new production
Types of
Modes of
time, subject, doer
(parallel/sequential; product/design;
user/producer) Æ (upgrading, variety,
versatility, customization)
Scope of AD
Set limits
3 limits
Avoiding terminology inconsistencies
Identify original
Study current
Specific AD,
General AD
General AD not established in
mechanical engineering design.
General AD
Seek an ideal
Imitate the ideal
rational functional
structure in the
physical structure
Function-based segmentation;
autonomous functional modules
assembled on spatial frames.
(frame and function architecture)
Study inherent
Find ways to
avoid constraints
Specific AD prior
to general AD
The overall AD methodology
Criterion is saving
Trade-off formula applicable to
specific AD only.
Table 6. 1: Highlights and organization of the thesis.
6.2. Discussions
This section discusses three issues related to this research. First, the concept of
“function” in function-based modularization for general adaptability will be discussed
to show that the distinction between specific AD and general AD is related to the
“scope of applicability” of mechanical subsystems and their functions. Second, the
measurement and minimization of “information content” and the inclusion of “quality”
in the assessment of a design will be discussed. Third, the circumstances that justify the
higher costs of making products more adaptable are discussed using the proposed
formula for measuring adaptability. These discussions will be utilized to draw the
conclusions of this research in the next chapter.
6.2.1. Function-Based Modularization
Mechanical systems are naturally constructed from subsystems or modules. Methods of
modular design further increase the modularity of mechanical systems. This thesis
proposed function-based modularization as the primary method for increasing the
general adaptability of mechanical designs. The distinction between this method and the
conventional methods of developing modular products is the emphasis which is given to
functions: while the division of subsystems in conventional design is driven by
problem-specific criteria which might not be related to functions, in ‘general AD’ the
division of subsystems exclusively aims at developing modules which perform
meaningful functions. Two implications of this distinction are discussed in this section.
Function as Division Criterion
In ‘general AD’, FRs at every level of decomposition are chosen in such a way that
each physical FR represents a useful function. Then for each FR an autonomous module
is developed which is self-sufficient and is capable of delivering its function
independently from the larger context it is placed in. These two etiquettes, the
usefulness of functions and the autonomy of their corresponding physical modules,
determine the modularization scenario in the architecture of a product. Thus the division
of subsystems is based on the meaning, recurrence, or usefulness of their functions. This
approach avoids the four shortcomings of conventional clustering methods discussed in
Chapter 2: combinatorial complexity does not exist because only a few rational options
have to be considered; uncertain numerical data are not utilized in data-sensitive
algorithms; functional ambiguity of modules is reduced because modules are explicitly
designed for meaningful and recurring functions; and the designer’s freedom in
changing the design of a module can be maintained because modules are autonomous
and their design/modification can be carried out independently.
A Function’s Meaning
The above discussion assumed that some modules perform a more meaningful function
than others. The fact is that there is no general agreement on the meaning of functions
in mechanical engineering design. In Systematic Design the function of a component is
defined as its effects on materials, energies, and signals [Pahl 1988] 1. This narrative
description does not reflect the purpose of a component and the rationale behind its
existence. In Axiomatic Design, on the other hand, functions are defined as the
necessary and sufficient description of goals [Suh 1990]. This definition does not reflect
the actual physical effects thus both physical and non-physical requirements are called
FRs in Axiomatic Design.
The opinions on how to ‘represent’ functions are even more diverse [Hashemian 1997a]. Of relevance to this discussion is the requirement for the functions to be represented
in solution-neutral terms. Appendix 1 presents a function representation scheme in
which a function is described by its actions on physical entities, where actions and
entities are chosen from predetermined taxonomies. Despite their academic research
value, such solution-neutral representation methods have not proven useful to the design
In Systematic Design the function of a component is defined as what it actually does;
the functional structure of a product is obtained by the replacement of every component
with its function. By this popular definition of a functional structure, the “subordination
of physical structure to functional structure” would be a redundant statement. Therefore,
general AD emphasizes that the physical structure should be subordinated to a rational
functional structure defined in Chapter 4. This means that relations between subsystems
should be causal. Few mechanical systems follow this structure in their designs.
practice in industries. The reason is that solution-neutrality for the meaningful
representation of functions is a relative parameter. It depends on the level of design
hierarchy to which the task at hand belongs. People, industries, companies, and design
teams use different vocabularies for describing functions. For a designer who has been
assigned with the task of designing a rack-and-pinion, the name of the device might be
an acceptable description of FRs. In a larger context, for the average user of a vehicle
“steering” is a meaningful description of a function while “rack-and-pinion” refers to a
particular mechanism. Therefore, solution-neutrality, meaning, and usefulness of a
function are relative parameters which depend on the scope of a module’s application.
More “generality” in this context means: the design task is at a higher level in the
design hierarchy; the function is applicable to a broader range of applications; the
function is of interest to a larger audience; and so on.
The beginning of this section emphasized that the distinguishing feature of functionbased segmentation is the development of modules which perform “useful” functions.
The above discussion revealed that this usefulness is a relative parameter which reflects
the “generality” of a module’s function. Therefore, the main difference between a
functional module developed by general AD and a product-specific module developed
by specific AD is in fact in their scope of applicability. That is, general AD aims to
increase the scope of applicability of mechanical systems. A similar statement can be
made on the difference between platform architecture and frame-and-function
architecture: while a platform’s function is usable within a portfolio of products, the
‘functional modules’ of a frame-and-function architecture perform functions which are
usable in a broader spectrum of applications.
6.2.2. Information Content
Chapter 4 defined the information content (IC) of a design as the ‘total costs’ of
materializing that design. These costs reflect the consumption of resources of all kinds
including natural resources such as materials, energy, pollution absorption capacity,
land, water, and so on; and artificial resources such as industrial infrastructure,
subcontractors, labor and expertise, software and hardware, and so on. This
interpretation of information content is somewhat different from the definition of IC in
Axiomatic Design, which relates IC to the “probability of success” and measures it by
“bits” [Suh 1990]. Three distinguishing aspects of this interpretation are discussed in
this section: the use of monetary value for measuring IC, the encouragement of a
minimalist view leading to such principles as “self-help” in design, and the inclusion of
“quality” in evaluating a design.
Measuring IC
The consumption of resources is most commonly measured by monetary means. For
instance, the consumption of Nature’s absorption capacity can be measured by financial
criteria; this valuation of “pollution” might include penalties and fines, loss of
customers due to adverse publicity, class action suits, and other consequences of
causing environmental impacts [Tipnis 1998]. Monetary means are also used for the
valuation of rare natural resources. For instance, the value of 200 million tons of topsoil
blown off the U.S. Great Plains in one 1934 dust storm has been estimated at $9 trillion
[NGM 2001]. The measurement of resources by financial means, however, requires a
fair and intelligent valuation and pricing system.
Minimizing IC
The IC of a design depends on how much the existing states have to be altered in order
for that design to materialize successfully. Therefore, the minimization of information
content means that a good design should require minimal alteration of the existing states.
This philosophy leads to the development of design principles such as the self-help
principle discussed by Pahl and Beitz [Pahl 1988].
The success of a design can rarely be decided in a binary fashion: successful or
unsuccessful. Success can be achieved at various levels. The level of success represents
the ‘quality’ of a design. Quality is an indication of how well the initial objectives are
fulfilled. Therefore, an evaluation schema should include the parameter of quality in
addition to the main parameter of IC. This is shown in Figure 6.1. In this figure, the
goal of design is to set the value of a functional requirement (FR on the horizontal axis)
within an acceptable range. Although every design which delivers the FR within this
range is considered ‘acceptable’ or ‘successful’, different values within this range yield
different values for quality. In this figure quality as a function of FR is shown on the
horizontal plane (XY plane). In a typical design situation, the final value of FRs can be
determined only probabilistically because of manufacturing variations and other noise
factors [Suh 1990]. In this figure the probability density function of FR is shown on the
vertical plane (XZ plane). Therefore, the level of success, measured by quality, can be
calculated by a dual integral which corresponds to the enclosed volume in the figure.
Quality as a
function of the
value of FR
Probability Density
function of FR
Figure 6. 1: Both quality and probability should be used in the evaluation of a design.
6.2.3. Justification of Design for Adaptability
The formula for measuring adaptability resulted in several rules and guidelines in
Chapter 4. Particularly, it was stated that a negative value for the adaptability factor (AF)
means that the adaptation task is not justifiable. When the procurement of a new
product is difficult due to the scarcity of resources, AF values will change and can
justify adaptability. This can be discussed in the context of an example.
Assume a product costs $100; it can be adapted to perform the function of another
product which costs $80 and the cost of adaptation is $500; the cost of an adaptable
design of the same product is $200, and the cost of adapting this design is $0. For the
original product, the AF for the adaptation task is:
AF (original) = (80-500)/80 = -5.25
(not justified!)
For the adaptable design of the product, AF for the adaptation task is:
AF (adaptable) = [(80-0) - (200-100)] / 80 = -0.25
(not justified!)
Now assume these products have to be shipped to a far location for an expedition (e.g. a
space expedition). The transportation cost for each product is $400. This has to be
added to the cost of products, reminding that the parameter in the formula is IC which
represents total costs. Thus:
AF (original) = (480-500)/480 = -0.04
(not justified!)
AF (adaptable) = [(480-0) - (200-100)] / 480 = 0.8
This example shows that the scarcity of resources justifies the higher costs of adaptable
design and the costs and difficulties of performing adaptation tasks.
Chapter 7: Conclusion
This chapter lists the conclusions and contributions of this research and briefly
discusses future work related to the subject.
7.1. Conclusions of This Research
Importance of AD
The potential benefits of AD encompass the environment, the user, and the producer. Its
environmental benefits are achieved through creating new usage for idle or obsolete
products, thus postponing product retirement and reducing product replacement. Its
customer benefits stem from the customizability and adaptability of products to varying
needs. Its producer benefits result from the ability to reuse proven designs and
production processes instead of creating new ones. Increasing environmental concerns,
rising customer expectations, and increasing competition among producers motivate
research and development of AD as an emerging paradigm for mechanical design.
The AD Framework
This thesis established a framework for AD as a new design paradigm based on a broad
definition of adaptability, which includes user adaptations and producer adaptations,
design adaptations and product adaptations, and adaptations which occur over the
course of time and those which are unrelated to time. The setting up of this framework
involved original classifications and definitions, and the study of circumstances and
parameters which influence adaptability-related decisions during design. In the
proposed framework, existing design methods can be categorized as “targeting
predetermined adaptabilities” during the design process.
Design for General Adaptability:
Function was identified as a general modularization criterion which can be considered
in addition to, or in the absence of, predetermined adaptability targets.
The fundamental arguments for utilizing functions in designing for general adaptability
are supported by the importance of functions in a service-based market on one hand,
and by the adaptability properties of “rational functional structures” on the other hand.
For the actual implementation of this idea in mechanical design, this thesis proposed the
construction of systems from autonomous functional units in a “frame-and-function
architecture”. This requires the overcoming of the inherent difficulties of mechanical
design, which were identified as structural connectivity and functional ambiguity. For
this purpose, guidelines and methods were developed.
In conclusion, the proposition for the development of adaptable products is supported
by logical reasoning, while the proposed methods and guidelines need to be tested
through practical design problems for further development and improvement.
The Overall Methodology
The AD methodology combines specific AD and general AD in a sequential manner. As
discussed in Chapter 6, the distinction between these two, similar to the distinction
between the platform architecture and the frame-and-function architecture, leads to a
fundamental discussion on the “meaning” of a function. It can be concluded that the
premise of the overall methodology, including both specific AD and general AD, is to
generalize the applicability of designs and systems via making their functions less
context-dependent and more universally usable. This conclusion also reveals the
relation between a function’s meaning and its generality of usage. This finding is
helpful for the development of function representation schemes such as the one
discussed in Appendix 1.
Measuring Adaptability
The proposed formula for measuring adaptability is directly based on the rationale of
decision making and results in a tangible quantity for adaptability. The methods of
evaluating and quantifying the elements of this formula, however, need to be further
developed. In particular, the relation between information content and total costs
requires extensive research in order to develop practical quantification schemes.
7.2. Contributions
Treating adaptability as a design characteristic similar to manufacturability, thus
setting grounds for a new design paradigm.
Identifying the extension of utility as the purpose of adaptation, thus establishing
a relationship between the scarcity of resources and the need for AD.
Developing a categorization of adaptabilities in engineering design, which led to
the conclusion that current methods are concerned with design for specific
adaptabilities, while there is no formal method for the design of mechanical
systems for general adaptability.
Developing a proposition that the general adaptability of a product can be
increased through the subordination of its physical structure to an ideal rational
functional structure.
Proposing the frame-and-function architecture for the general adaptable design
of mechanical systems. A system with this architecture consists of several
autonomous functional modules which correspond to a rational functional
structure and a spatial frame for the assembly of these modules according to the
embodiment requirements of a design problem.
Proposing methodology and guidelines for design for adaptability.
Introducing a novel interpretation for the concept of information content.
Developing a measure for adaptability based on the amount of saving which is
achieved by adaptation.
7.3. Future Work
The calculation of adaptability in this thesis is based on comparing the information
content of an adaptation task with the information content of producing a new product.
Therefore this method is applicable only when the adaptation task is known and
Inf(S1ÆS2) can be measured. An alternative approach is to calculate the adaptability of a
product based on its design characteristics. The design characteristics that affect the
adaptability of a product include: the correspondence between the functional and the
physical structures, the ratio between the number of functional components and the
number of product-specific components, the difficulty (information content) of
assembly and disassembly tasks, the flexibility and compatibility of interfaces among
modules, and the utilization of smaller sizes and generic forms. Developing methods for
the quantification of these characteristics for the calculation of adaptability would be an
extension of the current research.
While Axiomatic Design defines IC by the ‘probability of success’, this thesis defines it
as the ‘cost of success’. The probability of success, however, might be an indication of
costs. That is, a design that has a higher probability of success usually costs less to
succeed than a design with a lower probability of success. Since probability of success
is relevant only when design is incomplete, it might be possible to use it as a heuristic
rule for assessing the ‘cost of success’ of solution alternatives generated for FRs during
the design process. The study of the relation between probability and cost might lead to
the validation of the above heuristic rule.
The methods proposed in this thesis should be tested through more concrete case studies,
especially with respect to cost and feasibility.
[Albano 1993] Albano, L. D., Suh, N. P., "Information axiom and its implications,
Proceedings of the 1993 ASME Winter Annual Meeting, American Society of
Mechanical Engineers, Design Engineering Division, V. 66, 1993, pp 1-11.
[Allen 1962] Allen M. S. “Morphological Creativity”, Prentice Hall, Englewood Cliffs,
NJ, 1962.
[Altshuller 1984] Altshuller, G., “Creativity as an Exact Science, The Theory of the
Solution of Inventive Problems (TSIP)”, translated from Russian, Gordon and Breach
Pub. 1984.
[Altshuller 1998] Altshuller, G., "40 Principles: TRIZ Keys to Technical Innovation",
1st edition, Technical Innovation Center Inc., USA, 1998.
[Altshuller 1999] Altshuller, G., "The Innovation Algorithm" (Translated by L. Shulyak,
S. Rodman), Technical Innovation Centre Inc., USA, 1999.
[Ayres 1998] Ayres, R. U., “Toward a zero-emissions economy”, Environmental
Science and Technology, V. 32, No. 15, August 1, 1998, pp 366-367.
[Bechler 1997] Bechler, B., “Evolution of the virtual enterprise”, Annual International
Conference Proceedings - American Production and Inventory Control Society, 1997,
pp 65-67.
[Belhe 1996] Belhe, U., Kusiak, A., "House of quality in a design process",
International Journal of Production Research, V.34, No. 8, 1996, pp 2119-2131.
[Bhaskar 1995] Bhaskar, V., and Glyn A. (eds). “The North, the South and the
Environment”, Earthscan Publications and United Nations University Press., London,
[Boothroyd 1983] Boothroyd, G., Dewhurst, P., “Design for Assembly: A Designers’s
Handbook”, Boothroyd and Dewhurst Inc., Wakefield, Rhode Island, 1983.
[Boothroyd 1994] Boothroyd, G., "Product design for manufacture and assembly",
Computer Aided Design, V. 26, No. 7, 1994, pp 505-520.
[Brezet 2001] Brezet, J.C., A.S. Bijma, J. Ehrenfeld, and S. Silvester, "The Design of
Eco-Efficient Services: Method, Tools And Review of the Case Study Based
‘Designing Eco-Efficient Services’ Project", Design for Sustainability Program, Delft
University of Technology, Delft, The Netherlands,2001.
[Broadbent 1981] Broadbent, G., 1981, "The Morality of Designing", Design: Science:
Method, Proceedings of The 1980 Design Research Society Conference," pp. 309-328,
Westbury House, Guilford, England.
[Byrne 1993] Byrne, J. G., Barlow, T."Structured brainstorming: A method for
collecting user requirements", Proceedings of the Human Factors and Ergonomics
Society, V. 1, 1993, pp 427-431.
[Chen 1994] Chen, Wei, Rosen, David, Allen, Janet K., Mistree, Farrokh, "Modularity
and the independence of functional requirements in designing complex systems",
American Society of Mechanical Engineers, Design Engineering Division (Publication)
DE, Concurrent Product Design, Proceedings of the 1994 International Mechanical
Engineering Congress and Exposition, V. 74, 1994, pp 31-38.
[Chen 2002] Chen, Li-Chieh, Lin, Li, "Optimization of product configuration design
using functional requirements and constraints", Research in Engineering Design, V. 13,
No. 4, 2002, pp 167-182.
[Cherian 2001] Cherian, R.P., Midha, P.S., Smith, L.N., Pipe, A.G., "Knowledge based
and adaptive computational techniques for concurrent design of powder metallurgy
parts", Advances in Engineering Software, V. 32, No. 6, 2001, pp 455-465.
[Clausing 1994] Clausing D.P, Total Quality Development, ASME Press, New York,
[Cohen 1992] Cohen, R., Lipton, M.G., Dai, M.Q., Benhabib, B., "Conceptual design of
a modular robot", Journal of Mechanical Design - Transactions of the ASME, V. 114,
No. 1, 1992, pp 117-125.
[Condoor 1992] Condoor, S., Shankar, R., Brock, R., Burger, P., Jansson, G.,
"Cognitive framework for the design process", American Society of Mechanical
Engineers, Design Engineering Division, V. 42, 1992, pp 277-281.
[Condoor 1999] Condoor, S. S., Weber, R. G., “Model for conceptual design
methodology”, American Society of Mechanical Engineers, Design Engineering
Division (Publication) DE, V. 103, 1999, pp 57-64.
[Coyne 1987] Coyne, R. D., Rosenman, M. A., Radford, A. D., and Gero, J. S.
“Innovation and creativity in knowledge-based CAD, Expert Systems in ComputerAided Design”, ed. J. S. Gero, Amsterdam, North-Holland, pp. 435-465. 1987.
[Coyne 1990] Coyne, R. D., Rosenman, M. A., Radford, A. D., Balachandran, M., and
Gero, J. S. “Knowledge-Based Design Systems”, Reading: Addison-Wesley. 1990.
[Cross 1980] Cross, N. J. and Walker, D., "Design Method and Scientific Method",
Design: Science: Method, Wcstbury House, Guilford. England, 1980, pp. 18-27.
[Dahmus 2001] Dahmus, J.B., Gonzalez-Zugasti, J.P., Otto, K.N., "Modular product
architecture", Design Studies, V. 22, No. 5, 2001, pp 409-424.
[de Bono 1992] de Bono E., “Serious Creativity”, Harper Collins Publishers, New York,
[DeLit 2003] De Lit, P., Delchambre, A., "Integrated design of a product family and its
assembly system", Kluwer Academic Publishers, Boston, 2003.
[Dixon 1988] Dixon. J.R., "On Research Methodology Towards A Scientific Theory of
Engineering Design," Artificial Inelegance for Engineering. Design, Analysis and
Manufacturing (AI-EDAM). Vol. 1, No. 3, 1988.
[Domb 1998] Domb, E., Kowalick, J., "Applying TRIZ to develop new designs for the
future of the airbag", SAE Special Publications, Airbag Technology, V. 1333, 1998, pp
[Doriben 2003] Doriben, H. T., Durkopp, K., "Mechatronics and drive-by-wire systems
advanced non-contacting position sensors", Science Direct, Control Engineering
Practice, No. 11, 2003, pp 191-197.
[Dorst 2001] Dorst, K., Cross, N., "Creativity in the design process: Co-evolution of
problem-solution", Design Studies, V. 22, No. 5, 2001, pp 425-437.
[Du 2001] Du, X., Jiao, J., Tseng, M., "Architecture of product family: Fundamentals
and methodology", Concurrent Engineering Research and Applications, V. 9, No. 4,
2001, pp 309-325.
[Eppinger 1990] Eppinger, Steven D., Whitney, Daniel E., Smith, Robert P., Gebala,
David A.,"Organizing the tasks in complex design projects", American Society of
Mechanical Engineers, Design Engineering Division (Publication) DE, V. 27, Design
Theory and Methodology - DTM '90, 1990, pp 39-46.
[Farrell 2001] Farrell, R. S., Simpson, T. W., "Improving commonality in custom
products using product platforms", 2001 ASME Design Engineering Technical
Conference and Computers and Information in Engineering Conference, V.2, Sep 9-12
2001, pp 1075-1085.
[Finger 1989] Finger, S. and J. R. Dixon (1989). "A Review of Research in Mechanical
Engineering Design. Part I: Descriptive, Prescriptive, and Computer-Based Models of
Design Processes," Research in Engineering Design, Vol. 1, pp. 51-67, Springer-Verlag,
New York, NY, USA.
[Fujimoto 2003] Fujimoto, J., Umeda, Y., Tamura, T., Tomiyama, T., Kimura, F.,
"Development of Service-Oriented Products Based on the Inverse Manufacturing
Concept", Environmental Science and Technology, V. 37, No. 23, 2003, pp 5398-5406.
[Gero 1985] Gero. J.S. and Coyne, R.D., "Knowledge-Based Planning as a Design
Paradigm," Design Theory in Computer-Aided Design. Proceedings of the IFIP WG 5.2
Working Conference 1985. Tokyo. Yoshikawa, H. and Warman, E.A., eds., North
Holland, Amsterdam, 1985, pp. 261-295.
[Gero 1989a] Gero, J. S. (1989), “Prototypes: a basis for knowledge-based design”, in J.
S. Gero and T. Oksala (eds.), Knowledge-Based Systems in Architecture, Acta
Scandinavica, Helsinki, pp. 11-17.
[Gero 1989b] Gero, J. S. (1989), “Knowledge-based computer-aided design”, in F.
Kimura and A. Rolstadas (eds.), Computer Applications in Production and Engineering,
North-Holland, Amsterdam, pp. 13-20.
[Goldman 1995] Goldman, S., Nagel, R., Preiss, K., “Agile Competitors and Virtual
Organizations- Strategies for Enriching the Customer”, Van Nostrand Reinhold, NY,
[Gonzalez-Zugasti 2000] Gonzalez-Zugasti, J.P., Otto, K. N., Baker, J. D., "Method for
architecting product platforms", Research in Engineering Design - Theory, Applications,
and Concurrent Engineering, V. 12, No. 2, 2000, pp 61-72.
[Gou 2003] Gou, H., Huang, B., Liu, W., Li, X., “A framework for virtual enterprise
operation management”, Computers in Industry, V. 50, No 3, 2003, pp 333-352.
[Gu 1997] Gu, P. , Hashemian, M., Sosale, S. , "Integrated modular design
methodology for life-cycle engineering", CIRP Annals - Manufacturing Technology, V.
46, No. 1, 1997, pp 71-74.
[Gu 1999] Gu, P., Sosale, S., "Product modularization for life cycle engineering",
Robotics and Computer-Integrated Manufacturing, V. 15, No. 5, 1999, pp 387-401.
[Gu 2002] Gu, P., “Adaptable Design Using Bus Systems”, Proceedings of the 10th
International Manufacturing Conference in China (IMCC2002), Xiamen, China, Oct.
11-12, 2002.
[Gu 2003] Gu, P. and Slevinsky, M., “Mechanical Bus for Modular Product Design,”
Annals of CIRP, Vol. 52, No. 1, 2003, pp. 113-116.
[Gu 2004] Gu, P., Hashemian, M., Nee A., “Adaptable Design”, CIRP Annals - The
Proceedings of the CIRP General Assembly, Poland, 2004.
[Hashemian 1995] Hashemian M., “Assisting Conceptual Mechanical Design”, Masters
thesis, University of Calgary, 1995.
[Hashemian 1996] Hashemian, M., and Gu, P., "Representation and Retrieval of Design
Knowledge for Conceptual Mechanical Design", Proceedings of the 3rd CIRP
Workshop on Design and Implementation of Intelligent Manufacturing Systems, June
19-22, 1996, Tokyo, Japan, pp. 85-94.
[Hashemian 1997-a] Hashemian, M., Gu, P., “A function representation scheme for
conceptual mechanical design”, Proceedings of the 11th International Conference on
Engineering Design (ICED 97), V. 2, 1997, pp 65-71.
[Hashemian 1997-b] Hashemian, M., Gu, P., "Modeling the Creative Knowledge of
Conceptual Design", Proceedings of the 11th International Conference on Engineering
Design (ICED 97), V. 2, 1997, pp 211-214.
[Hashemian 2004] Hashemian, M., Gu, P., "Postponing Retirement in Engineering
Design”, Proceedings of the first CDEN conference, Montreal, 2004.
[Hauschild 1998] Hauschild, M., Wenzel, H., “Environmental Assessment of Products”,
Chapman & Hall Publishers, 1998.
[Hazelrigg 1996] Hazelrigg, G.A., “Systems Engineering, An Approach to InformationBased Design”, Prentice Hall, NJ, 1996.
[Hazelrigg 1998] Hazelrigg, G.A., “Framework for decision-based engineering design”,
Journal of Mechanical Design, Transactions of the ASME, V. 120, No. 4, 1998, pp 653658.
[Higgins 1996] Higgins, James M., "Innovate or Evaporate: Creative Techniques for
Strategists', Long Range Planning, V. 29, No. 3, 1996, pp 370-380.
[Hillstrom 1994] Hillstrom, F., “Applying axiomatic design to interface analysis in
modular product development”, American Society of Mechanical Engineers, Design
Engineering Division (Publication) DE, V. 69-2, 1994, pp 363-371.
[Hubka 1980] Hubka V., “Principles of Engineering Design”, Butterworth Scientific,
London, 1980.
[Hubka 1988] Hubka, V., Eder E. W., "Theory of technical systems : a total concept
theory for engineering design", Springer, Berlin, 1988.
[Hubka 1992] Hubka V., Eder W.E., “Engineering Design”, Heurista, Zurich,
Switzerland, 1992.
[Hubka 1996] Hubka, V., Eder, W. E., "Design science : introduction to needs, scope
and organization of engineering design knowledge", Springer Verlag, London, 1996.
[Hykin 1975] Hykin, D.H.W. and Lansing, L.C., "Design Case Histories: Report of a
Field Study of Design in the United Kingdom Engineering Industry," Proceedings of the
Institute for Mechanical Engineers, Vol. 189, No. 23, 1975.
[Ishii 1994] Ishii, K., Eubanks, C.F., DiMarco, P., “Design for Product Retirement and
Material Life Cycle”, Materials and Design, V. 15, No. 4, pp 225-233.
[Ishii 1995] Ishii, K., Juengel, C., Eubanks, C.F., “Design for Product Variety: Key to
Product Line Structuring”, ASME Design Technical Conference Proceedings, 1995, V.
2, pp 499-506.
[Ishii 1998] Ishii K., “Modularity: A key concept in Product Life-cycle Engineering”,
Stanford University, Stanford, 1998.
[Iyengar 1994] Iyengar, G., Lee, Chun-Liang, Kota, S., "Towards an objective
evaluation of alternate designs", Journal of Mechanical Design, Transactions Of the
ASME, V. 116, No. 2, 1994, pp 487-492.
[Jeong 2002] Jeong, I., Leon, V. J., "Decision-making and cooperative interaction via
coupling agents in organizationally distributed systems", IIE Transactions (Institute of
Industrial Engineers), V. 34, No. 9, 2002, pp 789-802.
[Kamrani 2003] Kamrani, A. K., Gonzalez, R., "A genetic algorithm-based solution
methodology for modular design", Journal of Intelligent Manufacturing, V.14, No. 6,
2003, pp 599-616.
[Koestler 1967] Koestler, A., “The Ghost in the Machine”, Hutchinson Publishers,
[Kolodner 1983] Janet Kolodner, “Maintaining Organization in a Dynamic Long-term
Memory”, Cognitive Science, Vol. 7, (1983), pp. 243-280.
[Kolodner 1992] J. L. Kolodner, "CBR for Design", in Pu, Faltings, Maher,
Navinchandran, Oxman, (eds.), "Case-Based Design Systems", Workshop Proceedings,
AID'92 Second International Conference on Artificial Intelligence in Design, Carnegie
Mellon University, Pittsburgh, Penn., USA, 1992.
[Kolodner 1993] J. L. Kolodner, “Case-based Reasoning”, Morgan Kaufmann, 1993.
[Kourmaev 2003] Kourmaev, R., Teplitsky, A., "Application of inventing problem199
solving theory (TRIZ) in pipeline technologies", New Pipeline Technologies, Security,
and Safety, V. 2, 2003, pp 1465-1471.
[Kushnir 2003] Kushnir, E., Sheehan, T., “Development of machine tool structure at the
early stages of design process”, American Society of Mechanical Engineers, Pressure
Vessels and Piping Division (Publication) PVP, V. 470, 2003, pp 121-127.
[Lee 2003] Lee, B., Saitou, K., "Assembly synthesis with subassebly partitioning for
optimal in-process dimensional adjustability", Proceedings of the ASME Design
Engineering Technical Conference, V. 2A, 2003, pp257-268.
[Li 1998] Li H., Wu X., “learning design knowledge as generalization”, in Cybernetics
and Systems, V. 29, 1988, pp. 181-207.
[Li and Zhang 1998] Li, Q., Zhang, W.J., "Computer comparison of design knowledge",
Proceedings of the Institution of Mechanical Engineers, Part B: Journal of Engineering
Manufacture, V. 212, No. B8, 1998, pp 635-645.
[Lindahl 2001] Lindahl, M., Ölundh, G., “The Meaning of Functional Sales,” in de Ron,
A.J. (ed.): Proceedings of the 8th CIRP International Seminar on Life Cycle
Engineering – Life Cycle Engineering: Challenges and Opportunities, June 18-20, 2001,
Scientific–Technical Union of Mechanical Engineering of Bulgaria, 2001, pp. 211-220.
[Lu 2003] Lu, B., Gu, P., “Systematic Life Cycle Design for Sustainable Product
Development”, ASME 2003 design engineering technical conference, September 2003.
[Luce 1957] Luce, R. D. and H. Raiffa, “Games and Decisions”, John Wiley, NY, 1957.
[Maher 1987] Maher, M.L., "A Knowledge-Based Approach to Preliminary Design
Synthesis," Engineering Design Research Center Paper EDRC-12-14-87, Carnegie
Mellon University, 1987.
[Maher 2003] Maher, M. L., Tang, H., "Co-evolution as a computational and cognitive
model of design", Research in Engineering Design, V. 14, No. 1, 2003, pp 47-63.
[Mann 2002] Mann, D., "Manufacturing technology evolution trends", Integrated
Manufacturing Systems, V. 13, No. 2, 2002, pp 86-90.
[Markeset 2001] Markeset, T., Kumar, U. , “R and M and risk-analysis tools in product
design, to reduce life-cycle cost and improve attractiveness”,
Proceedings of the
Annual Reliability and Maintainability Symposium, 2001, pp 116-122.
[Marshall 1999] Marshall, R., Leaney, P.G., Botterell, P., "Modular design",
Manufacturing Engineer, V. 78, No. 3, Jun 1999, pp 113-116.
[Martin 2002] Martin M. V., Ishii K., “Design for variety: developing standardized and
modularized product platform architectures, Research in Engineering Design, V. 13, No.
4, 2002, pp213 - 235.
[McNeill 1998] McNeill, T., Gero, J. S. and Warren, J “Understanding conceptual
electronic design using protocol analysis”, Research in Engineering Design, V. 10, 1998,
pp 129-140.
[Meadows 1992] Meadows, D. H., Meadows, D. L., Meadows, J. R., “Beyond the
limits : confronting global collapse, envisioning a sustainable future”, Chelsea Green
Pub., 1992.
[Mebratu 1998] Mebratu, D, "Sustainability and sustainable development: Historical
and conceptual review", Environmental Impact Assessment Review, V. 18, No. 6, 1998,
pp 493-520.
[Mistree 1990] Mistree, F., Smith, W.F., Bras, B.A., Allen, J.K., Muster, D. “Decisionbased design, A contemporary paradigm for ship design”, Transactions - Society of
Naval Architects and Marine Engineers, V. 98, 1990, pp 565-597.
[NGM 2001] The National Geographic Magazine, November 2001.
[Oksana 2000] Oksana, M., "Product Service Systems – Final Report", The
International Institute of Industrial Environmental Economics, Lund University,
Sweden, 2000.
[Olewnik 2003] Olewnik, A. T, Lewis, K. E., “On validating design decision
methodologies”, Proceedings of the ASME Design Engineering Technical Conference,
V. 3, 2003, pp 747-756.
[Osborn 1957] Osborn A. F., “Applied Imagination, principles and procedures of
creative thinking”, Scribner, NY 1957.
[Ostrosi 2003] Ostrosi, E., Ferney, M., Garro, O., "A fractal-based approach for
concurrent engineering", Concurrent Engineering Research and Applications, V. 11, No.
4, 2003, pp 249-265.
[Otto 1994] Otto, K., "Robust product design using manufacturing adjustments",
American Society of Mechanical Engineers, Design Engineering Division (Publication)
DE, 6th International Conference on Design Theory and Methodology, V. 68, 1994,
[Pahl 1988]
Pahl, G., Beitz, W., "Engineering design : a systematic approach",
translated by Arnold Pomerans and Ken Wallace , edited by Ken Wallace, London,
Springer-Verlag, 1988.
[Pender 2001] Pender, S., "Managing incomplete knowledge: Why risk management is
not sufficient", International Journal of Project Management, V. 19, No. 2, 2001, pp 7987.
[Pine 1993] Pine B.J., "Mass Customization, The New Frontier in Business
Competition", Harvard Business School Press,Boston, MA, 1993.
[Pratt 1993] Pratt, M. K., Deutsch, P. B., “Application of concurrent engineering to hitech product realization”, American Society of Mechanical Engineers, Design
Engineering Division (Publication) DE, V. 65, part 1: Advances in Design Automation,
1993, pp 43-50.
[Rai 2003] Rai, R., Allada, V., “Modular product family design: Agent-based paretooptimization and quality loss function-based post-optimal analysis”, International
Journal of Production Research, V. 41, No. 17, 2003, pp 4075-4098.
[Ramaswamy 1992] Ramaswamy, R., Ulrich, K., "Augmenting the House of Quality
with engineering models", American Society of Mechanical Engineers, Design
Engineering Division (Publication) DE, V. 42, 1992, pp 309-316.
[Reich 1992] Reich, Y., 1992, "Transcending the Theory-Practice Problem of
Technology," Technical Report EDRC 12-51-92, Engineering Design Research Center,
Carnegie Mellon University, Pittsburgh, PA.
[Reich 1995]Reich Y., “A critical review of General Design Theory”, Research in
Engineering Design, 7(1): PP 1-18, 1995.
[Rifera 1999] Rivera-Becerra, A., Lin, L. “Measuring environmental consciousness in
product design and manufacturing”, Concurrent Engineering Research and Applications,
V. 7, No. 2, 1999, pp 123-138.
[Rigo 2003] Rigo, Ph., Karr, D.G., Na, S.S., Sarghiuta, R., "Comparison of rationalbased and rule-based optimum design of ship structures", Transactions - Society of
Naval Architects and Marine Engineers, V. 110, 2003, pp 435-451.
[Sanchez 2002] Sanchez, R., "Fitting together a modular approach", Manufacturing
Engineer, V. 81, No. 5, 2002, pp 216-218.
[Sand 2002] Sand, J., Gu, P., Watson, G., "HOME: House of modular enhancement - A
tool for modular product redesign", Concurrent Engineering Research and Applications,
V. 10, No. 2, June 2002, pp 153-164.
[Seliger 1997] Seliger G., Müller K., Perlewitz H., “More Use with fewer Resources”,
Proceedings of the 4th CIRP International Seminar on Life Cycle Engineering 1997,
Berlin, pp 3-16.
[Seliger 1998] Seliger, G., Grudzien, W., Müller, K. “The acquiring and handling of
devaluation.” Proceedings of the 5th CIRP International Seminar on Life Cycle
Engineering 1998, Stockholm (Sweden), pp 99-108.
[Seliger 2000] Seliger, G., Basdere, B., S. Schumann, and M. Wapler: “Technological
Conditions for Selling Use instead of Selling Products,” in Kimura, F. (ed.), the
Proceeding of the 7th CIRP International Seminar on Life Cycle Engineering,
November 27-29, 2000, The University of Tokyo, Tokyo, 2000, pp. 17-22.
[Simon 1969] Simon, Herbert, “The Sciences of the Artificial”, Oxford University Press,
[Simpson 1998] Simpson, T.W., Rosen, D., Allen, J.K., Mistree, F., “Metrics for
assessing design freedom and information certainty in the early stages of design”,
Journal of Mechanical Design, Transactions of the ASME, V. 120, No. 4, 1998, pp 628635.
[Simpson 2001] Simpson, T.W., Maier, J.R.A., Mistree, F., "Product platform design:
Method and application", Research in Engineering Design - Theory, Applications, and
Concurrent Engineering, V.13, No. 1, 2001, pp 2-22.
[Simpson 2003] Simpson, T. W., "Product platform design and optimization: Status and
promise", Proceedings of the ASME Design Engineering Technical Conference, V. 2A,
2003, pp 131-142.
[Sosale 1997] Sosale, S., Hashemian, M., Gu, P., "Product modularization for reuse
and recycling", American Society of Mechanical Engineers, Design Engineering
Division (Publication) DE, V. 94, Concurrent Product Design and Environmentally
Conscious Manufacturing, 1997, pp 195-206.
[Stanley 2001] Stanley, R., "Enumerative Combinatorics", Cambridge University Press,
[Strawbridge 2002] Strawbridge, Zeke, McAdams, Daniel A., Stone, Robert B., "A
computational approach to conceptual design", Proceedings of the ASME Design
Engineering Technical Conference, V. 4, 2002, pp 15-25.
[Suh 1990] Suh, N. P., "The principles of design", New York, Oxford University Press,
[Suh 2001] Suh, N. P., “Axiomatic Design”, New York, Oxford University Press, 2001.
[Sydenham 2004] Sydenham, P. H., “Systems Approach to Engineering Design”,
Artech House Inc., Boston, London, 2004.
[Taguchi 1987] Taguchi, G., “System of Experimental Design”, vol. I and vol. 2,
UN1PUB-Kraus International Publications (White Plains) and American Supplier
Institute, Inc. (Dearborn), 1987.
[Tate 1995] Tate D., Nordlund M., “Synergies Between American and European
Approaches to Design”, Proceedings of the First World Conference on Integrated
Design and Process Technology (IDPT-Vol. 1), Society for Design and Process Science,
Austin, TX, pp. 103-111, December 7-9, 1995.
[Tebay 1984] Tebay, R., Atherton, J., and Wearne, S.H., "Mechanical Engineering
Design Decisions: Instances of Practice Compared with Theory," Proceedings of the
Institute for Mechanical Engineers, Vol. 1988, No. 6, 1984.
[Tipnis 1998] Tipnis, V. A., “Evolving Issues in Product Life Cycle Design: Design for
Sustainability”, in Handbook of Life Cycle Engineering, Kluwer Academic Publications,
1998, pp 413-460.
[Tolle 2003] Tolle, M., Vesterager, J., “VEM: Virtual enterprise methodology”, VTT
Symposium (Valtion Teknillinen Tutkimuskeskus), No. 224, 2003, pp 53-70.
[Tomiyama 1993] Tomiyama, T., Umeda, Y., Yoshikawa, H., "CAD for functional
design", CIRP Annals, V. 42, No. 1, 1993, pp 143-146.
[Tomiyama 1994] Tomiyama, T., "From general design theory to knowledge-intensive
Manufacturing: AIEDAM, V. 8, No. 4, 1994, pp 319-333.
[Tomiyama 1995] T. Tomiyama, T. Sakao, Y. Umeda, Y. Baba, “The Post-Mass
Production Paradigm, Knowledge Intensive Engineering, and Soft Machines,”, in
Krause, F.-L. and H. Jansen (eds.), Life Cycle Modelling for Innovative Products and
Processes, Chapman & Hall, London, 1995, pp. 369-380.
[Tomiyama 1996] Tomiyama T, Yoshikawa H., "Extended general design theory,
Design Theory for CAD", Elsevier Science Publishers, 1987. pp 95-130.
[Tomiyama 1997] Tomiyama, T., “A Manufacturing Paradigm Toward the 21st
Century,” Integrated Computer Aided Engineering, Vol. 4, 1997, pp. 159-178.
[Tomiyama 1999] Tomiyama, T., “The Post Mass Production Paradigm,” in Yoshikawa,
H., R. Yamamoto, F. Kimura, T. Suga, and Y. Umeda (eds.), Proceedings of the First
International Symposium on Environmentally Conscious Design and Inverse
Manufacturing (EcoDesign ’99), February 1-3, 1999, Tokyo, IEEE Computer Society,
1999, pp. 162-167.
[Tomiyama 2000] Tomiyama, T., “Knowledge Intensive Engineering towards
Sustainable Products with High Knowledge and Service Contents,” in Horváth, I., A.J.
Medland, and J.S.M. Vergeest (eds.), TMCE 2000, Third International Symposium on
Tools and Methods of Competitive Engineering, April 18-20, 2000, Delft University
Press, Delft, the Netherlands, 2000, pp. 55-67.
[Tribus 1969] Tribus, M., 1969, “Rational Descriptions, Decisions and Designs”,
Pergamon Press, Elmsford, NY.
[Tsai 1999] Tsai, Y., Wang, K., “Development of modular-based design in considering
technology complexity”, European Journal of Operational Research, V. 119, No 3, 1999,
pp 692-703.
[Tsai 2004] Tsai, H., Hsiao, S., "Evaluation of alternatives for product customization
using fuzzy logic", Information Sciences, V. 158, No. 1-4, 2004, pp 233-262.
[Tseng 1996] Tseng, M. M., Jiao, J., Merchant, M. E., "Design for mass customization",
CIRP Annals - Manufacturing Technology, V. 45, No. 1, 1996, pp 153-156.
[Tseng 1998] Tseng, M., Du, X., "Design by customers for mass customization
products", CIRP Annals - Manufacturing Technology, V 47, No. 1, 1998, pp 103-106.
[Ullman 1987] Ullman, D.G., Stauffer, L.A., and Dietterich, T.G., "Preliminary Results
of an Experimental Study of the Mechanical Design Process," Proceedings from the
NSF Workshop on the Design Process, Waldron, M.B., ed., Ohio State University,
Oakland, CA, February 8-10, 1987, pp. 145- 188.
[Ullman 1992] Ullman, D. G. ,"The Mechanical Design Process", McGraw-Hill Inc.,
[Ullman 1994] Ullman, David G., "Issues critical to the development of design history,
design rationale and design intent systems", American Society of Mechanical Engineers,
Design Engineering Division (Publication) DE, V.68, 1994, pp 249-258.
[Ulrich 1991] Ulrich, K., Tung, K., "Fundamentals of product modularity", American
Society of Mechanical Engineers, Design Engineering Division (Publication) DE, V. 39,
Issues in Design/Manufacture Integration, 1991, pp 73-79.
[Wallace 2000] Wallace, K.M., Blessing, L.T.M., "Observations on some German
contributions to engineering design in memory of professor Wolfgang Beitz", Research
in Engineering Design - Theory, Applications, and Concurrent Engineering, V. 12, No.
1, 2000, pp 2-7.
[Wassenaar 2001] Wassenaar, Henk Jan, Chen, Wei, “An approach to decision-based
design”, Proceedings of the ASME Design Engineering Technical Conference, V. 4,
2001, pp 21-30.
[WBCSD 1998] WBCSD, “Cleaner Production and Eco-Efficiency – Complementary
Approaches to Sustainable Development”, The World Business Council for Sustainable
Development and United Nations Environment Program – Industry and Environment,
Paris, 1998.
[Webb 2002] Webb, Alan, "TRIZ: An inventive approach to invention", Manufacturing
Engineer, V. 81, No. 4, 2002, pp 171-177.
[Whitney 1993] Whitney D., Nevins J., DeFazio T., Gustavson R., "Problems and
Issues in Design and Manufacture of Complex Electromechanical Systems", C. S.
Draper Lab Report R-2577, December 1993.
[Willems 2003] Willems, B., Seliger, G., Duflou, J., Basdere, B., “Contribution to
Design for Adaptation: Method to Assess the Adaptability of Products (MAAP)”,
Proceedings of EcoDesign2003: Third International Symposium on Environmentally
Conscious Design and Inverse Manufacturing, Tokyo, Japan, December 8-11,2003.
[Yoshikawa 1981] Yoshikawa, H., "General Design Theory and a CAD System", ManMachine Communication in CADICAM. Proceedings of the IFIP WGS.2/5.3 Working
Conference 1980, Tokyo, North Holland, Amsterdam, 1981, pp. 35-58.
[Yoshimura 2003] Yoshimura, M., Izui, K., Fujimi, Y., "Optimizing the decision-
making process for large-scale design problems according to criteria interrelationships",
International Journal of Production Research, V.41, No. 9, 2003, pp 1987-2002.
[Yu 1999] Yu, J.S., Gonzalez-Zugasti, J.P., Otto, K.N., "Product architecture definition
based upon customer demands", Journal of Mechanical Design, Transactions of the
ASME, V. 121, No. 3, 1999, pp 329-335.
[Yu 2003] Yu, T., Yassine, A., Goldberg, D. E., "A genetic algorithm for developing
modular product architectures", Proceedings of the ASME Design Engineering
Technical Conference, V. 3, 2003, pp 515-524.
[Zeng 1999-a] Zeng, Y., Gu, P., "Science-based approach to product design theory. Part
I: Formulation and formalization of design process", Robotics and Computer-Integrated
Manufacturing, V. 15, No. 4, 1999, pp 331-339.
[Zeng 1999-b] Zeng Y, Gu P., "A science-based approach to product design theory. Part
II: formulation of design requirements and products", Robotics and Computer
Integrated Manufacturing, V. 15, No. 4, 1999, pp 314-352.
[Zha 2002] Zha, Xuan F., Lu, Wen F., "Knowledge intensive support for product family
design", Proceedings of the ASME Design Engineering Technical Conference, 28th
Design Automation Conference, V. 2, 2002, pp 603-612.
[Zwicky 1948] Zwicky, F., “Morphological Analysis and Construction”, Wiley Interscience, New York, 1948.
[Zwicky 1969] Zwicky, F., “Discovery, Invention, Research through Morphological
Analysis”, MacMillan, New York, 1969.
Appendix 1: A Function Representation Scheme
for Conceptual Mechanical Design
The function of a designed object represents the intent or purpose of creating that object
in terms of what it must do. The intent of the design object depends on the design
domain. For example, in the design of an organization the purpose might be to provide a
certain type of service to customers. In the context of mechanical design, which without
mentioning will be the context of all discussions hereafter, the term “function” refers to
a physical function, as defined below:
Definition: The function of a device in mechanical engineering design is to effect
its physical surrounding in an intended manner.
From this definition we will draw a guideline to determine whether a requirement is or
is not function in the context of mechanical engineering design.
A non-physical requirement such as “cost”, in the context of
mechanical engineering, is not a function. A requirement is a
physical requirement (a function) if it can be described as some
effects on the physical environment that consists of material, energy,
signal, and force.
Thus, in order to define the function of an object we must describe both the surrounding
of the object (physical environment) and the effects of the object on the physical
environment. This appendix presents a taxonomy of entities to describe the physical
world, and a taxonomy of different types of effects that an object may have on its
physical environment. Then it shows a scheme for the construction and representation
of mechanical functions using entities chosen from these taxonomies.
A1.1. Function Operands (Physical Entities)
A function is defined as the effect of an object on some entities that describe the
physical environment around the object. The entities whose attributes are affected by an
object are called "function operands". The initial and final states of function operands
are called the inputs and outputs of a function; inputs and outputs are not necessarily
physical entities that enter or exit the design object. Therefore, the definition of a
function requires the identification of the entities that sufficiently describe the physical
world for the purpose of mechanical design.
First, “material” is recognized as a category of physical entities. This category presents
all physical objects in the environment, including solids, liquids, and gases. The second
type of entity is "energy". Energy in some cases can be treated as an attribute of
material. For instance, the thermal energy of a hot piece of metal may be considered an
attribute of that material object. However, energy might not require a material carrier,
thus cannot be expressed as an attribute of material (e.g. the electro-magnetic energy).
Also, in many cases the material carrier is of no importance to the design task. For
example, the input to a gearbox is better described as rotational energy than described
as a shaft with rotational energy as one of its attributes. For these reasons energy is
considered an independent category of physical entities.
The third category recognized is “force”, which is a measurable phenomenon observed
as pulling or pushing things. Force cannot be described by materials, thus this category
is independent from the first category. However, it is often considered as belonging to
the category of energy in the design literature [Pahl 1988]. A reason for not considering
force as an independent category is that the practical application of “force” in
engineering is typically associated with ‘displacement’, thus force can be described by
energy. Similar reasoning may also question the independence between materials and
energies. These discussions are avoided here because the classification of entities may
be made arbitrarily for the purpose of developing a practical representation scheme for
mechanical functions.
The above three categories can describe a broad range of entities. However, physical
objects usually are not disconnected entities without any "order". The arrangement of
multiple objects, their shapes, and the temporal sequence of events carry "information".
In engineering design, the information conveyed in the existing order among and within
entities represents the human notion of the meaning, behavior, or purpose of entities.
For example, a word written on a piece of paper contains information; which is a
particular order between the material entities (the paper and ink). Information usually
requires a physical carrier in the form of material or energy; however the carrier itself
may not be important to the design problem. Therefore, information is recognized as an
independent category, which is not expressed by the previous three categories. In
Systematic Design, information together with its physical carrier is called "signal" [Pahl
1988]. By this definition, signal is the physical manifestation of information; signal
refers to a certain order that exists in the properties and attributes of entities and events.
It is necessary to discuss whether or not “time” should be treated as an independent
entity. There is no device whose function is to affect time; therefore time cannot be a
function operand and is not considered an independent entity. Instead, time is an
important part of "actions" that describe the effect of an object on its physical
environment. Actions and the role of time in describing them will be discussed in the
next section.
In the rest of this appendix, we refer to the above four basic categories, material, energy,
force and signal, as "principal entities" or PEs for short. PEs in turn can be sub-divided
into more specific types. For example material can be solid or liquid or gas. PEs and
their sub-types form a taxonomy of function operands. Figure A1.1 shows a prototype
taxonomy of function operands developed for the description of mechanical functions.
couple (torque)
rotary (rotating torque)
linear (moving force)
hydraulic (flow)
pneumatic (flow)
muscular (human-generated)
compression (gas)
data (information,
not an attribute)
condition (status or attribute of operands)
activator (signal activates action)
input (signal needed for action continuation)
controller (values of outputs for given inputs)
Figure A1. 1: The hierarchical taxonomy of function operands.
Locating an entity in the above taxonomy does not sufficiently describe it. To describe
entities precisely, either the sub-division of types must continue into minute details, or
the characteristics of entities must be further specified by their attributes. Exhaustive
sub-division is not practical since for almost every new entity a new sub-category must
be defined which requires the continuous modification of the taxonomy. On the other
hand, using attributes is advantageous since attributes (e.g. velocity, length, weight,
dimension, color, smell, frequency, taste, magnitude, direction, etc.) can be shared by
all relevant entities.
We have developed a pool of attributes, where each attribute is represented by a list of
"item-value" pairs. First an entity (function operand) is selected from the above
taxonomy. Then it is further specified via selecting its relevant attributes from the
attribute pool and evaluating the values for attributes. For example the entities
"rotational mechanical energy" and "metal bar" require attributes "torque" and "length"
Table A1.1 shows some of the attributes developed for entities. The column "type"
shows whether an attribute is specified by numerical or symbolic values. The column
"representation" indicates how the value of an attribute is represented in our scheme, e.g.
by lower and upper limits or by strings. The column "access" shows the rules of
accessing operands in databases. For example, numerical attributes can be accessed by
"greater than" or by their exact values. The column "taxonomy" shows where in the
operand taxonomy (and all branches thereafter) the operand is applicable. In the "Type"
column of the table, n[L-U] indicates that the attribute value is numerical and is
presented by lower and upper limits; strings are chosen from a set of pre-specified
words; code string is a particular type of string which is either generated by the
standard vocabulary or is copied from an operand within the function.
We have found it more practical not to consider the "location" of an entity as one of its
attributes; otherwise we would have to define a fixed spatial reference in order to
specify the location of entities. The absolute location of entities with respect to a fixed
reference is rarely of importance in mechanical design problems. Instead, a change in
location (or prevention of change) is usually important. We deal with location in the
"action" part of a function, to be discussed in the next section.
The four categories of entities in Figure A1.1, together with their attributes in Table
A1.1, can generally describe most entities of interest in conceptual mechanical design.
Since the location of an entity is not considered as one of its attributes, an entity may be
described by its type in the PE taxonomy, the value of its relevant attributes, and its
n, [L - U]
material, gravity
liquid, gas
n, [L - U]
liquid, gas
n, [L - U]
n, [L - U]
n, [L - U]
n, [L - U]
n, [L - U]
body, surface
n, [L - U]
n, [L - U]
n, [L - U]
force-single, mechlinear, muscular
circle, square, etc.
wood, plastic, metal, …
type of force
n, [L - U]
force-couple, mechrotary
n, [L - U]
n, [L - U]
n, [L - U]
n, [L - U]
n, [L - U]
n, [L - U]
n, [L - U]
n, [L - U]
n, [L - U]
material, thermal
n, [L - U]
mat-liquid, mat-gas, hydraulic, pneumatic, compression
flow rate
n, [L - U]
pneumatic, hydraulic
code string
specifies the target condition
that is being monitored
code string
carrier of a signal
frequency of direction variation
Table A1. 1: The list of operand attributes.
A1.2. Actions
We regard the function of an object as its effect on its physical surrounding. This effect
might be to change the characteristics of function operands or it might be to prevent
them from changing. In our scheme, the part of a function that describes such effects is
called the "action". An action affects the properties of one or multiple function operands,
which are physical entities. It was mentioned in the previous section that entities are
specified by their types, attributes; and location. Since an action may affect any of these
parameters, three types of actions can be identified: change-type, change-attributes, and
change-locations of entities.
Also, for two reasons we introduce the action "change-number" (connect or separate) as
a distinct type of action. First, entities from the categories of "energy" and "signal" may
need physical carriers. It may be necessary to add the carriers to these entities in the
functional structure; or it may be necessary to remove the carriers. The addition or
removal of entities can be achieved via the actions “connect” and “separate”. For
example, when mechanical energy contracts a spring, we may consider this function as
"connect energy and material". Second, the function of a device can be to split or join
entities in circumstances that keeping track of the original entities is no longer useful to
the design task. For example, consider a machine that mixes two substances for further
processing in a larger system. The mixture may be then represented as one entity in the
function structure of the overall system when there is no need to trace its ingredients.
The action “connect” represents the function of the mixer, which is to replace the initial
state (two entities) with the final state (one entity) in the functional structure.
Therefore, this function representation scheme recognizes four main types of actions:
change_type, change_attribute, change_location, and change_number. The opposite of
an action is also considered an action. That is, ‘prevent from change’ may be applied to
the type, attribute, location, and number of entities. For example, the action of a thermal
insulation material can be described as “prevent from change in location”. Further
detailing of action types generates a hierarchical taxonomy similar to the one developed
for operands. The taxonomy of actions is shown in Figure A1.2.
preserve (prevent from change)
maintain (prevent from change)
channel (provide a conduit or medium)
guide (channel with controlled direction)
transport (carry entities, provide drive)
store (fix location for a time)
contain (isolate material from surrounding)
isolate (don’t provide media for channel)
Keep (prevent
from a change
in location)
hold (signals control time, involves force)
restrict-motion (stop + DOF, no signal or time)
position (hold + position and orientation)
connect (create an entity from entities, decrease number,
may have physical significance or may represent loss of
interest in an entity in the function structure)
extract (also create from)
controlled (controlled separation such as sifting)
uncontrolled (uncontrolled separation like split)
Figure A1.2: The taxonomy of actions.
In the same way principal entities were further specified by their attributes, actions are
further specified by their "specifiers". Action specifiers explain the way an action is
performed. For instance, "distance: 2 meters", "time: 12 seconds", and "accuracy: 1%"
are three specifiers for actions "transfer", "store", and "change-attribute" respectively.
Table A1.2 shows a partial list of the specifiers developed for our pilot computer system.
n, [L - U]
n, [L - U]
n, [L - U]
X, Y, Z, CW, CCW
n, [L - U]
duration of an action
n, [L - U]
values in %
X, Y, Z, CW, CCW
code string
change-type and
shows what sub-type or
attribute of the input changes
code string
code string
infinity, for continuous
Table A1. 2: The list of action specifiers.
Figure A1.3 summarizes the elements of function operands and actions. A function
operand is described by its type, attributes and its "connection code". Connection code
describes how the operand is connected to its action, and explains the role of an operand
of a function within a functional structure. For example, if the connection code of an
operand is “input”, this operand represents the initial state of the physical environment;
if the code is “output”, the operand represents the final state of the physical
environment after the function has been executed. Thus the relation between actions and
operands is described as a part of ‘operands’, and an action can be simply described by
its two elements: its type chosen from the action taxonomy, and its specifiers chosen
from the list of available specifiers.
Entity (Function Operand)
sub-type att_1
sub-type att_2
Connection Code
input, output,
activating signal,
utilized energy
Figure A1.3: The constituting elements of function operands and actions.
A1.3. The Structure of a Function
The previous two sections described two constituting elements of functions: function
operands and actions. Function operands represent the physical entities that are affected
by the function of an artifact; an action describes the effect. This section explains how
these elements are assembled together to build functions.
Basic Functions
In the proposed function representation scheme, a function is constructed from certain
building blocks called basic functions. A basic function consists of one action and a
number of function operands. Any number of basic functions may be used to
sufficiently describe a function. A simple function that can be sufficiently described by
one action consists of one basic function. A simple function has all the properties of
complex functions and can be decomposed or altered. Figure A1.4 shows how a basic
function is constructed from actions and operands, specified by their types/specifiers
and types/attributes respectively.
list of
list of
Figure A1.4: Elements of a basic function in the proposed scheme.
Constructing Larger Functions
A function may be expressed in terms of a number of basic functions and is recognized
in the functional structure by its function boundary. Larger functions are composed of
other functions. Thus, we can build a hierarchical functional structure where the overall
function is constructed from other functions, and these functions from smaller functions
down to basic functions at the lowest level of the hierarchy. Figure A1.5 shows the
structure of a function.
a) Structure of a Basic Function
boundary of the
overall function
boundary of a
simple function
boundary of a highlevel function
b) The Overall Function of a system
Figure A1.5: Construction of functions in the proposed scheme.
In the proposed scheme a function operand (an entity) is described by its type, selected
from the taxonomy, and its attributes, selected from the attribute list. There are only a
small number of types and attributes to choose from. However, the designer can
describe a large number of physical entities through the morphological combination of
different types and attributes. Similarly, actions are constructed by their types and their
specifiers. A large number of actions can be constructed through different combinations
of action types and specifiers. Also, actions and operands may be combined in various
ways to produce a large number of possible basic functions, and any number of basic
functions can be used to describe a function. Thus, the proposed scheme provides for
the possibility of morphological combination of different elements within a function. As
a result, a small vocabulary can describe a large number of complex functions.
A1.4. Examples
In this section a few examples are provided to demonstrate the expressive power of the
proposed function representation scheme. For each example, we consider a device and
its nominal function, and then show how this function can be described using the
proposed scheme.
Example 1: an analogue measurement device (e.g. a thermometer)
The inputs of this function are the temperature to be measured and a pre-specified scale.
The first operand is of the type “signal-condition”. It is further specified by the attribute
“target attribute”. The second input, also of type signal, is recorded data. The action is
to “connect” these two signals in order to generate a new display signal. Generally, the
function "compare" can always be described as connecting input signals to produce an
output signal. Also, the function "measure" is a special case of "compare" in that a
signal is compared against pre-specified units. This example is illustrated in Figure
Function boundary
SignalÎ dataÎvisualÎanalogue
Î connect
Figure A1.6: Representing the function of a thermometer.
Example 2: a current controller for a servomotor (e.g. for speed control)
This function includes the “measure” function explained above, as well as other actions
to change the value of electrical energy. The function consists of three basic functions.
First, the signal that is the condition of the monitored quantity (e.g. speed) is connected
to a pre-specified target value to produce the difference signal. This signal is a
command pulse that can have a carrier of type, for example, electrical energy. Then this
signal is connected to a source of energy to release the exact amount of energy required
to correct the initial value. This energy is then connected to the main source to produce
the final output. This representation in general can describe the function "control". It is
shown graphically in Figure A1.7.
signal: condition (attribute of operand)
signal: command
signal: data: recorded (target value)
energy: electrical
energy: electrical (correcting energy)
energy: electrical (main source)
energy: electrical (output)
Figure A1.7: Representing the function of a controller.
Example 3: a shaft
This is a simple example that does not involve a signal, since a shaft transmits torque
unconditionally. This function is represented as shown in Figure A1.8:
Figure A1.8: Representing the function of a shaft.
The following examples show how some common functions are represented in the
proposed scheme:
Compare = change-numberÆconnect (signal + signal = signal)
Measure = a type of compare
Control = compare + change-attributeÆvary + change-attributeÆvary
Store = change-locationÆkeepÆstore
Create = change-numberÆseparateÆextract
A1.5. Software Implementation
We have developed a pilot software system (TARRAUH) for the proposed function
representation scheme. The interface includes a canvas divided into two areas as shown
in the following figures. The left area of a canvas is the hierarchy of a functional
structure, and the right area shows the details of functions and their constituting ‘basic
functions’, as well as the relations between all the elements involved in the construction
of functions.
A1.5.1. Operands
TARRAUH allows the user to represent an operand by its type and its attributes using
the standard vocabulary. This is illustrated in Figure A1.9. Each operand is represented
by a link with a red node and a blue node on its two ends. The thickness and color of a
link shows the type of operand (e.g. material or energy). A link attaches to "actions"
through its nodes. When the red node attaches to an action, it indicates the output of that
action. When the blue node attaches to an action it indicates the input to that action. The
unattached nodes are larger in size. Such large nodes indicate the main inputs and
outputs of the overall function. TARRAUH pops up the proper dialogue boxes so that
the user can select the right "type" of the operand from the taxonomy through menu
options. This is shown in the figure.
Figure A1.9: Specifying the types of function operands.
Once the type of an attribute is defined, TARRAUH automatically retrieves the relevant
attributes from the attribute pool and asks the user to enter their values. For numerical
attributes there are boxes for lower and upper limits of their values. For "string"
attributes, a small pull-down menu allows the user to select a word from the list of
possible strings that are available in the program. This is illustrated in Figure A1.10.
Figure A1.10: Specifying the relevant attributes of a function operand (solid, material).
A1.5.2. Actions
In TARRAUH actions are denoted by small boxes. There are four different shapes of
boxes for the representation of four main actions (e.g. change-type and change-location).
This is demonstrated in Figure A1.10.
TARRAUH allows the user to specify the ‘type’ of an action by picking the right item
from within the dialogue boxes (Figure A1.11). Actions together with their operands are
the building blocks for constructing functions. Thus, actions are put together to build a
function. TARRAUH shows the grouping of actions under a function in the left area of
the canvas as shown in Figures A1.10 and A1.11.
Figure A1.11: Different "types" of actions in TARRAUH.
Once the type of an action is fully specified, TARRAUH pops up the proper dialogue
boxes that contain pull-down menus to select the action specifiers. This is shown in
Figure A1.12. The figure shows the relevant specifiers for the action changeattributeÆdecrease. In this case the relevant specifiers are no-of-ratios, target-attribute,
and ratio.
Figure A1.12: Quantifying "action specifiers" for the actions of known types.
A1.5.3. Building Larger Functions
In the proposed scheme, functions consist of building blocks and also of other smaller
functions, resulting in a hierarchic functional structure. This hierarchy exists in all
design problems regardless of the representation scheme that is used. Generally, in a
design problem a functional structure is the result of a decomposition process. In a
redesign task, however, the functional structure of a product might be initially
constructed by studying an existing product. We recognize the following two distinct
and equally important characteristics in a functional structure.
First, a functional structure is a graph that can be also called a network. The nodes of
this network are functions at various levels of the hierarchy down to the end-node
functions. The “end” level of decomposition is where functions can be systematically
replaced by readily-available solutions, that is, the solutions which can be obtained from
outside sources and their further decomposition is not a part of the design task in hand.
The links within this network represent different types of functional interactions among
functions or function carriers (solutions).
Second, a functional structure is a hierarchical structure that can be called a tree. This
tree shows the decomposition history hence the design rationale. The initial functional
requirements of a design problem are the root of the tree. These complex functions may
require complex solutions which are not readily available; therefore they are
decomposed into the functional requirements of their solutions. The decomposition
process continues until readily-available solutions for all functions are found.
Figure A1.13 schematically shows these two important aspects of a functional structure.
The network is represented as several end-node functions that are linked together
through their sharing operands. The tree is represented by a series of nested function
boundaries. From a top-down view, the tree is the nested segmentation of the overall
functions; and from a bottom-up view, it is a recursive grouping of smaller functions.
Basic functions
Functions directly composed of basic functions
Function boundaries
Complex functions built up of other functions
The overall function of the system
Function operands (inputs and outputs)
Figure A1.13: Representing networks and trees in a functional structure.
TARRAUH is capable of representing both the tree and the network in a functional
structure. The right area of the canvas represents the network. The left side represents
the decomposition tree. The functions on the left are directly linked with the "basic
functions" on the right. Any changes in either side are automatically reflected in the
other. Functions can be moved and regrouped on the left side. This is equivalent to
shifting function boundaries in the above figure.
The following figure shows how the function of a car, which is to transport people and
objects, is represented and decomposed to form a functional structure. The
decomposition history is represented on the left. The function network is represented on
the right by actions and operands.
Figure A1.14: Representation of the function of a car in TARRAUH.
Appendix 2: Four-Wheel Servo Steering
This appendix discusses some added capabilities which can be made possible in a fourwheel servo steering vehicle.
A2.1. Steering Axis Offset
In front wheel steering, which was depicted in Figure 5.17 of Chapter 5, the
instantaneous center of rotation (ICR) is always located on a line that connects the rear
wheels. Thus, steering action can be translated to the shifting of the location of ICR on
this line. The distance between ICR and the centerline of the vehicle is the only variable
parameter in steering. This distance varies between infinity (vehicle moving straight)
and the minimum turning radius (when the driver steers to the maximum).
When rear wheels also turn, ICR can be anywhere on the two dimensional plane. If we
call the line drawn from ICR perpendicular to the vehicle’s centerline ‘the steering axis’,
then two variable parameters can be defined for this steering mechanism. One is the
location of the steering axis on the vehicle’s centerline, and the other one is the location
of ICR on this axis.
Here we measure the location of the steering axis on the vehicle’s centerline from the
rear axis, and call this parameter the ‘offset’. When the offset is zero, the vehicle
behaves like a regular front wheel steering car. When the offset equals the distance
between the front and rear axes, denoted by L, the vehicle behaves like a warehouse
forklift. Figure A2.1 shows the shifting of the steering axis and the effect of this offset
on steering angles.
0 < Offset < L
Offset = L/2
Steering axis in the middle
Steering axis shifted forward
Offset = L
Rear steering
Figure A2. 15: The offsetting of the steering axis.
Since there are two variable parameters in this system, there could also be two driver
control devices. One is the steering wheel which determines the location of the ICR on
the steering axis, and the other is a shifter which determines the offset of this axis from
the rear wheels. The latter determines the ‘feel’ or ‘behavior’ of steering as discussed
above. This feature also greatly enhances the maneuverability of the vehicle.
A2.2. Unlimited Steering
In the absence of mechanical connectors between wheels, it is possible to design a
steering mechanism for unrestricted turning of the wheels. Figure A2.2 illustrates a
conceptual design for the unrestricted steering of a slow moving vehicle. The wheel is
mounted on a long bracket which can spin around the vertical axis. This bracket has a
gear on top that engages the pinion of a servomotor. The figure shows this gear and the
servomotor’s mounting holes on the chassis, but it does not show the servomotor itself.
Figure A2. 16: Unrestricted turning of the wheel.
With such unrestricted steering as the one provided by the above design, the distance
between the ICR and centerline can be made as small as zero. This enables the vehicle
to turn in a very small radius, or even turn around on the spot.
A2.3. Adjustable Sensitivity
Here the term ‘sensitivity’ refers to the ratio between the driver’s steering and the actual
turning of the wheels; high sensitivity means that for a small turning of the steering
wheel the vehicle turns to a large degree. As a safety feature, it might be desirable to
vary the sensitivity of steering adversely proportional to the ‘speed’ of the vehicle; as a
performance feature, it might be desirable to vary the sensitivity of steering for different
turning radii. Figure A2.3 illustrates this adjustability.
Steering limit
Driver’s steering
More sensitive
(low speed)
Less Sensitive
(high speed)
Turning Radius
Figure A2. 17: Adjustable sensitivity in servo steering.
The horizontal axis on top is the driver’s steering, ranging from zero to the device limit.
For example, if a steering wheel is used as the driver’s interface this limit might be 720
degrees (two whole turns). The vertical axis on the left is the turning radius that varies
from infinity to a minimum, which in this design can be as small as zero. The figure
shows three sensitivity curves for three different speeds. Also, the steering radii are
color coded. Red represents a radius smaller than half of the wheelbase, which means
the vehicle spins around itself and the inner wheels have to move in the reverse
direction of the outer wheels. The green zone represents larger thus safer radii, and the
yellow zone represents smaller radii which are still large enough to keep all four wheels
moving in the same direction. It can be seen that as the speed increases, the sensitivity is
reduced. The high speed curve on the right requires more steering to achieve the same
radius that would be achieved by a small steering at low speed. Further, it can be seen
that the high speed curves do not reach the red zone. These curves are examples of
many performance profiles which can be easily programmed for the servo steering
A2.4. Wheel Alignment
Wheel alignment in a conventional steering mechanism assures that at ‘zero steering’ all
four wheels are paralleled along the vehicle’s centerline. Once this is achieved, the
differential rotation of right and left wheels is automatically achieved through the
geometry of steering knuckles. In the servo steering mechanism proposed here, wheel
alignment means both the calibration of servo motors and the resetting of ‘zero
rotation’ for straight driving.
The resetting process is similar to the process of conventional wheel alignment. All
wheels are paralleled along the centerline and the reference point (zero) for servo
motors and for their associated feedback sensors is reset to zero. The calibration process
involves the ‘training’ of servo motors in order to achieve the desired sensitivity
discussed in Section A2.3. Both resetting and calibration can be performed easily
because these processes only require the adjustment of software variables.
Appendix 3. Task and Information Processing
Chapter 4 defined a task as a process that begins with a set of goals and ends when these
goals are achieved according to the task’s design. Design is a plan, recipe, or a set of
instructions for performing a task. Execution of a design requires work. This thesis uses
the term information processing (IP) to refer to the ‘work’ or ‘effort’ spent to execute a
design. The reason for this terminology is that any work can be theoretically translated
into the execution of primitive instructions.
A task has a hierarchical structure, and is always decomposed into smaller tasks. As
explained in Section 4.1, the decomposition ends arbitrarily depending on which
detailed levels of subtasks are pertinent to the task at hand. For example, the
decomposition of a task may continue to extremely detailed levels such as the
movements of individual basic particles, if such details are important. However, such
level of detail is rarely relevant to an engineering task. A reasonable end-level of
decomposition detail for engineering tasks might be to treat the human body with all its
functionalities and capabilities as an available resource. In this case, the decomposition
may end when the design is broken down into a set of “primitive instructions” (e.g.
neural pulses) that will create the required body motions to finish the task.
In practice, however, this level of detail is unnecessary. Decompositions actually end at
much higher levels than individual body movements. Humans can learn procedures and
skills, and can on their own plan the required course of action for simple tasks. Thus
these skills may be treated as a higher-level available resource; these resources can then
internally generate the required primitive instructions.
Chapter 4 explained that a design (set of instructions) can be treated as an end-node
only if its fulfillment can be assured by the available resources and thus does not require
further decomposition by the designer. The above discussion reveals that such resources
are available at different levels and are capable of performing tasks at different levels of
complexity. For example, a manufacturing firm can be treated as an available resource.
Once the manufacturing drawings are given to this entity, the delivery of the physical
product is reasonably assured, and there is no need for the designer to decompose the
task and specify the details of production processes.
Figure A3.1 shows the hierarchical structure of technological resources discussed above.
A ‘task robot’ in the figure is an emulation of a human body with the same physical
abilities. It requires detailed instructions (or programming) in order to perform a simple
function such as cutting a metal bar. The middle part of the figure shows a higher level
resource, which includes a set of skills. In this case, it is unnecessary for the task to be
decomposed into detailed instructions. Skilled resources have the ability to process
information and prepare the detailed instructions for task robots from a set of high level
The bottom of the figure shows a higher level resource, which is a service provider such
as a company. A service provider internally consists of a hierarchical structure of lower
level resources such as labor, skills, machine power, and computing power. A task can
be assigned to these service providers at high levels of complexity, while we can be
reasonably certain of the outcome. These service providers internally do the information
processing and generate the set of lower-level instructions for the task. That is, they
internally decompose the task into detailed instructions for their subservient entities,
and allocate resources accordingly.
High Level
(less detail)
More abstract instructions:
More complex goals with less
specification of ‘how to’
Figure A3. 18: Hierarchy of service providers, the resource for high-level functions.
This discussion reveals that accomplishing a task, regardless of its size, involves the
processing of information and ultimately decomposing the task into a set of “primitive
instructions”. Thus, “working” is in fact “information processing”, and the amount of
work can be understood as the number of “primitive instructions” which need to be
executed. In summary:
A design is a plan or course of action for achieving a goal.
A design can be ultimately translated into a number of “primitive instructions”.
The “information content” of a design is in fact a “price tag” which is attached
to it. This tag indicates how many primitive instructions have to be executed in
order for the design to materialize.
Was this manual useful for you? yes no
Thank you for your participation!

* Your assessment is very important for improving the work of artificial intelligence, which forms the content of this project

Download PDF