Innovative Conceptual Design

Innovative Conceptual Design
Conceptual design, along with need identication and analysis, make

up the initial stage of the design process. Need analysis transforms the
often vague statement of a design task into a set of design require-
ments. Conceptual design encompasses the generation of concepts and
integration into system-level solutions, leading to a relatively detailed
design.
This book is devoted to the crucial initial stage of engineering
design. In particular, it focuses on parameter analysis, a systematic
yet exible methodology that leads the user through the design
process, helping to identify critical issues (parameters) of the design
and propose conguration-specic solutions to these issues. To
illustrate the principles discussed, the text presents numerous exam-
ples and a variety of real-world case studies. The emphasis through-
out is on innovation. The ideas developed by the authors encourage
the derivation of original solutions to new design problems or fun-
damental improvements to existing designs.
Innovative Conceptual Design will be a useful text for advanced
undergraduate and graduate students, as well as a handy reference for
practicing engineers, architects, and product development managers.
Ehud Kroll is on the faculty of the Mechanical Engineering Depart-

ment at Ort Braude College in Karmiel, Israel. He was previously a
professor at Texas A&M University and The University of Missouri
Columbia/Kansas City. His areas of interest include design theory
and methodology; design for manufacturing, assembly, and disas-
sembly; and automatic assembly planning.
Sridhar S. Condoor is a professor in the Department of Aerospace

and Mechanical Engineering at Saint Louis University. His primary
areas of interest are product design, design for manufacturability,
and computer-aided design.
David G. Jansson is the principal in David G. Jansson & Associates. He

was previously a professor and director of the Innovation Center at
the MIT and a professor, founder, and director of the Institute for
Innovation and Design in Engineering at Texas A&M University.
This page intentionally left blank
Innovative
Conceptual
Design
Theory and Application
of Parameter Analysis
Ehud Kroll
Ort Braude College
Sridhar S. Condoor
Saint Louis University
David G. Jansson
David G. Jansson & Associates
PUBLISHED BY CAMBRIDGE UNIVERSITY PRESS (VIRTUAL PUBLISHING)
FOR AND ON BEHALF OF THE PRESS SYNDICATE OF THE UNIVERSITY OF
CAMBRIDGE
The Pitt Building, Trumpington Street, Cambridge CB2 IRP
40 West 20th Street, New York, NY 10011-4211, USA
477 Williamstown Road, Port Melbourne, VIC 3207, Australia
http://www.cambridge.org
Cambridge University Press 2001

This edition Cambridge University Press (Virtual Publishing) 2003
First published in printed format 2001
A catalogue record for the original printed book is available

from the British Library and from the Library of Congress
Original ISBN 0 521 77091 2 hardback
Original ISBN 0 521 77848 4 paperback
ISBN 0 511 01928 9 virtual (netLibrary Edition)

To Our Families EK, SSC, DGJ
Contents
Preface page xi
1 Introduction 1
1.1 What Is Conceptual Design? 1
1.2 Parameter Analysis: A Conceptual Design
Methodology 3
1.3 Overview of the Engineering Design Process 4
1.4 Structure of the Book 7
2 Need Identication and Analysis 9
2.1 The Importance of Need Identication 9
2.2 Need Identication in Practice 14
2.3 The Need Analysis Methodology 19
2.4 Performance Considerations in Need Analysis 23
2.5 Value Considerations in Need Analysis 29
2.6 Size Considerations in Need Analysis 32
2.7 Safety Considerations in Need Analysis 32
2.8 Special Considerations in Need Analysis 33
2.9 Development of Design Requirements 34
2.10 Discussion and Summary 37
2.11 Thought Questions 38
2.12 Bibliography 39
3 Need Identication and Analysis Case Study:
Packing Factor of Sand in Electrical Fuses 41
3.1 Background 41
3.2 The Initial Need 42
vii
Contents
3.3 Need Identication 43

3.4 Need Analysis 44
3.5 Design Requirements 49
3.8 Bibliography 52
4 Introduction to Parameter Analysis 53
4.1 A Look at an Invention 53
4.2 The Nature of Conceptual Design 56
4.3 Theoretical Model of Conceptual Design 61
4.4 Parameter Analysis as a Conceptual
Design Methodology 64
4.7 Bibliography 70
5 Parameter Analysis Put to Work 71
5.1 Parameter Analysis Process Overview 71
5.2 Technology Identication 73
5.3 Parameter Identication 75
5.4 Creative Synthesis 86
5.5 Evaluation 90
5.6 Parameter Analysis Revisited 95
6 Conceptual Design Case Study: HVAC Airow Sensor 105
6.2 Abbreviated Need Analysis 106
6.3 Some Design Requirements 108
6.5 Parameter Analysis 111
7 Conceptual Design Case Study:
Cut-Edge Sensor for Flooring Removal 123
viii
Contents

Low-Cost Industrial Indexing Systems 137
8.5 Parameter Analysis of Conceptual Design I 143
8.6 Parameter Analysis of Conceptual Design II 147
Equal-Channel-Angular-Extrusion Metalworking 157
10 Need Analysis and Conceptual Design Case Study:
Ball Mover 184
10.2 Need Identication 186
10.3 Need Analysis 187
10.4 Design Requirements 192
10.6 Parameter Analysis of Conceptual Design I 194
ix
Contents
10.7 Parameter Analysis of Conceptual Design II 199

11 Technology Observation 207
11.1 Improving Design Abilities 207
11.2 Domestic Gas Water Heaters Example 208
12 Conclusion 216
12.1 The Essence of the Methodology 216
12.2 Cognitive Aspects of Parameter Analysis 217
12.3 Relation to Other Design Methodologies 220
12.4 Whats Next? 222
Index 227
x
Preface
Over the past several years, we have taught project-based courses on

product design to mechanical engineering students at several uni-
versities. We have used a variety of textbooks, but mainly our own
sets of teaching notes. At the same time, we have been doing research
in the area of design theory and methodology. Although most text-
books and design classes offer a decent coverage of the relevant top-
ics, their main weakness is in handling the conceptual design stage.
This is the most important phase of the design process, yet the
books fall short in providing the readers with a consistent and sys-
tematic methodology to apply for different design tasks. Our lecture
notes and research into parameter analysis as an approach to con-
ceptual design are the basis of this book.
Innovative Conceptual Design: Theory and Application of Parameter
Analysis is both an undergraduate- and graduate-level textbook for
design courses in engineering programs. Practicing engineers, man-
agers, and other product development professionals will also nd
the book valuable. Although our experience is with teaching the
material primarily to mechanical engineering students, the applica-
bility of the book is broad so that it may be used for teaching
advanced problem solving and design in various contexts, such as
other engineering disciplines and architecture.
The text focuses on the initial stages of the design process, namely,
need identication and analysis and conceptual design. We dene
need analysis as the process of transforming an often vague state-
xi
Preface
ment of a design task into a set of design requirements. In this trans-

formation process, the designer develops substantial insight into the
design task, in terms of both functionality and constraints. Under
conceptual design we encompass much more than other authors do.
We include the generation of concepts, the integration of these con-
cepts into system-level solutions, and the realization of the designs
as solid and concrete working congurations. Our conceptual
design phase ends with a relatively detailed design, which has been
shown to satisfy the requirements of the need. Our emphasis
throughout the text is on innovation. We encourage the derivation
of original solutions to design problems that have not been solved
before. If solutions already exist, we accentuate the need for sub-
stantial and fundamental improvements.
Most existing design books guide the reader to various search
and enumeration methods in order to generate a multitude of ideas
on how to solve the design task. These books refer the designer to
patent and literature surveys to identify potential solutions that may
be adapted to solve the task at hand. Because these approaches rely
heavily on previously generated solutions, they do not foster inno-
vation. Furthermore, they fail to provide the designer with a starting
point in a totally new design situation.
Most books also offer variations of brainstorming techniques
whose purpose is to organize the generation of ideas. Although these
techniques provide a useful starting point for the designer, they do
not help to transform an initial, rough idea into a good conceptual
design. An innovative design is a synergy of a number of good ideas,
not just a single good idea. Brainstorming techniques, however, do
not help in identifying ideas and integrating them into a cohesive
solution.
Because this book focuses on the parameter analysis methodol-
ogy to conceptual design, it eliminates the frequent sight of design
students and designers staring at a blank sheet of paper and desper-
ately trying to create a workable design. Parameter analysis is a sys-
tematic, yet exible, methodology that leads the user through the
design process. It helps the user identify the critical issues (parame-
ters) of the design and propose conguration-specic solutions to
these issues. Parameter analysis fosters innovation by requiring the
xii
Preface
designer to continually incorporate new ideas into the design. It is

also benecial as a tool for educating designers to think in a system-
atic and creative manner.
The content of the book is ideal for project-based courses on
engineering design. Such courses are already prevalent at the senior
and graduate levels, and the current trend is toward earlier teaching
of design methodologies. The engineering accreditation agency,
ABET, and other authoritative bodies have been advocating for some
time the integration of design across the entire engineering curricu-
lum. As a result, engineering departments are struggling to nd
appropriate teaching materials for design. The present volume uses
an approach to educating designers that is not limited to a specic
level of studies. We have taught parameter analysis, the main thrust
of the book, mainly at the senior and graduate levels; however, one of
us (S. C.) has taught the methodology to second-year students.
Faculty who teach design at a higher level than machine ele-
ments rarely adopt a textbook authors philosophy in its entirety.
Rather, they assemble their courses from articles, case studies, and
miscellaneous textbook chapters. This style is common in engineer-
ing schools within the eld of design. The perspective that guides
our book will enable faculty either to adopt it as the sole textbook
for a comprehensive and in-depth coverage of conceptual design or
to use it as a resource on conceptual design if the class also deals
with other topics. In both cases, the focus of our book on concep-
tual design will ll an existing void at both the undergraduate and
graduate levels.
The parameter analysis methodology was developed by Y. T. Li
and his colleagues at the Massachusetts Institute of Technology
(MIT), including David G. Jansson. Dr. Jansson and his colleagues at
MIT and later at Texas A&M University, including the other authors
of this book, carried out further developments and renements. In
particular, for a number of years, Dr. Jansson taught a graduate class
at MIT, called Invention, with Prof. A. Douglas Carmichael.
Parameter analysis was a primary part of that course. Dr. Jansson is
indebted to Dr. Carmichael for the many hours of discussion, inter-
change of ideas, renement of examples, and improvements in
teaching methodology that were the fruit of this partnership.
xiii
Preface
Dr. Condoor and Dr. Kroll have been teaching parameter analy-
sis at Texas A&M University, Saint Louis University, The University
of MissouriColumbia/Kansas City, Ort Braude College, and the
TechnionIsrael Institute of Technology. They are grateful to the
many undergraduate and graduate students, academic colleagues,
and professionals from industry and government who have con-
tributed both directly and indirectly to this book.
xiv
1
Introduction
Conceptual design is the thought process of generating and imple-

menting the fundamental ideas that characterize a product or sys-
tem. This process significantly affects the product novelty,
performance, robustness, development time, value, and cost. This
chapter presents the framework for the entire design process, within
which the upstream stages of studying the need and developing a
design concept t. It concludes with a brief road map to the book.
1.1 What Is Conceptual Design?
The core technical concepts developed during conceptual design fun-

damentally differentiate one product from other competitive prod-
ucts. The degree to which a new product is based on core concepts
that are dissimilar from those of existing products determines its level
of innovation. Innovative solutions may be viewed as designs that
incorporate novel concepts and exceptional functionality. Such solu-
tions can provide companies with a competitive advantage.
Core Technical Concepts of Computer Printers
Ink jet and laser printers serve the same function of producing an
image on paper. The ink jet printer transfers small droplets of ink
1
onto the paper, and these droplets form dots, which in turn
become the image. On the other hand, the laser printer uses a
laser source to trace the image on a drum. When the paper rolls
over the drum, the image is transferred onto it. Although both
products accomplish the function of applying ink to paper, their
core technical concepts are different. This difference results from
activities that took place during conceptual design.
To extend this example further, two major technologies
(thermal and piezoelectric) are currently in use to inject the ink
droplets in ink jet printers. The difference between the ink jet
technologies, however, is smaller than that between ink jet and
laser printers. Nevertheless, this difference is also the result of
the conceptual design stage.
The context in which engineering design activities occur is

much larger and more comprehensive than the conceptual design
stage. It includes the roles of marketing, nance, planning, and over-
all management. Furthermore, engineering design is carried out in
an integrated and concurrent manner, and requires effective com-
munication and cooperation among the team members. The coop-
eration of team members is not just cross-functional but may
include team members participating within a single functional area.
For example, the interaction of team members within the concep-
tual design stage itself is often quite advantageous as team members
build on or play off of the ideas of others. Note that even though
designs are often done by teams, in this book we frequently refer to
the creator of the product as the designer.
The task of conceiving of new concepts and their corresponding
physical congurations to meet the demands of a market need is not
a simple one. In addition to the traditional engineering and scien-
tic disciplines, this complex process involves human cognition, a
eld regarded as fundamentally outside of the engineers interest or
expertise. How do humans think? How do we create new ideas?
What are the cognitive processes involved in a successful conceptual
design activity? In spite of these questions and to the relief of most,
2
Introduction
if not all, of the readers, this book is indeed an engineering text, not
a textbook on human cognition.
1.2 Parameter Analysis: A Conceptual Design

Methodology
The primary focus of this book is the conceptual design process

itself. We present a methodological approach to conceptual design,
which we call parameter analysis. The development of parameter
analysis involved observation of numerous successful and unsuc-
cessful conceptual design processes, attempts to understand the rel-
evant thought processes, and identication of what occurred as
these new ideas were created and took shape. Thus, the formulation
of the parameter analysis methodology mirrors successful concep-
tual design activity.
The term methodology, rather than the term method, is used
purposefully to describe the parameter analysis approach. This may
be a somewhat subtle philosophical distinction, but as the authors
see it, the word method more commonly implies an orderly, step-
by-step, prescriptive process with a predictable outcome.
Methodology, on the other hand, is indicative of a process that is
based more on a set of general guiding principles than a series of
steps. Parameter analysis does indeed describe a set of principles
that are important in conceptual design and guides the designer in
developing an initial idea into a design. However, through extensive
experience in teaching this approach, the authors have discovered
that in the early stages of learning parameter analysis, the methodi-
cal use of a formal, step-by-step process is very helpful in commu-
nicating the principles involved. Thus, this textbook uses somewhat
of a forced march approach. When the principles are well under-
stood, the extra effort required by such an orderly process is not
always necessary.
Just as other textbooks may expand on certain elements of the
larger process, this book puts a magnifying glass onto the conceptual
design stage. In a sense, one can argue that conceptual design is the
most difcult or at least elusive part of the engineering design
3
process since it involves the creation of something new. (The

authors recognize that often the insightful recognition of a need is
the most crucial contribution to the success of a product, or that it
is a clever marketing approach or stunning product aesthetics which
will drive a product to the top.) At the risk of overgeneralizing, we
contend that parameter analysis is profoundly different from some
other approaches to conceptual design, which are more oriented to
listing what is already known or searching the state-of-the-art for
the best existing solution. Some of these approaches tend to cause
the designer to xate onto existing solutions, thereby hindering
innovation. As the reader will see in later chapters, the parameter
analysis methodology recognizes the nature of the task and focuses
on ways to unlock the unknown.
One further introductory comment about parameter analysis is
in order. As the book will show, a solid understanding of the fun-
damentals of the physical sciences and engineering disciplines is an
important component of parameter analysis. It is often true that
the core concept of a clever conceptual design emerges from a thor-
ough understanding of the fundamental physics involved. Thus,
there is no substitute for continuous learning of the basics of engi-
neering science. Learning to analyze existing products in a way that
uncovers the underlying physics of a conguration is a useful and
important by-product of the parameter analysis approach to con-
ceptual design. We call this process technology observation and dis-
cuss it in Chapter 11. The reader will recognize the linkage between
parameter analysis and technology observation in that technology
observation essentially represents an effort to discover and learn
from the core concepts behind the successful conceptual designs of
others.
1.3 Overview of the Engineering Design Process
Numerous terms are used to refer to the process or parts of the

process by which products are created, developed, and delivered to
the marketplace. Some of these terms are technological innovation,
product design, product development, product realization, inven-
4
Introduction
Need Identification
and Analysis
Conceptual
Design
Technology
Identification
Parameter
Analysis
Concept
Selection
Realization
Embodiment
Design
Detail
Design
Prototyping
Figure 1.1 Overview of the engineering design process.
tion, and, of course, engineering design. Practitioners and academ-

ics often differ on the denitions of these terms. In addition, there
are many different models of the engineering design process, but
they all include the following elements in some form or another:
A stage of identifying and analyzing a need prior to initiating

conceptual design.
A conceptual design stage to create new ideas that satisfy the
need.
Activities through which a concept is turned into an overall
product or system layout.
A stage of nalizing the design details.
A high-level view of the engineering design process is represented by

the owchart shown in Fig. 1.1.
5
The purpose of the upstream stage called need identication and

analysis is to set the stage for conceptual design. As we will see in this
text, the effectiveness of the conceptual design process depends in
part on how well a need in the marketplace is understood (ana-
lyzed). The two major functions of need identication and analysis
can be described in broad terms as:
1. Discovering what the real need is, and

2. Analyzing the need such that the best possible solutions to
meet its requirements are not precluded by the way the need
is understood or described.
In a sense, both the roles of need identication and analysis

relate to removing bias from the process. Identifying the real need
deals with recognizing and removing preconceptions of the need
denition. Furthermore, ensuring that the best possible solutions
are not inadvertently ruled out of the process is accomplished by
analyzing the need in a way in which the inuence of preconceived
solutions to the problem is objectively removed from the analysis.
These two inuences on the whole process are inherently quite sim-
ilar and suppress the natural tendencies of human problem solvers.
Therefore, the overt process of ensuring objectivity at this early stage
is an essential part of good engineering design practice.
The second stage in the design process is conceptual design.
The designer will always need a spark of ingenuity to come up
with innovative initial ideas. These ideas can come from the step
we call technology identification or from recognizing critical issues
during need analysis. One of the strengths of parameter analysis is
in providing a systematic methodology to develop an initial, rough
idea into a viable conceptual solution. Compared to other
approaches, we observe that conceptual designs developed
through parameter analysis are considerably more detailed and
firm. In the conceptual design stage, the designer typically devel-
ops three or more solutions to increase the likelihood of innova-
tion. These solutions are then evaluated against each other to
determine which one will be developed further. This stage is
known as concept selection.
6
Introduction
The broadly dened downstream design and development

activities identied as realization in the simple model of Fig. 1.1
include embodiment design, detail design, and prototyping. These
stages may include other functions, such as proof-of-concept testing
and design for manufacturing. Although this textbook does not
address these downstream design activities, it is vital to recognize
that the entire process is not complete until every element of the
process has been accomplished.
Notice that the model recognizes that the interfaces between
conceptual design and realization, and between need identication
and analysis and conceptual design, are bidirectional. That is, it is
possible that a conceptual design will be sent back to the drawing
board or rejected altogether if the results of the downstream efforts
reveal issues overlooked during the need analysis and conceptual
design stages. In other words, a bad conceptual design cannot
become a good design regardless of the quality of the realization
effort and may therefore call for additional work. This can cause
lengthy delays and increased product development costs, and will
eventually have an adverse effect on the success of the product.
1.4 Structure of the Book
This book is heavily dependent on examples to elucidate the

methodology presented, and the examples are carefully detailed in a
way that claries the thought processes undertaken. The student
reader is encouraged to study the examples and case studies fully
rather than merely read them through rapidly. Furthermore, the stu-
dent is encouraged to enhance his or her understanding of the
methodology by detailing each part of the thinking process in the
formal way that is presented in the book. As already mentioned,
when the parameter analysis methodology becomes more familiar,
the formalism of this approach will be less important. However, suc-
cessful, high-quality conceptual design comes about through prac-
tice and hard work, as well as inspiration. It is often argued that
clever, creative conceptual design skills are a gift, not something that
can be acquired by learning. However, the authors have seen rst-
7
hand the value of the coaching method called parameter analysis.

Just as an incredibly gifted athlete must practice (with coaching!) to
reach the highest level of competitiveness, so, too, can designers
benet from a methodology that sharpens their ability to think in
ways that foster creative solutions.
The authors have considerable experience in teaching parame-
ter analysis. This approach to conceptual design has been tested in
both graduate and undergraduate design subjects and in numerous
industry short courses and projects. The combination of theoreti-
cal discussion of the principles of parameter analysis and the use of
numerous examples and exercises has proved to be an effective way
to learn this material.
Chapter 2 introduces a methodology for identifying and analyz-
ing needs as the preparatory step to conceptual design. Chapter 3
presents a detailed example of need analysis to illustrate the for-
malisms of Chapter 2. While Chapter 4 introduces the parameter
analysis methodology, Chapter 5 offers an expanded discussion and
provides many guidelines for implementing the methodology.
Chapters 6 and 7 are case studies of applying parameter analysis to
relatively simple sensor design tasks. Development of alternative
conceptual solutions to meet an industrial need is demonstrated in
Chapter 8. Chapter 9 shows how parameter analysis was applied to
the design of a machine for a novel extrusion process. Chapter 10
contains a comprehensive need identication and analysis, together
with the subsequent conceptual design for entry in an engineering
students competition. Technology observation is presented in
Chapter 11 as an approach that will sharpen the skills of designers
by understanding the concepts behind the designs of others. The
book concludes with Chapter 12, which provides direction on how
to proceed downstream in the design process, and discusses some of
the cognitive aspects of parameter analysis.
8
2
Need Identification and Analysis
Every design process begins by identifying and analyzing the need,

and dening it in terms of the design requirements. Clarifying and
quantifying the design task helps the designer to gain crucial
insights that will facilitate the creation of innovative products. This
chapter provides a methodology for carrying out these initial steps.
The methodology rst calls for identifying the real need in qualita-
tive and solution-independent terms. Next, the overall need is stud-
ied in light of ve basic categories of functions and constraints:
performance, value, size, safety, and special. These functions and con-
straints should be fullled and satised by any design solution that
will emerge later in the design process, yet any premature specica-
tion of design solutions is carefully avoided during this stage. The
need analysis is summarized at the end of the process as a consistent
and quantitative set of design requirements.
2.1 The Importance of Need Identication
Design tasks arise from a variety of sources, sometimes referred to

as customers. Because these sources are often nontechnical, the
tasks may not be dened in engineering terms. Furthermore, initial
task statements sometimes describe perceived problems as opposed
to real needs. The natural tendency of novice designers is to start
thinking of solutions that satisfy the perceived problem before gain-
9
ing full understanding of what should really be designed. This ten-

dency, coupled with a lack of proper denition of the task, may mis-
lead the designer and waste his or her time and effort in solving the
wrong problem.
Successful identication and denition of a market need is a
prerequisite in developing innovative products. Many companies
actively identify new market needs and technologies through mar-
ket surveys, product benchmarking, technology observation, and
forecasting. These activities enable companies to identify immedi-
ate, interim, and long-term market needs, as well as current and
emerging technologies. Later, during the design process, the design-
ers will focus their efforts on establishing synergy between market
needs and existing and new technologies.
An incorrect need denition focuses attention on solution-spe-
cic issues, and so the designer will be xated on developing an
apparent solution into a product. In the process, many superior
solutions that t the overall design need better may be overlooked.
Competition can leapfrog by better dening the need and conceiv-
ing more innovative solutions that match the current technologies
with the real need. The importance of recognizing and understand-
ing market needs and matching new technologies with the changing
needs of society is illustrated by the following example of the evolu-
tion of typewriters.
Market Needs: Insights from the Evolution of Typewriters
The typewriter, a great American invention, shaped our society

and the mode of communication for over a century. The origin
of typewriters can be traced back to 1868, when C. L. Sholes
and several of his associates created the rst typewriter proto-
type. E. Remington and Sons began selling a rened version of
this machine in 1874. This typewriter was popularly known as
the blind writer because the paper was hidden from the typ-
ists view. The character set consisted of uppercase letters only,
but an important feature was the arrangement of letters on its
keyboard, known as QWERTY (the rst six letters at the left
10
Need Identication and Analysis
end of the top row). This arrangement, which helped to avoid

jamming of the keys at high speeds by separating the letters that
are often typed in succession into opposite sides of the key-
board, persisted over the years and is even used in present-day
computer keyboards.
To improve the usefulness of typewriters, the Smith Premier
Typewriter Company introduced the Smith Premier, which used
a full keyboard with separate keys for upper and lowercase let-
ters. Remington, on the other hand, came up with the innovative
Remington Model 2, which used the shift key to type both upper
and lowercase letters using a single character set on the key-
board and dual-faced typebars. This keyboard represented an
ergonomic improvement, reducing the mechanical movement of
the hand. Even though the demand for typewriters was virtually
nonexistent at the time, the few typewriter manufacturers rmly
believed in the great potential of the product.
Let us now step back and examine the possible needs that
typewriters could satisfy in the later part of the nineteenth cen-
tury. Typewriters could satisfy a need for clear and legible doc-
uments; however, typewritten documents were not considered
socially acceptable then, so this need did not exist. On the other
hand, with the increasing speed of communication at that time,
a need existed for high-speed transcription. Typewriters could
potentially satisfy this need, but very few people knew how to
type. There were neither books nor instructions to teach typing.
As a result, typing speeds were much lower than manual writ-
ing, and it did not make sense for a customer to buy a type-
writer when he or she could write faster. Although the inventors
believed that typewriters could help in high-speed transcription,
they did not understand the big picture in which the typewriter,
and the training to use it, together formed the complete solution.
The initial market failure of typewriters can be attributed to this
lack of recognizing the real market need.
The breakthrough for typewriters came in 1881, when the
Young Womens Christian Association (YWCA) thought of typing
as a career for women and started offering typing courses. The
business community immediately recognized the need for typists,
11
which also made way for women to enter the workforce in large
numbers. To cope with the demand for training courses, several
institutions opened across the country. The training, along with
the typewriters themselves, now fullled the need for high-speed
transcription. It sparked a great increase in the sales volume and
in the competition as well. But while the demand for typewriters
and typists was increasing, users still had to cope with the prob-
lem of blind writers. In 1895, recognizing the need for see-
ing the document while typing, the Underwood Company
introduced the revolutionary Underwood No. 5 typewriter. This
was the rst writing-in-sight or visible-writing typewriter.
The next innovation in typewriter design addressed the
need for portability and noise-free performance. Electric tech-
nology was incorporated into typewriters in 1920, creating the
electric typewriter. It had several advantages over the manual
typewriter. The electrical energy was used to actuate the keys
instead of nger pressure and to do the line spacing and car-
riage return functions. Consequently, the performancetyping
speed and quality of printwas greatly improved. Recognizing
these benets, all typewriter companies moved quickly toward
electric typewriters.
In 1934 Dvorak, one of the founding fathers of industrial
engineering, developed a new keyboard arrangement for type-
writers. The arrangement, shown in Fig. 2.1, was based on twelve
years of studying languages and the physiology of the hand. In
this keyboard, 70% of all English words can be typed using the
middle-row keys as opposed to 30% in the QWERTY arrange-
ment. Also, the most common consonants are placed on the right
side of the keyboard, whereas the most frequently used vowels are
on the left. As a result, the typist continuously uses both hands in
every word. This arrangement is easier to learn and use, improves
typing speed by about 10%, increases accuracy, and reduces
fatigue. Despite its tremendous advantages, marketing this key-
board by Smith-Corona failed. While designing the keyboard,
Dvorak overlooked an important implicit constraint: the entrench-
ment of the existing technology brought about by millions of peo-
ple being procient at using QWERTY keyboards. A change
12
" < > ? +

' , . P Y F G C R L / =
_
A O E U I D H T N S -
:
; Q J K X B M W V Z
Figure 2.1 Dvoraks keyboard arrangement which separates fre-

quently used consonants and vowels into the right and left sides of
the keyboard to improve typing efciency.
would require retraining people and buying new typewriters. This

constraint made the revolutionary keyboard impractical.
Even though the QWERTY arrangement reduced the ten-
dency of the printing arms to collide, all typewriters, both elec-
tric and manual, were jamming at higher speeds. Recognizing
this limitation of mechanical levers, IBM introduced a revolu-
tionary concept, dubbed Selectric, in 1961. This machine
used a ball with letters embossed on it. The ball was mounted
on a carriage, which continuously moved in the horizontal
direction during typing as opposed to moving the paper relative
to stationary keys. In addition to obviating the problem of jam-
ming the levers, the new concept reduced the overall size by
eliminating the carriage motion. It also reduced the moving
mass, which in turn improved typing speed. The ball could also
be changed easily to accommodate different languages and
fonts, or replaced when worn.
The electronic era found the typewriter companies trying to
cope with the new technology by introducing electronic typewrit-
ers. These devices allowed the typist to check and revise each line
before printing to eliminate mistakes and the need for corrections.
Electronic typewriters were further improved by incorporating a
monitor to allow checking whole documents. However, these
changes could not keep up with the consumers need and appre-
ciation of the greater exibility offered by personal computers.
Recognizing the coming end of the typewriter era in the
United States, the Brother Company began increasing sales in
13
foreign countries. By producing different-language typewriters,

they successfully exploited the residual value of the typewriter
technology while switching to the next-generation products. The
company diversied into printers, fax machines, word proces-
sors, and electronic stationery such as labeling systems and
stamp creators. The Smith-Corona Corporation, on the other
hand, was unable to make the leap from electrics to electronics
and led for bankruptcy on July 5, 1995. Interestingly, on the
same day, Bill Gates, the founder and chairman of Microsoft
Corp., was announced to be the richest person in the world.
2.2 Need Identication in Practice
Design tasks may originate from sources external or internal to the

organization. Externally, the market is the driving force for develop-
ing new products. Design tasks may also originate internally from
within the organization. For example, while designing a large system,
the system integration team may dene a subsystem as the design task
for another team. Design tasks often originate from the manufactur-
ing department as a need for improvement in the product manufac-
turability or a need for new jigs and xtures. Regardless of the source
of the task, the design process is always initiated by a task statement.
Many design task statements tend to be highly congurational. In
other words, they identify the required end product rather than
emphasize the need. Design a shaft,design an encoder, and design
a ship canal are some typical statements. They quickly evoke existing
congurations in our minds. Our natural inclination is to take these
general congurations and make them more concrete. However,
recalling existing congurations tends to xate the designer onto sim-
ilar and common solutions and to prevent the conception of highly
innovative and effective products.
Companies often make a similar mistake by dening their core
business in terms of products and not the need satised by those
products. As the products become obsolete, the companies, too, face
extinction, like most typewriter manufacturers. Successful companies
redirect themselves with changing technologies. The next example
14
illustrates how a new market need can emerge and emphasizes the
importance of viewing the core business in terms of customer needs.
Zero Emission Vehicles
Automobiles contribute well over half of the emissions that dam-

age the ozone layer and a signicant percentage of particulate
matter in the atmosphere. To address the problem of air pollution,
the California Air Resources Board adopted the Low Emission
and Clean Vehicle Regulation in 1990. According to this regula-
tion, 10% of the cars sold in the year 2003 should be Zero
Emission Vehicles (ZEVs). Other states and countries are follow-
ing this lead and adopting strict federal standards for vehicle
emissions. This regulation created a market need for ZEVs and
spurred an interest in the development of electric vehicles. Several
technological leaps have been made, including the development
of advanced batteries that do not use lead and the commercial-
ization of fuel cells and other alternative technologies.
This regulation had a profound impact on the oil industry.
Petrochemical companies are radically restructuring their core
business. Whereas earlier they viewed their core business as
supplying petrochemicals, now they redene it as meeting
energy needs. This new denition allows the companies to
expand into new markets, such as electric and hydrogen recharge
stations, and might be crucial for their long-term survival.
To suppress the tendency of being conguration- or product-ori-

ented, the designer must always begin by determining what the real
needthe benet sought by the customeris. The need for a shaft, for
example, can be dened as transmitting torque. Leading companies
spend considerable effort in dening their core business in terms of
customer needs. Kodak, for instance, denes its core business as imag-
ing, which includes digital photography. Such denition facilitates the
transition into new technologies in a systematic manner. The follow-
ing two examples expand on the advantages of a good need denition.
15
NASAs Mars Sample Return Mission
One possible initial task statement for NASAs Mars sample

return mission may be to design a space mission to Mars that
will deploy a vehicle for collecting rock samples at several loca-
tions and bring the rocks back to Earth. From this statement,
one can very easily jump into initiating a design process of a
spaceship that will travel to Mars, land on the surface, release
a ground vehicle for collecting samples, take off from Mars with
the samples, and travel back to Earth. However, the danger in
designing this mission according to the highly conguration-spe-
cic initial task statement is that more innovative and economi-
cal options may be ignored.
Scientists have recently identied a rock found in Antarctica
as a Martian rock blown into space as a result of a meteorite
impact on Mars. So perhaps we can send a powerful bomb to
explode on Mars, blow pieces of it into space, and wait for
some of them to arrive here. We can even attempt to capture the
rocks in Earth orbit by the space shuttle. Alternatively, we could
send a spaceship to Mars orbit, drop bombs and collect rock
fragments while in orbit, and travel back to Earth. Or, if we
insist on landing a sample-collecting device on Mars, perhaps it
does not necessarily have to be ground based. One Russian
proposal for a similar mission consisted of a balloon, which
inates and deates automatically due to the daynight temper-
ature differences on Mars and is carried around by Martian
winds. Another option could be to analyze the soil on Mars and
only send the results to Earth.
Such innovative solutions would have been considered had
the need been stated as to obtain Martian soil samples for
analysis or to analyze Martian soil samples. Thus, the way
the need is stated may affect the degree of innovation in the
design solutions.
16
Ship Canals
Let us examine another task statement, that of design a ship

canal to connect two bodies of water. The rst thought that this
statement brings to mind is digging a simple moat, which is a
possible solution when the water level of the two seas is the same.
An example is the Suez Canal, which connects the Mediterranean
with the Gulf of Suez at the head of the Red Sea, as shown in
Fig. 2.2a. Because it is a sea-level canal, the elevation of a ship
does not change during its journey. The Suez Canal remains the
worlds longest canal and can still be widened or deepened rel-
atively easily to accommodate bigger ships.
After building the Suez Canal, the French tried to build the
Panama Canal to connect the Caribbean Sea with the Pacic
Ocean. Their attempts to dig a sea-level canal lasted nine years
but ended in failure. This led the Americans to build the Panama
Canal with locks that raise and lower the waterway, as shown
in Fig. 2.2b. When a ship enters a lock in the canal, the gates
are closed behind it and the water level is raised or lowered,
Ship Ship
Mediterranean Red Sea
Sea
(a)
Locks Locks
Ship Lake
Atlantic Ocean Pacific Ocean
(b)
Car Railroad
Ship
(c)
Figure 2.2 Longitudinal proles of (a) the Suez Canal, (b) the
Panama Canal, and (c) canal on wheels (not to scale).
17
thereby changing the elevation of the ship. Then the next set of
locks ahead of the ship opens to allow the ship to proceed. In
this way, the locks eliminate the need to maintain constant water
level in the canal. Thus, a better denition of the original design
task might have been to connect the two bodies of water.
Note that the last statement is less congurational (no canal)
and more functional (connect).
If we investigate the need furtherby asking, do we really
want to connect the two bodies of water?then the answer
reveals that the real need is to transport ships across a land-
mass in a cost- and time-effective manner. The designers at
Ronquires in Belgium utilized this insight in constructing a sys-
tem for transporting ships across a 1-mile stretch of land with a
220-ft rise between Brussels and Charleroi. The gradient for this
system is much larger than the Panama Canals, which elevates
ships by 85 ft over about 2 miles. Because of the steep gradi-
ent, it was difcult to implement a canal system that uses locks
to raise the waterway. The designers conceived a novel system
that uses 300 40-ft water-lled railroad cars to transport ships
up and down the ve-degree slope as shown in Fig. 2.2c. Had
the designers followed the natural instinct of designing a ship
canal that connects two bodies of water with water, they would
have ignored innovative solutions such as the canal on
wheels. Therefore, before making any congurational commit-
ment, it is necessary to understand the need in terms of its func-
tional requirements.
The preceding examples and insights show that the real need, as
opposed to a perceived need, should be identied as a rst step in
design. In other words, what the designer is told to design is not nec-
essarily what the customer really needs, but rather what the cus-
tomer thinks that he or she needs. The designer should therefore use
the customers input to dene in qualitative terms what the primary
purpose or goal of the product is. This denition constitutes the
need statement.
18
Energy Energy
Materials Materials
Information Design Information
Generalized Artifact Generalized
Displacements Displacements
Generalized Generalized
Forces Forces
Figure 2.3 Black-box model of a design artifact with inputs and outputs.
To develop a good need statement, it is often useful to view the

product-to-be as a black box with inputs and outputs. The purpose
of the design artifact is to transform the inputs into desired outputs.
Typically, inputs and outputs for engineering products are con-
cerned with energy, material, and information. In mechanical engi-
neering, it is often convenient to view some inputs and outputs in
terms of generalized forces (including moments) and generalized
displacements (including rotations) rather than energy. Figure 2.3
shows the representation of a design artifact as a black box that
transforms inputs into outputs. The designer should identify all the
inputs and outputs and dene the transformation that the product
must perform. In some design tasks, the required output is clear, and
the designer is free to select the inputs. In such situations, the need
statement should reect only the outputs. Table 2.1 shows the need
statements for a few common devices, generated by applying the
above procedure. These statements in turn can lead to alternative
solutions and thereby foster innovation.
2.3 The Need Analysis Methodology
Once the real need has been identied, considerable effort should be
put into thoroughly understanding and precisely dening this need.
On an abstract level, we can think of the collection of all possible,
still-unknown solutions to the design task as forming an imaginary
solution space. As shown in Fig. 2.4, this space is enclosed by
19
Table 2.1 Need statements for some products based on the black-box
model.
Black Box Description
Alternative
Product Inputs Outputs Need Statement Solution
Slider-crank Linear Rotational Convert linear Rack and
linkage motion motion motion into pinion
rotational
motion.
Tachometer Rotational Speed Measure the Encoder +
motion information rotational speed Clock
of a shaft.
Rolling-element Rotational Rotational Transfer radial Magnetic
bearing motion; motion; and axial forces bearing
forces forces while allowing
relative motion.
boundaries that are the constraints of the task. During the need
analysis stage, the designer outlines the solution space with precise
and quantitative boundaries. Although this process can easily be
done by considering a specic solution, the result will be biased and
narrow in scope, eliminating many potentially good designs and
hampering innovation. The objective during need analysis is to
maximize the size of the solution space, so the designer must refrain
from inventing a solution while dening the boundaries. Indeed,
thinking in solution-independent and generic terms is one of the
most challenging aspects of the entire design process. The benets
of such abstract thinking include a larger solution space, better
potential for innovation, and greater exibility.
Need analysis is concerned with two main aspects of the design:
functions and constraints. The purpose, or main function, of the
product to be designed has already been recognized during the need
identication stage. Now, however, this function needs to be further
studied, rened, quantied, and perhaps broken down into subfunc-
tions. The brief and qualitative task statement will be turned into
more elaborate and quantitative descriptions of what particular
characteristics the product is required to have. A complex design may
require division of the overall task into smaller, more manageable
subsystems that can be assigned to several designers or design teams.
20
Solution Space
Boundaries
(Constraints)
Figure 2.4 Schematic of a solution space bounded by constraints. The task

of need analysis is to maximize the size of the solution space.
In an ideal situation, the designer is concerned with satisfying

functions only, so the solution space is innitely large or unbounded.
A high degree of innovation is possible because any solution is accept-
able. In the real world, however, the solution space is bounded by con-
straints (see Fig. 2.4). The constraints limit the ways in which functions
can be realized and thus reduce the size of the solution space. For
example, the need to archive ideas can be satised by several solu-
tions such as using pens, pencils, audio and video recorders, and com-
puters. Posing a constraint, for instance, adding the phrase on paper
to the need, eliminates several potential solutions. Adding the word
erasable as another constraint crosses out most solutions other than
pencils. In this way, each additional constraint can drastically reduce
the solution space, even to the point where no solution is possible.
Market surveys provide a large amount of information about
customer wishes and preferences. Before turning these preferences
into constraints, the designer should evaluate each item to check
whether it is real or ctitious. A real constraint is one for which its
violation renders the product nonprotable or nonviable. The list of
constraints should include all the real constraints because these
eliminate impractical solutions and focus the design effort on feasi-
ble solutions only. Fictitious constraints, on the other hand, should
be ignored to increase the solution space and foster innovation.
21
Constraints can be generated by examining the original design

task statement, relevant regulations, and previous design require-
ments for similar tasks. They can be classied into explicit and
implicit constraints. Explicit constraints are those listed in the initial
task statement or directly derived from it and are therefore easy to
identify. Implicit constraints, on the other hand, are those that must
be generated during the need analysis stage and are much more
important and difcult to recognize. Explicit constraints are typi-
cally listed rst, and all the product designers are made aware of
them to eliminate the possibility of creating a product that violates
these constraints. Any changes to the explicit constraints require the
customers approval. In contrast, implicit constraints are found by
studying the life-cycle environment and doing order-of-magnitude
calculations. If an implicit constraint, generated during need analy-
sis, is found to cause a problem later in the design process, the
designers may reexamine and revise it.
Need analysis is a discovery process wherein designers nd criti-
cal issues and gain crucial insights into the design task. To facilitate
the discovery process, need analysis is divided into ve general cate-
gories as shown in Fig. 2.5. These categories allow the designer to
focus on a few issues at a time while continuously acquiring more
understanding of the task. Studying trade-offs, such as performance
versus value, is a very important activity that may even lead to chang-
ing the original need denition. In trade-off studies, the designer
identies potential inconsistencies and tries to resolve them. This
adds to the designers understanding of the task, prevents conicts
between designers working on different aspects of the design in later
stages of the design process, and assures that the nal result would
have a good balance between the often conicting requirements.
Need analysis is not necessarily a structured and sequential
process, and designers often backtrack, both within the need analy-
sis process, and from later in the design process back to need analy-
sis. The designer may gain new and important insights during the
conceptual design stage, which would force a revision of the need
analysis and the design requirements. However, although designers
should regularly examine and update the design requirements, the
goal of need analysis is to minimize such time-consuming itera-
tions. This can be achieved by investing the adequate effort in a good
22
1. Performance
2. Value
3. Size
4. Safety
5. Special
Figure 2.5 The ve general categories of functions and constraints.
need analysis and considering downstream design issues before

starting with conceptual design. Keeping the number of modica-
tions to the design requirements to a minimum usually improves
the overall time-to-market of a new product.
The recommended practice for doing need analysis is rst to list
all the questions and issues that should be considered, preferably
classied under the ve general categories of Fig. 2.5. Next, the nec-
essary research and investigations are carried out, and the results
used to answer the questions and ll the relevant issues with infor-
mation. The next ve sections discuss the general categories in more
detail, followed by a description of how to turn the need analysis
into design requirements.
2.4 Performance Considerations in Need Analysis
The primary concern when analyzing a need is to identify every-

thing that the product must do and sometimes also how it should
be done. Performance deals with functions, which are the actions
or activities that a product should perform. An important charac-
teristic of functions is that they can be decomposed into smaller or
lower-level functions. This decomposition helps to break the over-
all task into manageable chunks, allowing several design teams to
work independently on different functions and thereby greatly
reducing the time-to-market. However, a bad or incorrect func-
tional decomposition may deprive the teams of their independent
23
decision-making ability. In such a situation, the teams are faced

with conflicts on a regular basis and need to repeatedly negotiate
with each other. This drastically increases the product develop-
ment time. In large design tasks, the common practice is to have a
system integration team that is in charge of high-level functional
decomposition, interface definition, and conflict resolution. The
following example illustrates some of the complexities faced by
integration teams.
Undersea Remotely Operated Vehicles
Remotely operated vehicles (ROVs) are used extensively in

deep-water applications in which maintaining a human diver for
any length of time is cost-prohibitive. A typical ROV (schematic
shown in Fig. 2.6) consists of a robotic arm that performs the
desired operations and a vehicle that maneuvers the arm. The
vehicle is controlled and powered from the surface through a
tether. The design of the tether is an interesting challenge
because the teams working on other subsystems raise conict-
ing demands on the tether designers.
Mother ship
Drag force
Tether
ROV
Seabed
Figure 2.6 Remotely operated vehicle system.
24
The tether is the lifeline of the ROV: it transfers energy for

navigation and control signals to the vehicle, and it transmits
visual information to the mother ship. This information and
energy transfer is crucial for successful operation. The tether is
neutrally buoyant to reduce the gravitational forces on the ROV.
However, the drag force acting on the tether due to ocean cur-
rents could be signicant, requiring the ROV to continuously
resist it. As a result, the tether is not only a lifeline, but also a
burden on the ROVs limited resources. These considerations
result in three functions in the tether design task: transfer energy,
transfer information, and reduce drag force.
To satisfy these functions, the tether design team can modify
several key variables, for example, the cable diameter and length.
A short cable with a large diameter is preferred for effective
power transmission because it reduces power losses and voltage
drop. Such a cable also increases the communication bandwidth
and rate. The initial task requirements usually establish a lower
bound for the length, leaving the designer with more freedom to
change the diameter. However, an increase in diameter affects the
drag force, which in turn increases the power requirement.
Different subsystem teams, such as communication, power,
and navigation, often place unreasonable demands on tether
designers to increase the rates of energy and information transfer
while reducing the drag force. It is the duty of the system integra-
tion team to decide on the acceptable magnitude of the energy
transfer rate, information transfer rate, and drag force in the initial
stages of the design process. The skill of the integration team lies
in anticipating potential conicts ahead of time and in nding suit-
able trade-offs that optimize the overall performance of the ROV.
Functions can be classied into primary and secondary. Primary

functions are those functions that must be satised to fulll the
need. Therefore, all design solutions must address the primary func-
tions. In contrast, secondary functions are not directly related to the
need and tend to be solution-specic. Typical issues addressed by
secondary functions are product inefciency and protection from
25
undesirable inputs. The following gearbox example elaborates on

the differences between primary and secondary functions.
The primary function of any gearbox is to provide a constant
speed ratio between input and output. Any solution, such as spur or
helical gear pairs, must therefore satisfy this function. Helical gears,
however, always generate a thrust load. If a successful implementa-
tion using helical gears is sought, then the designer must address the
additional functionresist the thrust loads. This function is solu-
tion-specic because another solution, using spur gears, does not
require it. Such functions are known as secondary functions.
Dissipate the heat generated in the bearings is another secondary
function because it does not directly involve fulllment of the over-
all need.
Designers take an active approach when addressing functions:
they incorporate certain features, components, or assemblies to han-
dle each function. As the number of functions increases, both the
magnitude of the required design effort and the product cost also
increase. The designer must reduce the number of functions to pos-
itively affect these two issues. Designers have very little leverage
when it comes to primary functions because these are required to
satisfy the overall need. Secondary functions, however, can often be
eliminated. Because secondary functions often address issues such
as robustness, the designer must trade cost against the product
improvements related to secondary functions. The designer should
therefore spend time at the beginning and decide what functions are
more important. The following example of the difference between
oppy disks illustrates the importance of identifying the signicant
secondary functions.
Computer Floppy Disks
Let us examine the need satised by oppy disks. At an abstract

level, both 5.25 and 3.5 disks are aimed at fullling the same
primary function: to store information. Both disks satisfy this need
using the same concept. However, a drastic usage shift away
from 5.25 disks and toward 3.5 disks took place in the early
26
1990s for several reasons. These reasons will help us to under-

stand the importance of identifying the correct set of functions.
From a cursory examination, it is obvious that a 3.5 disk
stores signicantly more information in less space. The 3.5
disks beneted from advances in technology and exhibit supe-
rior portability. When 5.25 disks were introduced, their infor-
mation storage density corresponded with the state-of-the-art
technology of that time. Consequently, these disks were rather
large, and portability became a major concern. A detailed
examination reveals that 3.5 disks satisfy additional functions;
for example, they can be slipped into a shirt pocket without wor-
rying about possible damage to the information. In other words,
3.5 disks address the additional function provide information
safety, and this function is realized in the form of a rigid cover
and a protective slide shutter. The protective shutter prevents
accidental damage to the magnetic medium when not in use.
The 5.25 disk is perfectly square (as shown in Fig. 2.7),
and as a result, it can be inserted into the drive in eight differ-
ent ways. This is a minor nuisance for the user whose attention
is required during disk insertion. In contrast, 3.5 disks (also
shown in Fig. 2.7) are rectangular, measuring 3.5 3.7, and
the slot in the disk drive measures 3.6. Because the disk can-
not be inserted along the 3.7 side, four possible orientations
are eliminated. Two more erroneous orientations are taken
Routine Design
HD
3.7"
5.25"
Innovative
Conceptual
Design
3.5"
5.25"
Figure 2.7 Shape and dimensions of 5.25 and 3.5 oppy disks.
27
away by the tactile feel provided by the metal shutter, and the
nal position for correct insertion is determined by the chamfer
at the right-hand corner. This design therefore addresses the sec-
ondary function: eliminate wrong insertion of the disk.
The designers of 5.25 disks addressed the function of
information storage and paid very little attention to secondary
functions. Although the 3.5 disk designers could benet from
their predecessors mistakes, a good need analysis could addi-
tionally have revealed these important secondary functions to
the 5.25 disk designers. These designers could have created a
more robust disk easily, although it would probably have been
more expensive. The designer of a successful product must
therefore identify a good set of functions from the numerous
demands and wishes of the customer.
The product development time may be long when new technol-

ogy is involved. In such cases, it becomes very important to accu-
rately estimate the future performance requirements. Companies
often perform detailed technology forecasting and market studies to
identify trends. For instance, the oil industry found that the trend in
fuel usage has been to reduce the carbon-to-hydrogen ratio. Fuel
cells, which combine hydrogen with oxygen to produce electricity
and water, use no carbon and therefore will have the lowest rating
on this scale. This trend helps companies to determine in which
technologies to invest. For example, DaimlerChrysler planned to
invest about $1.4 billion in ve years before the release of its rst
commercial fuel cell-powered car in 2004.
In the semiconductor industry, Moores law predicts the
increase in the number of transistors in a chip: the number of tran-
sistors in a unit volume doubles every eighteen months. The law is
based on observations by Gordon Moore, cofounder of Intel
Corporation, and it helps predict the desired performance for future
computers and electronic products.
Although performance considerations involve mainly functions,
performance-related constraints should also be recognized and pre-
cisely dened. One of the most relevant types of constraint is time-
28
related. Temporal constraints include the relative sequencing of dif-

ferent events and transfer rates. For example, the required time for
deploying an automotive airbag is very short, of the order of one-twen-
tieth of a second. This constraint requires that an explosive release of
gases be used. The required deployment time for a side-impact airbag
is even shorter, making the design task very difcult. Typically, tem-
poral constraints can be dened by recognizing which events have to
be performed before others and organizing them in a owchart.
Real-life designs are rst and foremost a compromise among
various performance objectives. The disk brakes mounted on the
wheels of a passenger airliner, for example, do not contribute at all
to in-flight performance such as payload, speed, and range.
However, maximum braking performance is expected on the very
rare occasion of an aborted takeoff: a fully loaded plane accelerating
on the runway when the pilot decides to stop because of an engine
problem or an obstacle ahead. Under these severe conditions, apply-
ing the brakes to stop the aircraft would cause so much kinetic
energy to be dissipated as heat that the brakes would practically
melt. However, this design is very efcient: the plane does not have
to carry the extra weight of more durable brakes at the expense of
decreased payload, and the brakes can be replaced after an aborted
takeoff, which happens very rarely.
Several calculations are often done to identify possible conicts
between the extent to which various functions can be satised. These
trade-off studies are important because they can provide an order-of-
magnitude sense of the difculties in realizing the design together with
signicant insights into the task. The calculations may begin with
energy and mass balances, and require making several approximations.
These calculations should provide information about the characteris-
tics of ideal solutions and the ideal performance values. Then, the
designer can strive to incorporate these characteristics into the design
solution and achieve the ideal performance during the design process.
2.5 Value Considerations in Need Analysis
Addressing the issue of value means much more than just cost
because sometimes value cannot be directly measured in monetary
29
terms. For example, when the design frees humans from dangerous
or repetitive jobs, it makes a social contribution. A space exploration
mission may make scientically invaluable contributions, and an
improvement to automotive emission control systems may have
environmental value. Even the Eiffel Tower has value as a great engi-
neering marvel, although it may not have much monetary value.
Although cost is usually associated with a specic solution, an
analysis of the value of the new design can be done before its con-
ception. Value is strongly related to the original need, the function
of the product, and the level of performance attempted. There is
usually a direct mapping between value and performance, as shown
schematically in Fig. 2.8. A low level of performance is associated
with low value, and high value is perceived as commanding superior
performance. Consider an expensive luxury car that costs much
more than what one would pay for a more basic vehicle. The luxury
car probably represents good value to some customers, who seek the
high performance levels reected in such factors as speed, accelera-
tion, handling, safety, comfort, reliability, and more.
In addition to requiring that the design provides certain valu-
able functions, value-related constraints should also be investi-
gated. Marketing is a key element in the overall product development
strategy. To better meet the needs of customers, markets are seg-
mented based on economic and social considerations. The target
market segment dictates the product cost and the anticipated sales
volume. Assuming that the market is rational (which may not always
be true), increasing the price decreases the sales and vice versa.
During need analysis, attention should be paid to the target product
cost and the sales volume.
From the manufacturers perspective, capital cost, payback
period, and return-on-investment are some of the key constraints
related to value. The customer, on the other hand, is interested in a
low product cost and a high product value. Often, the interests of
customers and manufacturers are in conict with one another. An
attempt should be made to strike a balance between these compet-
ing demands.
The price of a product refers to the amount that a customer
would pay to buy it. By investing in this purchase, the customer
either generates revenue or saves expenditure. This monetary bene-
30
Value
Performance
Figure 2.8 Schematic of valueperformance correlation.
t to the buyer is the value of the product. A successful product

should exhibit a good match between price and value; that is, cus-
tomers would usually be willing to pay more to get better value (and
thereby performance).
The calculations to determine the performancevalue relation-
ships are very important. For instance, in designing time-saving
machines, these calculations may redene the task and impose new
constraints on the product cost.
Making a prot is often the key motivation for developing a new
product. Prot depends on the product manufacturing cost, price,
and production volume, all of which cannot be varied independ-
ently. Increasing the production volume may lead to adopting mass
production techniques that drastically reduce the manufacturing
cost and price. In most markets, a lower price can help in selling the
product in larger quantities. The designer should decide on the
desired manufacturing cost and production volume based on input
from the nance and marketing departments. For a manufacturing
rm that is starting a new product line, it is important to under-
stand the investment needed to establish the required equipment
and facilities. To this end, the rm may borrow money, and the prof-
its generated should satisfy the payback period and return-on-
investment criteria.
The targeted manufacturing cost and production volume
should be established during need analysis because they will consid-
erably affect several downstream design decisions. This kind of
trade-off study was one of the critical issues in Ferdinand Porsches
31
historic proposal of the Volkswagen Beetle. He suggested the low

target price of 900 DM ($215) for the Beetle, based on a very high
production volume. Interestingly, with more than 21 million Beetles
made, it remains the worlds most produced car.
2.6 Size Considerations in Need Analysis
Size is used here as a general term to denote anything that has to

do with constraints on dimensions, shape, weight, and other related
physical properties. It is in this area that designers often have the
most preconceived ideas, or bias, about the solution. However, any
overconstraining of future solutions should obviously be avoided
during need analysis.
Indeed, the natural evolution of products is such that they
become lighter and more compact. In some industries, such as aero-
space and electronics, product success hinges heavily on how well the
designer handles the size and weight constraints. Designers are there-
fore under constant pressure to work within tight size constraints.
These constraints in turn may eliminate certain material and geo-
metrical choices, thereby limiting the solution space unnecessarily.
Most products must also interface with users, other equipment,
and the environment. The term environment is used here in a
broad sense to include both operating and manufacturing condi-
tions. The interface constraints establish the required compatibility
between the product and its surroundings. For instance, ergonomic
constraints dictate the maximum force that can be applied by a
human user. Interface constraints are often nonnegotiable because
they originate from sources external to the product being designed.
2.7 Safety Considerations in Need Analysis
Safety concerns dominate almost every aspect of modern design.

Accidental deaths are more common in industrialized nations than
in developing or rural countries. This can be directly attributed to
the greater usage of machines. With the increasing emphasis on
safety, health, and the environment, new products must address not
32
only functionality, but also safety. Safety constraints often require

designing redundant and fail-safe systems.
As an example, let us look at some safety aspects in designing a car.
In the case of collision, primary restraint systems (seat belts and head-
rests) and secondary restraint systems (airbags) are designed to hold
the passengers in place. Many other components in the passenger com-
partment are designed so as not to inict injury to the occupants. For
example, the steering column is made collapsible. Safety aspects are
also incorporated in the structural design of the car body, which needs
to be strong enough to sustain rollover, and at the same time provide
collapsible, energy-absorbing sections around the passenger compart-
ment. Emission control measures and sound-attenuating exhaust sys-
tems are required to protect the health of the population at large.
2.8 Special Considerations in Need Analysis
This category is concerned with market conditions and trends, gov-

ernment regulations, and other, special, or unusual task-related cir-
cumstances, even including issues such as international political
uncertainties. Government regulations play a crucial role in many
product developments. Documents pertaining to existing regula-
tions in the specic area of product design should be consulted dur-
ing need analysis and before generating the requirements. The
regulations may also specify standard tests that the product must
pass before it can be released to the market. Testing standards are
often more demanding than the actual service conditions and,
therefore, may pose additional constraints. The following example
illustrates how regulation can play a dominant role in design.
Power Source for Space Exploration
In many space missions, the technically preferred choice of

power source for a spacecraft is a Radioisotope Thermoelectric
Generator (RTG), a device that converts heat energy generated
by decay of a radioisotope into direct current electricity. RTGs
were used in several space missions including Apollo and
33
Galileo. However, since the 1986 Challenger accident, it has

been difcult to get approval for launching a radioactive device
from Earth, even though the probability of explosion in Earths
atmosphere is relatively small and the possible consequences of
such an explosion are quite mild.
If the constraint no launching of radioactive materials is
recognized, the design team will focus on developing alternative
solutions that use other energy sources. If this constraint is not
identied early in the need analysis, much effort can be wasted
on designing a whole space mission around the wrong power
source solution. This may lead to more conicts and expensive
delays. For example, NASA had to receive White House
approval for the October 1997 launch (with an Earth yby two
years later) of the RTG-equipped Cassini mission to Saturn.
2.9 Development of Design Requirements
Design requirements are a list of quantitative criteria that any solu-

tion must satisfy. They summarize the need analysis and constitute
the input to the next stage, conceptual design. In some industries,
the requirements are referred to as design specications. However,
it is important to note that design specications or design require-
ments are different from product specications. Product specica-
tions are written at the end of the design process in order to
summarize the performance and properties of the actual product.
For example, the product specications may include the actual
weight, size, lowest and highest operating temperatures, and so on.
In contrast, design specications or requirements stipulate the min-
imum required performance and other characteristics. The require-
ments are written at the beginning of the design process and are
used for evaluating the goodness of different solutions. For exam-
ple, a design requirement may specify the maximum allowable
weight for the design.
The requirements are a compilation of both functions and con-
straints in the form of short and quantitative statements. They tell
34
what the system or artifact must do and how well it must be done.
The requirements should be quantied as precisely as possible to
avoid ambiguity, misunderstanding, and potential conicts later in
the design process. The magnitude of functional measures and con-
straints needs to be established. Obtaining the information and per-
forming the necessary calculations are often difficult and
time-consuming tasks; however, these will turn out to be useful
throughout the design process. The design requirements should be
consistent: there should be no conicts between requirements, and
any potential conict must be resolved during the study of trade-offs.
The design requirements should be broad in scope to allow the
consideration of innovative solutions. Sometimes, however, the set
of design requirements is narrower and more focused than the need
analysis. Consider the example of designing a machine for pre-recy-
cling of aluminum beverage cans. In order to reduce the cost of stor-
age and transportation of the cans, it is desirable to process them so
that they take up less space, before shipping them to a recycling
facility. The need analysis should be very broad so as not to preclude
any possibility. Without actually creating a solution, a designer may
look into the economical aspects of various processing technologies
such as crushing, shredding/grinding, and melting. Regardless of the
specic implementation, the energy required to process a single can
by each technology may be estimated and compared to the cost sav-
ings in storage and transportation. Suppose it is concluded that only
crushing is a viable option. The requirements developed may reect
a decision to design a crushing-type machine and may disregard the
other processing techniques altogether.
The requirements need to be precise in denition and veriable,
so that they can be used later to help evaluate the design and select
among alternatives. The engine cooling system should provide suf-
ficient cooling to prevent overheating is a bad requirement.
Sufcient is very vague in denition, it does not help in sizing the
cooling system during design, and it does not allow quantitative
comparison of alternative designs. On the other hand, the statement
The cooling system should provide a cooling capacity of 10 kW is
both precise and veriable. Good requirements guide the design
process and affect the quality of its outcome. Eventually, it is the
35
design requirements that drive the cost and schedules of the design
process, the skills and resources needed, and the verication and
operational procedures used during and after design.
The designer begins developing the requirements by identifying
the variables that dene each function and constraint. For example,
the aluminum can processing machine may require a denition of its
capacity by the number of cans processed per unit time and a char-
acterization of its output by density (i.e., the weight of aluminum per
unit volume). Next, the variables are quantied by setting target val-
ues for the design, using the information obtained in the need analy-
sis. Some variables may be assigned a single value, whereas others
would have a range of acceptable values. Single values may be either
maximum or minimum. To continue with the previous example, we
can require that the maximum weight of the can recycling machine
is 1000 N or that the minimum processing rate is 2000 cans/hour. A
range of variables for this design may be assigned to the height of the
loading opening, say between 1 and 1.5 meters above the oor.
Once the design requirements are formulated and quantied,
they should be organized into various categories. The categories
typically correspond to those of Fig. 2.5 (i.e., performance, value,
size, safety, and special). Within each category, there may be one or
more requirements. This organization is useful and convenient dur-
ing design because it helps focus on one aspect at a time.
The design requirements list is a binding document to which
every design team must conform. The requirements may be modi-
ed in the later stages of the design process; however, this would
require approval from all the parties involved in the product devel-
opment and may pose complications. An effort should therefore be
made at the front end, that is, during need analysis, to consider all
the downstream issues and compile the best possible set of require-
ments. On the other hand, requirements should be challenged by
everyone involved in the design process, as more knowledge and
understanding of the problem domain is gained. For example, con-
sider the following scenario regarding the Mars sample return mis-
sion described earlier. The system integration group may analyze the
initial need by studying old photographs of the Martian surface
taken by several spacecrafts in the 1970s, and may generate a
36
requirement for a Martian ground vehicle that is capable of travers-

ing obstacles as big as one meter across. This requirement is dictated
to the vehicle design team, which concludes that the resulting vehi-
cle would be prohibitively large. The vehicle designers, therefore,
challenge the original requirement by negotiating with the integra-
tion team, which subsequently modies the requirement to state
that the vehicle should be able to traverse obstacles that are 0.3 m in
size, and handle larger obstacles by detecting and avoiding them.
2.10 Discussion and Summary
Need identication and analysis answer the following questions:
What is the problem Im trying to solve?

What particular characteristics is the new product required to
have?
How will I know that Im done?
Need identication is the step where a possibly nontechnical

and sometimes congurational design task is converted to a real
need. This need is extensively studied during need analysis to iden-
tify both functions and constraints in ve general categories: per-
formance, value, size, safety, and special. Throughout need analysis,
the designer should be objective and careful to avoid creating solu-
tions or developing a bias through false constraints. This will ensure
maximizing the size of the solution space.
Analyzing the need also deepens the designers understanding of
the task and its domain. This will be extremely helpful later in the
design process and is facilitated by being very quantitative and spe-
cic. The knowledge gained during need analysis, together with its
summary as design requirements, is carried over to the conceptual
design stage. Here, satisfying all the requirements serves as the crite-
rion for completion of conceptual design. Moreover, a good set of
design requirements will provide measures for the goodness of the
design, thus facilitating conception of an optimal solution, not just
a solution.
37
The approach presented in this chapter is a methodology for

need identification and analysis. It is not a step-by-step procedure
as are the following two alternative approaches. Quality Function
Deployment (QFD) has been touted in recent years as a struc-
tured method of developing design requirements. It leads the
designer through several steps, including listing of customers
requirements and benchmarking the competition. Alternatively,
some textbooks provide comprehensive checklists of generic areas
and issues that should be considered, and they ask the designer to
choose the relevant points and fill-in the blanks. Both of these
approaches may be used to supplement the methodology of this
chapter.
2.11 Thought Questions
1. Several design congurations may be used to bridge across a

body of water, such as the English Channel or the Hudson River.
Write a general need statement for this task, and examine how
the need is met by bridges, causeways, tunnels, and ferries.
2. What is the real need satised by videocassette recorders (VCRs)?

Identify competing technologies and describe the advantages and
disadvantages of each one.
3. Write a need statement and list the inputs and outputs for the
black-box model of the following products:
(a) Clothes washer
(b) Computer mouse
(c) Voltage transformer
(d) Gearbox
(e) Automobile brakes
(f) Automobile tires
4. The evolution of tires involved several signicant changes:

(a) From bare metal wheels to solid rubber tires
(b) From solid rubber to pneumatic tires with inner tubes
38
(c) From tubed tires to tubeless bias-ply tires

(d) From bias-ply tires to bias-ply belted and then to
radial-ply tires
Investigate the evolution of tires to determine the needs satised

by the different congurations and the performance trade-off
issues in their design. What are the performance requirements
for an ideal automobile tire?
5. Identify the trends in the laptop computer market with the goal
of developing a new laptop computer in the next year. Using
these trends, determine a target weight, size, screen size, and res-
olution for the new laptop.
6. Identify the important safety considerations in the design of a

food processor.
2.12 Bibliography
Quality Function Deployment is widely described in the literature.

A popular paper is:
Hauser, J. R. and Clausing, D. The House of Quality. Harvard
Business Review, 66, No. 3 (1988): 6373.
A more extensive coverage of QFD can be found in:

Clausing, D. Total Quality Development. New York: ASME Press,
1994.
The following two textbooks use QFD as a tool for understanding

the design task and developing the requirements (dubbed engi-
neering specications):
Dixon, J. R. and Poli, C. Engineering Design and Design for
Manufacturing: A Structured Approach. Conway, MA: Field
Stone Publishers, 1995.
Ullman, D. G. The Mechanical Design Process. 2nd ed. New York:
McGraw-Hill, 1997.
39
The use of checklists is demonstrated in:

Pahl, G. and Beitz, W. Engineering Design: A Systematic Approach.
London: The Design Council, 1988.
40
3
Need Identification and Analysis
Case Study: Packing Factor of
Sand in Electrical Fuses
In sand-lled electrical fuses, the packing factor of sand controls the

overall fuse performance. This chapter discusses an ill-dened, per-
ceived need involving the packing factor. The methodology outlined
in the previous chapter is applied to dene the real need by establish-
ing the scope of the design task at the proper level. The need is then
analyzed in order to fully understand and quantify the key issues. The
results of the need analysis are summarized as a set of design require-
ments. The case study demonstrates a denitive link between good
need identication and analysis and the potential for innovation.
3.1 Background
Sand-lled fuses are used to protect electric mains and feeders, cir-
cuit breakers, heating and lighting circuits, motors, transformers,
semiconductors, and more, against current surges. They are charac-
terized by their interrupt rating, which is the maximum current that
a fuse can stop while maintaining its mechanical integrity. Fuses that
have high interrupt rating are lled with sand. A schematic of a
sand-lled fuse is shown in Fig. 3.1. The components are:
Fuse element: made of silver with one or more weak spots

depending on the interrupt rating.
End-caps: connect the fuse in an electrical circuit. A hole is
provided in one of the end-caps to allow sand lling. The fuse
41
Fuse element Casing Pin

Plug
End-cap
Sand
End-cap without hole
with hole
Figure 3.1 Schematic of a sand-lled fuse.
element is soldered to the end-caps. The external protrusions

are the electrical terminals of the fuse.
Sand: lls the space in the casing around the fuse element.
Fiberglass casing: houses the fuse element and the sand.
Plug: closes the hole in the end-cap after sand lling.
Pins: attach the end-caps to the casing.
When the current in the circuit exceeds the prescribed limit, the
weak spots in the fuse element are designed to melt and create a dis-
continuity. An arc forms across the discontinuity and continues to
conduct current. During this arcing process, the sand serves two key
purposes:
1. It melts due to the intense heat of the arc and absorbs energy
equivalent to the heat of fusion. As a result, it diminishes the
amount of energy available for heating the gases in the cas-
ing and thereby reduces the possibility that the casing will
explode due to expanding gases.
2. The porosity of the sand lling allows the molten metal to
escape from the arcing interface. This reduces the time to
interrupt the circuit.
Thus, the sand lling helps to maintain the mechanical integrity of

the fuse while reducing the interrupt time.
3.2 The Initial Need
The time to interrupt the circuit is a key performance feature of the

fuse. A short interrupt time is desirable because it decreases the load
on various components during the short-circuit condition. The
42
Need Identication and Analysis Case Study
time to interrupt the circuit is highly dependent on the packing fac-

tor (PF) of sand. The PF is dened at the macroscopic level as the
fraction of space occupied by sand grains in the fuse cavity. The gen-
eral expression for the PF is:
Volume of sand
Packing Factor =
Volume of fuse cavity
A typical sand-lled fuse has a PF of 0.63 to 0.68. As the PF decreases,

the interrupt time increases and becomes unacceptable at 0.61 for
most fuses. A fuse manufacturer asked that a quality control device
be designed and incorporated in the production line to identify
faulty fuses. The stated initial need was to check the PF of each fuse
before shipment.
3.3 Need Identication
The black-box approach can be used to study and dene the real
need. For this particular task, the desired output is the PF. The fuse
is both the input and the output for the black box as shown in Fig.
3.2. Based on this model, the need can be formulated as determine
the PF of the fuse.
Now we must ask ourselves whether the need statement is not too
constraining; that is, whether it will permit us to conceive innovative
solutions. This particular need statement suggests that the PF be
determined after the fuse is completely assembled. This limits the
range of possible solutions to just a few nondestructive evaluation
techniques. On the other hand, if the implied constraint the PF must
be determined only after assembly is removed, then the designer is
free to conceive more innovative solutions. For instance, the PF can be
determined based on measurements taken during the manufacturing
process. Thus, the same need can be formulated at a more generic
level as determine the PF during the manufacturing process.
If we can determine the PF before closing the end-cap with the
plug, then it may be possible to add more sand to achieve the desired
PF. This raises the interesting question of whether we want merely
to nd the PF or whether we actually want to control it. The latter
43
Need Statement:
Design Fuse
Fuse Artifact Determine the PF
Packing
factor of the fuse
Figure 3.2 Initial black-box model of the design task and the resulting need
statement.
choice expands the scope of the design task, so the new need can be
stated as ensure the correct PF during the manufacturing process.
A black box for this particular need is shown in Fig. 3.3.
3.4 Need Analysis
We rst list relevant questions that come to mind, such as:
Can the PF be determined from the weight or mass of the fuse?

What are the sources of variation of the PF?
What is the ideal range for the PF?
What are the production rate and volume of the fuses?
What is the cost of a fuse?
How much would the fuse manufacturer be willing to pay for
the designed equipment?
How large can the equipment be?
Next, we attempt to develop a thorough understanding of the func-

tions and constraints involved with this design task, under the ve
categories of performance, value, size, safety, and special.
Performance
The primary function of the design is to ensure the correct PF.
For fullling this goal, it is important to identify:
The maximum possible value for the PF.

The design variables that determine the PF.
The primary sources of variation in the magnitude of the PF.
44
Casing
Sand
Need Statement
Fuse element Fuse with
Design the correct
End caps Artifact Ensure the correct
packing factor
PF during the
Plug manufacturing process
Pins
Figure 3.3 Revised need denition for the design task.
To determine the PF, we can assume the sand grains to be

incompressible spheres. From material science we know that the PF
is dependent on the spatial arrangement and independent of the
grain size; that is, two different grain sizes with the same spatial
arrangement will give the same packing factor. For any grain size,
the PF can range from 0.52 for a simple cubic structure to 0.74 for
an hexagonal close-packed structure. However, including interstitial
grains can increase the PF.
Now let us understand the design variables that determine the
PF. As discussed earlier, the PF is the ratio of the volume of the sand
to the volume of the cavity. The volume of the sand is proportional
to its mass. The volume of the cavity depends on the inside diame-
ter of the casing and the length of the cavity. By substituting terms
in the general expression for PF we get:
(Mass of sand/Density of sand) (M/)

PF = =
Volume of fuse cavity (d 2/4) l
where M is the mass of the sand, is its density, d is the inside diam-
eter of the casing, and l is the length of the cavity.
The mass of the sand can be determined by subtracting the mass
of all the components from the total mass of the fuse. However, due
to manufacturing tolerances, the mass of the fuse components can
vary signicantly, as shown in Table 3.1. Similarly, dimensional tol-
erances on the casing and the assembly process can cause the cavity
volume to change. These variations are summarized in Table 3.2.
45
Table 3.1 Mass variation of the components of sand-lled fuses.
Component Mean (g) Standard Deviation (g)

Fuse element 0.9 0.0025
End-cap without hole 37.75 0.175
End-cap with hole 36.25 0.175
Fiberglass casing 19.00 0.05
Plug 1.25 0.002
Pins 1.00 0.005
Table 3.2 Cavity diameter and length variation.
Design Variable Mean (mm) Standard Deviation (mm)

Diameter 8.00 0.01
Length 26.3 0.1
The two tables reveal that if the total weight of the fuse is used
for determining the PF, then the variation in the mass of the end-
caps and in the length of the cavity can result in signicant error.
The mass variation of other components and the casing diameter
variation result in much smaller errors.
Value
The value of the new system would be the increased quality of
fuse production. Quality Loss Function (QLF) is a systematic and
rational methodology for estimating loss of quality due to off-target
performance in monetary terms. This methodology will be used
here to analyze and quantify the value of the new design.
We know that ensuring a good PF results in high-quality fuses.
As the PF deviates from the desired value, the quality of the product
decreases, and this, in general, manifests itself as a loss to society.
These losses can be due to the product failing to deliver on-target
performance, to harmful side effects of the product, and to down-
time of the equipment. Quality loss is often represented by the fol-
lowing quadratic function:
L(y) = ky 2
46
where L(y) is the loss function, k is a constant known as quality

loss coefficient, and y is the difference between the desired and
actual PFs. For this problem, we can assume the desired PF to be
0.74. The quality loss of a fuse is zero (i.e., no loss in quality) when
the actual PF is equal to the desired value. As the PF deviates from
the desired value, the quality loss increases. Customers take eco-
nomic action when the quality loss becomes unacceptable to them.
The customer tolerance 0 is defined as the value of y at which half
of the customers would take economic action. Based on an infor-
mal survey of fuse customers, half of the customers who have fuses
with a PF of 0.61 would switch to another manufacturer.
Therefore, the customer tolerance was set at 0.13 (desired PF of
0.74 minus actual PF of 0.61).
To quantify the loss, we need to estimate the value of the qual-
ity loss coefcient k. This can be done by identifying the loss A0 at
the customer tolerance of the PF. In reality, the estimation of loss is
quite complex since the cost should include the downtime and the
replacement of damaged equipment for a typical application. For
simplicity, it is assumed that the estimated average replacement cost
(including the cost of a new fuse and the installation cost) is $2.
Thus, the QLF for this problem is:
L(y = 0) = k02 = A0
A0
k= = 118.3
02
and the quality loss of a fuse with packing factor PF can be written as:
L(y) = 118.3 (Desired PF Actual PF)2
Our interest is in nding the average loss for a given set of fuses
manufactured by the current production process. The average qual-
ity loss can be computed as:
(Quality loss of each fuse)

L(y) =
Total number of fuses
47
(y12 + y22 + y32 + + yn2)

L(y) = k
n
k n
L(y) = n yi2 k(S 2 + y 2)
i=1

where L(y) is the average quality loss, n is the number of fuses, y is
the mean y (the mean difference between the actual and desired
PFs), and S 2 is the variance of the PF. The last equation shows that
the quality loss can be decomposed into two parts:
1. Quality loss due to the mean being off target.

2. Quality loss due to the variance.
The quality loss can be reduced by bringing the mean PF close to the
target (0.74) and reducing the variance. Figure 3.4 shows the same
insight: the average quality loss decreases by increasing the PF or by
reducing the variance.
By substituting values for the fuses manufactured through the
current system (PF = 0.66, y = 0.08, and S 2 = 0.01), the average
quality loss can be calculated as $1.94. The target is set at PF = 0.67,
y = 0.07, and S 2 = 0.005. The average quality loss at this target is
$1.17 from Fig. 3.4. Thus, the target performance will result in an
increased value of $0.77/fuse ($1.94 $1.17). Based on a produc-
tion volume of 100,000 fuses per year, the new system would reduce
the losses due to bad quality by $77,000 in the rst year. Even
though this is not a cash ow to the manufacturer, it is realized in
terms of greater customer retention and increased customer satis-
faction. A target cost for the equipment was set at $40,000 based on
the possible quality improvements by the new system. Note that the
relatively low production volume relates to the particular size of the
high-performance fuse.
Size
Because of the tight space constraints in the plant, the maxi-
mum oor space available for the equipment is 8 5. The maxi-
mum permissible height is 8.
48
2.5
2
Packing
Factor
Average quality loss ($)
0.64
1.5
0.65
0.66
1 0.67
0.68
0.5
0
0 0.001 0.002 0.003 0.004 0.005 0.006 0.007 0.008 0.009 0.01
Variance
Figure 3.4 The average quality loss for different combinations of PF and
variance.
Safety
Any device on the shop floor should meet the relevant
Occupational Safety & Health Administration (OSHA) regulations.
These regulations often make some solutions less practical. For
instance, this requirement may rule out the possibility of using an
X-ray machine on the shop oor to determine the PF.
Special
Any modication to the fuse design should meet Underwriters
Laboratories (UL) regulations. The UL regulations set limits on sev-
eral factors such as I 2t (proportional to the energy let through the
system).
3.5 Design Requirements
Table 3.3 summarizes the requirements generated in each of the

need analysis categories.
49
Table 3.3 The design requirements list.
Category Design Requirement

Performance 1. The device must ensure a minimal packing factor of
0.62.
2. Target packing factor: 0.67 mean with 0.005 variance.
3. Mass specication of components:
Standard
Component Mean (g) Deviation (g)
Fuse element 0.9 0.0025
End-cap without hole 37.75 0.175
End-cap with hole 36.25 0.175
Fiberglass casing 19.00 0.05
Plug 1.25 0.002
Pins 1.00 0.005
4. Current cavity specication:

Design Standard
variable Mean (mm) Deviation (mm)
Diameter 8.0 0.01
Length 26.3 0.1
5. Sand: 460 grade
6. Production volume: 100,000/year
7. Production time: 1.5 minute/fuse
Value 8. The equipment cost should be no more than $40,000.
Size 9. Available oor space: 8 5
10. Maximum height: 8
Safety 11. Must meet OSHA regulations for safety.
Special 12. Any modication to the fuse design should meet UL
regulations.
During the need identication step, the need statement changed

from determine the PF to ensure the correct PF. Such expansion
of the scope increases the solution space and the probability of nd-
ing an innovative solution, as discussed in this section.
The first need statement, determine the PF of the fuse, may
result in solutions that nondestructively evaluate the PF. One
such solution uses acoustic emissions. When a fuse is shaken, the
50
sand grains impact each other and the housing. In the process,
they produce sound. The intensity of acoustic emission is
inversely proportional to the PF. In other words, a loosely filled
fuse emits more noise. This energy measurement can be corre-
lated with the PF.
The modied need statement, ensure the correct packing factor
during the manufacturing process, expands the solution space to
include various measurements during manufacturing, such as the
length of the cavity during assembly and the mass of the sand dur-
ing the lling process. Based on the insight that there is very little
variation in the cavity diameter, these measurements can be used to
determine the PF.
Even more innovative designs are facilitated by shifting the
emphasis from quality control to process modification. Identifying
the sources of variation in packing factor may lead to a simple, yet
innovative, solution involving a fixture to control the length of the
cavity. Since the volume now becomes almost constant, the focus
of the design task is shifted to filling a predetermined quantity of
sand.
Since we learned that changing the uniform grain size has
very little effect on the PF, we can focus on introducing a second
grain size to fill the interstitial voids. This can increase the PF
without significantly affecting the manufacturing process. In
summary, a good need identification coupled with a good need
analysis results in crucial insights that can translate into innova-
tive design solutions.
1. Quantify the sensitivity of the packing factor to various design

variables.
2. Determine the accuracy of computing the PF by weighing the fuse.
3. Identify three different ideas that may be used in a design to

ensure the PF.
51
3.8 Bibliography
The following book is a general reference for fuse design:

Wright, A. and Newbery, P. G. Electric Fuses. 2nd ed. London:
Institution of Electrical Engineers, 1994.
Quantifying the loss of quality due to off-target performance is dis-

cussed in:
Fowlkes, W. Y. and Creveling, C. M. Engineering Methods for
Robust Product Design: Using Taguchi Methods in Technology
and Product Development. New York: Addison Wesley, 1995.
52
4
Introduction to
Parameter Analysis
This chapter presents the systematic methodology for conceptual

design, namely, parameter analysis. First, we look at an invention
and try to reconstruct the designers thoughts that led to it. We
explore the nature of the conceptual design process and explain it as
continuous movement between concept space and conguration
space. This theoretical model is used to demonstrate the invention
described earlier and to outline the parameter analysis methodology
for conceptual design.
4.1 A Look at an Invention
A tiltmeter, or inclinometer, is a device capable of measuring

extremely small angles of tilt with respect to the local gravity vector.
When used for sensing earth tide, the distortion of the Earths
crust due to the gravitational pull of the moon, the tiltmeter needs
to measure changes in angle of the order of 106 radians. Another
use is in predicting earthquakes, where the required resolution is of
the order of 107 to 109 radians. Figure 4.1 is a schematic represen-
tation of a tiltmeter that was invented in the 1970s. The lines repre-
sent stiff members, the circles are the pendulum weights, and the
solid dots are hinges. A small tilt, , given to the base disturbs the
equilibrium of Fig. 4.1a and produces a much larger, and therefore
easier to detect, inclination of the pendulums, Fig. 4.1b.
53
(a) (b)
Figure 4.1 (a) Schematic of the tiltmeter with no input angle and (b) the
tiltmeters large response angle when measuring a small tilt angle (not
to scale).
It is fairly obvious, as it was to the inventor, that a simple pen-

dulum (Fig. 4.2) is a conguration that provides a measurement of
tilt angle through a measurement of the lateral displacement of the
pendulum weight. However, the size of the device presents a prob-
lem. The lateral displacement of the weight at the bottom of a sim-
ple pendulum is large enough to be measured at small angles of tilt
only when the pendulum is extremely long. For example, a 50-m
long pendulum would produce a displacement of 5 m when the
angle is 107 radians. This is a relatively obvious concept. However,
the inventor also realized that one could represent this displacement
relationship as a simple spring. A displaced spring pushes back with
a force f, which is proportional to the displacement x (f = k x).
For the pendulum of Fig. 4.2, the restoring force f is also propor-
tional to the lateral displacement:
x mg
f = mg sin = mg = x = k x
l l
Therefore, the statement that the pendulum must be very long

is the same as saying that k, the stiffness of the pendulum, must be
54
Introduction to Parameter Analysis
mg
Figure 4.2 A simple pendulum with a restoring force f may be viewed as a

spring.
very small. Continuing this logic, the inventor then recognized that
there is another way to obtain a small spring constant in addition to
extending the length of a simple pendulum. The difference between
two large spring constants (short pendulums) can yield a small
spring constant (effectively long pendulum). This relationship is
represented by f = (k1 k2) x, but it requires a negative spring in
order to obtain the k2 term. A negative spring would pull in the
direction of the initial displacement with a force proportional to the
displacement. Many statically unstable devices exhibit a similar
behavior; for example, some light switches jump to a different
state when pushed slightly. The inventor of the tiltmeter noted that
an inverted pendulum is a negative spring. Thus, all that remained
was the coupling between the two pendulums at a point at which the
resultant spring constant (k1 k2) is sufciently small but positive to
yield very high sensitivity (large lateral displacement) for small
angles of tilt.
This very brief description represents only a small portion of
the process that took place in creating the new tiltmeter. However, it
55
does describe the kernel events around which the whole concept
revolves. We shall return to the tiltmeter later, but let us rst exam-
ine a few fundamental issues relating to conceptual design and
attempt to establish some principles with regard to the process.
4.2 The Nature of Conceptual Design
When we look at inventions such as the tiltmeter, we often ask basic

questions such as:
Why is conceptual design so difcult?

Where do new ideas come from?
Why didnt I think of that?
These questions seem almost trivial or trite, but they are fundamen-
tal to conceptual design, and an attempt to answer them helps to
elucidate the subject.
Let us rst consider why conceptual design or the generation of
new ideas is such a difcult task. Implicit in this question, of course,
is the task of generating very good ideas as opposed to just generat-
ing ideas. It is quite a simple matter to generate many, many ideas
when the demands on you do not include the quality or goodness
of the ideas. As we all know, many more ideas are generated than are
developed; many more are developed to some level of completion
than are made into products; and many more products are made
than are successful. The real issue comes down to generating ideas of
high quality and of good value to others.
Figure 4.3 is a schematic view of why conceptual design is
fundamentally more difficult than many other stages in the
design process. The result of conceptual design is, hopefully, a
new concept or configuration. It is this newness that makes the
process so elusive. A new configuration that heretofore did not
exist has been created, and the process of reaching that point is
very different from the activity that takes place during the refine-
ment of the concept downstream in the process. The horizontal
axis in Fig. 4.3 is a one-dimensional representation of the many
56
New concept
generation
"Goodness"
Design
optimization
Design variables
Figure 4.3 The relationship between conceptual design and design opti-
mization.
design variables that can be changed within a configuration. The

vertical axis is a one-dimensional representation of the quality or
goodness of a concept. Such a measure may be imagined to be
the ratio of performance to cost since, fundamentally, we would
all like to pay less for more performance. In any event, such a
measure is a convenient way to describe the desirability of con-
cepts or configurations.
Consider now the curve on the left and imagine that this curve
is actually a multidimensional surface in space and that the points
on this surface represent different combinations of the design vari-
ables that can be reached. All points on the surface are various real-
izations of a single concept.
At this point, two simple examples will help. Imagine that the
curve on the left describes a fountain pen. That is, the core technical
concept is described by the fact that the pen contains a nib through
which ink ows by capillary action from some sort of a reservoir to
the paper. Points on this surface may describe pens that range from
a quill pen to a sophisticated rellable fountain pen to a fountain
pen that has a very wide tip used for calligraphy. These pens differ in
57
the values of many design variables, but the core technical concept
is still the same. Movements from one point to others on the same
surface may come relatively easily by making minor changes to a
design variable. These changes might include modifying the type or
size of the reservoir of ink, changing the physical dimensions of the
nib, or making small changes to the properties of the ink. All of the
realizations are still on the surface because they all share the same
core technical concept. Each of the points represents one realization
of the common attribute of being a fountain pen. These move-
ments may be described as design optimizationa selection of cer-
tain existing design variables in order to meet a performance
specication.
In contrast, conceptual design is represented by a leap from one
surface in the design space, the left curve in Fig. 4.3, to a new sur-
face, represented by the curve on the right. The new curve may not
even be described by all of the same design variables. Before the leap
was made, the new core technical concept did not exist. To extend
the example, imagine that the new curve represents a ballpoint pen.
In fact, some of the design variables that are pertinent to the foun-
tain pen are also important to ballpoint pens. However, these two
congurations do not share a number of design variables, and there
is a fundamental difference between the core technical concepts of a
fountain pen and a ballpoint pen. The fountain pen uses capillary
action to feed the ink while the ballpoint pen does it through the
mechanical rolling motion of the ball. All points on the curve to the
left are fountain pens, all points on the curve to the right are ball-
point pens, and the generation of the ballpoint pen concept was not
obvious in the consideration of the fountain pen designs. It is the
fact that this process is a leap to a nonexistent curve that makes con-
ceptual design so difcult.
As a second example, suppose the left curve in Fig. 4.3 describes
the concept of reciprocating internal combustion engines, while the
right curve represents rotary engines. The reciprocating engine
curve would consist of points representing various congurations
(V-shaped, in-line, horizontally opposed, radial), number of cylin-
ders, methods of valve actuation (side valves, overhead cams), and
so on. However, all these congurations would be based on the con-
58
cept of reciprocating pistons in cylinders, connecting rods, and

crankshafts (Fig. 4.4a). The rotary engine curve may represent dif-
ferent geometries (for example, a triangular rotor in a figure-eight
housing as in the Wankel engine, Fig. 4.4b, or any of the many
other theoretical possibilities), numbers of rotors, and so on.
Again, movement along the same curve stands for changing some
of the design variables, such as the number of spark plugs used or
the shape of the combustion chamber. In contrast, the creation in
the 1950s of the concept of rotary engines is denoted by the estab-
lishment of the new surface on the right in Fig. 4.3 and represents
a conceptual breakthrough.
This discussion of the generation of new ideas has been only
philosophical, but it should help to clarify why the creation of truly
new, high-quality congurations or ideas is so difcult and therefore
an event too rare in engineering practice today. It also shows that
good ideas do not usually emerge from doing conceptual design in
the form of a search over existing solutions. Rather, conceptual
design is a discovery process.
Let us now try to answer the other two questions posed at the
beginning of this section. Study of the process of conceptual
design and a look at many inventions allow us to make the follow-
ing observations:
Human creativity is more successful or productive when one is

attempting to solve simple problems rather than complex ones.
The best ideas are usually quite simple conceptually.
These notions need some explanation. The rst notion deals prima-
rily with the number of issues or aspects that must be considered
during the solution process. Thus, low-order problems, those that
require consideration of very few parameters or factors, tend to be
much more easily solved than higher order problems, those with a
large number of variables to be considered. That is why all of us try
to break down problems into smaller pieces in an effort to make
them more tractable. This simple notion comes as no surprise but
has profound implications for the nature of successful creative
processes.
59
(a) (b)
Figure 4.4 (a) A reciprocating internal combustion engine and (b) a rotary
Wankel engine.
The question, Where do good ideas come from? may be

addressed by examining the second notion, which relates to the sim-
plicity of the results of the creative process. It is not that the result-
ing products based on these good ideas (core technical concepts) are
simple, rather, the concepts themselves are most often based on sim-
ple but creatively new insights. Often, these simple ideas can be
described as resulting from transformations of the key issues in a
problem into new relationships or sets of relationships uniquely
identied by the creative designer.
As a result, the answer to the frequently asked question, Why
didnt I think of that? may be twofold: Perhaps I failed to break
down the task in a way that would allow me to focus on one issue at
a time; or, while in retrospect the ideas behind the design are simple,
the pathway to reach them may not have been at all obvious to me.
60
4.3 Theoretical Model of Conceptual Design
The notions regarding the nature of conceptual design lead to several

conclusions about a model for the conceptual design process. Figure
4.5 presents a simple schematic model. The output of the process is
generally a physical object or group of objects that here are called
congurations. Thus, the various outputs as the process unfolds can
be thought of as elements in conguration space. However, consider-
ation of the preceding observations concerning problems and solu-
tions within conceptual design indicates that movement from one
point in conguration space to another point in conguration space
is generally made at the level of concepts. Therefore, the model also
contains another space, called concept space, which contains the ideas
or concepts that provide the basis for the elements of conguration
space. The conceptual design process can be viewed as an iterative
process of moving from conguration space to concept space, mak-
ing changes or movements within concept space, and then moving
back to new points within conguration space corresponding to fur-
ther creative generation of congurations based on these new ideas.
The process of moving from conguration space to concept
space can be thought of as abstraction or generalizationthat is, gen-
eralizing from the particulars of a physical conguration to a con-
ceptual understanding in order to change it or improve upon it.
Movement from concept space to conguration space can be
regarded as realization or particularizationbringing to reality, in
particular physical form, the technical concepts arrived at within
concept space. Readers should note that in this process description,
the elements of conguration space are not actually physical objects
but only diagrams, sketches, or other representations of physical
objects motivated by the conceptual elements of concept space,
which bring some real form to the thoughts created in concept space.
This model should be viewed as much more than a philosophical
statement. Readers are encouraged to examine their own experiences
in conceptual design to be further convinced of its signicance. As
ideas come to life, there are many points along the way which are either
intermediate congurations (elements in conguration space) or con-
ceptual statements or relationships (elements in concept space). Rarely
61
Realization
(particularization)
Concept Configuration
space space
Abstraction
(generalization)
Figure 4.5 A theoretical model of conceptual design.
within this process is a new conguration created from the previous

one without an excursion to concept space, with new conceptual
insight being the driving force for the new congurational result.
Let us return now to the tiltmeter example. Of particular
importance is the nature of the events that took place within con-
cept space. First, there was simplification of the problem by con-
sidering only the pendulum conguration and ignoring, temporarily,
many other issues that obviously are important. Second, there was
a transformation from the normal way of looking at a pendulum
to viewing a pendulum as a spring. Third, there was the creative
step of recognizing the relevance of the difference between two
large numbers to the situation at hand. Finally, we should not
ignore the additional creative steps to generate the double pendu-
lum configuration.
Figure 4.6 is a modication of Fig. 4.5 in order to include an
important observation. On the right-hand side, within congura-
tion space, are represented two realizations of pendulum devices: a
simple pendulum and the new coupled-pendulum tiltmeter. On the
left-hand side, within concept space, are two concepts: the simple
spring relationship and the spring relationship represented by the
difference between two stiff springs. The transition from the simple
pendulum to the complex pendulum within conguration space is
not likely to take place by itself. On the other hand, the transition
from the concept of a simple spring, correlating with the properties
of a pendulum, to the second concept is motivated by recognizing
62
1
f=kx
3
f=(k1k2)x
space space
Figure 4.6 A simplied map of the creation of the new tiltmeter.
the elements of these conceptual issues. Thus, the creative motion

within conguration space is actually a movement driven by motion
within concept space. In simple terms, the process unfolds with the
movements as labeled numerically in the gure.
Notice also that the movement labeled 1 was in itself a creative
step. It represents a simplication and transformation of the typical
understanding of pendulums to a form that, downstream, led to a
very interesting result.
The reader should be aware that this diagram of movements is
extremely simplied. In this example of creative conceptual design,
many more motions were taking place back and forth between con-
cept and conguration which have not been included in the dia-
gram. Furthermore, the process at this stage of the description is far
from over. Many more conceptual issues remain, such as how to
obtain hinges with the properties necessary to allow the new device
to perform as required and how to measure the lateral displacement
of the pendulum weight.
Of course, conceptual design problems have many solutions, as
does this problem of measuring extremely small angles of tilt. For
example, Fig. 4.7 shows three other congurations that could be
used instead of the double pendulum to implement the concept of
measuring tilt with a long pendulum. The virtual length of the
63
Radius of
curvature
P
(a) (b) (c)
Figure 4.7 Three realizations of the long-pendulum concept: (a) a ball

rolling on a slightly concave surface, (b) a ball moving in a pipe connecting
two reservoirs, and (c) a mechanism whose point P moves along a trajectory
with innite radius of curvature.
pendulum of Fig. 4.7a equals the radius of curvature of the surface

on which the ball rolls, which can be made very large. The congu-
ration of Fig. 4.7b uses a large difference in area between the two
reservoirs and the connecting pipe to obtain a large gain. The move-
ment of the small ball, or the interface between two nonmixing u-
ids, is used to detect the tilt. Figure 4.7c is only one of many
examples of mechanisms that have a point whose trajectory has an
innite radius of curvature. The signicance of the double pendu-
lum example is not the particular solution presented but the types
of information that were used during the process and the relation-
ship between the elements as the process moved forward.
Readers will remember instances when they were stuck in a
rut. One may represent this situation as an effort to undergo cre-
ative movements within conguration space, when in fact it is most
benecial to make signicant movement within concept space and
between concept space and conguration space.
4.4 Parameter Analysis as a Conceptual Design

Methodology
Understanding the nature of the conceptual design process and the

process model proposed above leads us to generate a methodological
64
approach to carrying out conceptual design. We must rst state very

clearly that this approach is not to be considered a method but rather
a methodology. The importance of this understanding cannot be min-
imized since the word method implies that a procedure that unfolds
in an orderly and sometimes predictable manner results from such an
approach. However, successful design methodologies must be much
more open, iterative, and exible than a highly systematized method
would allow. Therefore, the methodology presented here is a general
approach based on the fundamental notions regarding the nature of
conceptual design. A methodology gives one the general direction of
how to approach problems rather than outlining a rigid procedure.
Readers should also appreciate that the structured framework in
which the methodology is presented serves only to describe the
approach in an effective way. Thus, although the methodology
appears in a very graphical and orderly form, this is only to illustrate
the general principles on which the methodology is based.
A general word about problem solving techniques is in order. As
a way of describing various techniques to enhance creativity and, in
particular, conceptual engineering design, one might describe vari-
ous approaches as being environmental or methodological.
What is meant by environmental here is that creativity is encour-
aged or stimulated by establishing an environment in which creativ-
ity is more likely to occur. Methodological approaches, such as that
which is described here, consider the enhancement of the creative
process from the point of view of active intervention in the unfold-
ing of the process rather than of simply causing conditions to be met
that will encourage creativity.
The methodology for conceptual design presented in this book
is called parameter analysis and is represented by the schematic of
Fig. 4.8. Figure 4.9 indicates the correspondence between the theo-
retical description of the process and this methodological approach.
The three types of activity within parameter analysis form an itera-
tive loop. The rst element, parameter identication, consists prima-
rily of the recognition of the dominant parameters or issues in a
problem. The word parameter is used to describe in a very general
way any issue, factor, concept, or inuence that plays an important
part in developing an understanding of the problem and pointing to
potential solutions.
65
Parameter Creative
Identification Synthesis
Evaluation
Figure 4.8 Schematic of the parameter analysis methodology.
Realization
Parameter Creative
Identification Synthesis
space space
Evaluation
Abstraction
Figure 4.9 Correspondence between the theoretical model of conceptual

design and the parameter analysis methodology.
The best descriptions of what takes place within parameter identi-

cation are simplication and transformation. The parameters within
a problem are not xed but rather are those that are creatively recog-
nized as the process moves forward. The recognition of interesting new
66
parameters, based on transformations of more commonly understood

parameters, triggers new, unique solutions to problems. Also, tempo-
rary purposeful oversimplications of a problem, made to focus on
only one of the key parameters that characterize it, help to concentrate
creative activity so that it can produce quality ideas. Identication of
these key parameters through simplication and transformation does
not yield results unique to a problem but is a highly creative process.
The second part of parameter analysis is creative synthesis. This
part of the process represents the generation of a physical congu-
ration based on a concept recognized within the parameter identi-
cation step. Since the process is iterative, one should expect many
physical congurations to be generated as the process unfolds, not
all of which will be very interesting. However, the physical congu-
rations allow one to see new key parameters, which will yet again
stimulate a new direction for the process.
This process of moving from a physical realization back to
parameters or concepts is facilitated by the third part of the param-
eter analysis methodology, the evaluation step. Evaluation is impor-
tant because one must consider to what degree a physical realization
is a possible solution to the entire problem. As parameter identica-
tion and creative synthesis proceed, one should expect many physi-
cal congurations to be generated. While these are far from being
valid solutions to the problem, nevertheless, they are important to
the stimulation of the creative process. Evaluation should not con-
tain analysis of physical congurations that is any deeper than that
required to create a fundamental understanding of its underlying
elements. We might best call this process appropriate analysis. It
must be carried out in the light of the entire need, not just with
respect to the particular simplied parameter of each iteration.
The major implication of the theoretical model that stands
behind parameter analysis is that the burden of truly creative activity
is shifted from what here is called creative synthesisthe generation of
physically realizable congurationsto parameter identication, the
creation of new conceptual relationships or simplied problem state-
ments (all lumped under the term parameters), which will lead to
the desired congurational results. Thus, the task of creative synthesis
along the way is only to generate congurations that, through evalua-
67
tion, will enlighten the creative identication of the next interesting

conceptual approach. Each new conguration does not have to be a
good solution, only one that will further direct the discovery process.
Several characteristics are at the foundation of parameter analysis:
Behind every congurational attribute of a design there must

be a conceptual reason or justication.
Developing a design requires focusing on one or very few crit-
ical conceptual issues at a time.
The most creative aspect of conceptual design is not the inven-
tion of a new conguration, but rather the designers ability to
discover a conceptually new way of looking at the problem.
The key to good conceptual design is a thorough understand-
ing of the underlying physics of design artifacts.
Conceptual design is not pure synthesis; rather, it consists of a
partnership between synthesis and analysis.
Good design is a synthesis of a series of good ideas, or concepts,
not just one good idea. The parameter analysis methodology
helps to create ideas and incorporate them in a single design.
Conceptual design by parameter analysis can be done at differ-

ent levels of abstraction when considering the degree of innovation
involved. The distinction made in Section 4.2 between design opti-
mization and new concept generation depends on the scope of the
particular design. For example, the invention of the rotary Wankel
engine may be perceived as the creation of a new concept (and the
generation of a new surface in Fig. 4.3) in the broad context of
designing internal combustion engines. At this level, the invention
of the spherical piston head, as opposed to at-top pistons, is
merely an optimization step illustrated by movement along an
existing surface to a point with a higher goodness value. However,
had the scope of the design task been limited to improving the ef-
ciency of reciprocating engines, the degree of innovation intro-
68
duced by spherical piston heads would certainly entail the creation

of a new surface or concept. Similarly, the leap from external to
internal combustion engines (and the future leap from internal
combustion engines to batteries and fuel cells) is a bigger innovation
than the move from reciprocating to rotary engines.
Parameter analysis can also be applied to routine design tasks,
those that we would not normally associate with innovation. Many
designs are developed through a process that is more evolutionary
than revolutionary. Most developments in the automotive industry,
for example, are of this nature. This does not imply that todays cars
are not good designs. On the contrary, and perhaps thanks to the
long evolutionary development of automobiles, they represent a
very solid and rened solution to the basic need of providing a
means of transportation. Evolutionary designs do not necessarily
lack innovation. Electronic ignition systems, fuel injection, antilock
brakes, and airbags are all novel designs. And parameter analysis can
be applied to rening an existing design as much as to tasks that are
totally new and do not have any existing solution.
1. Find additional examples of design optimization versus new con-

cept generation, as described in Section 4.2. How do these exam-
ples depend on the level of innovation, as discussed in Section 4.5?
2. Study the operation of felt-tip pens. Do they represent a design

optimization on an existing fountain pen or ballpoint pen curve,
or do they constitute a new concept?
3. Look back at a design process, either yours or someone elses, and

classify its elements (e.g., decisions, ideas, sketches) into those
that belong in concept space and those that are members of con-
guration space.
4. The tiltmeter described in this chapter requires near frictionless

hinges. Find out how this is done in other applications with sim-
69
ilar demands. Then, check what is suggested for this purpose in

the original tiltmeter patent (see Bibliography).
4.7 Bibliography
Much of the material in this chapter is also found in:

Jansson, D. G. Conceptual Engineering Design. Chapter 23 in
Design Management: A Handbook of Issues and Methods,
Oakley, M. (ed.). Oxford: Basil Blackwell, 1990.
The broad context of parameter analysis, even beyond its use as a

methodology for conceptual design, is described in:
Li, Y. T., Jansson, D. G., and Cravalho, E. G. Technological
Innovation in Education and Industry. New York: Van Nostrand
Reinhold, 1980.
The inventor of the tiltmeter used in this chapter is Y. T. Li, the father
of parameter analysis and a former colleague of D. G. Jansson. A full
description of the tiltmeter is in the following patent record:
Li, Y. T. Sensitive Tiltmeter. U.S. Patent 3,997,976, 1976.
Some information on the dynamic aspects of the tiltmeter can be

found in:
Redeld, R. C. and Krishnan, S. Towards Automated Conceptual
Design of Physical Dynamic Systems. Journal of Engineering
Design 3, No. 3 (1992): 187204.
70
5
Parameter Analysis Put to Work
The previous chapter presented the theory behind the parameter

analysis methodology. This chapter introduces the preliminary
steptechnology identicationin which several core technologies
are lined up to form the starting points for design. The three basic
building blocks of parameter analysisparameter identication,
creative synthesis, and evaluationare then described separately.
Several guidelines and illustrative examples are provided to help the
reader execute each of these steps efciently. Finally, we focus on
practical aspects of applying parameter analysis to conceptual design,
and we outline some important characteristics of the process.
5.1 Parameter Analysis Process Overview
Figure 5.1 shows what this chapter is all about. Suppose that we have
already nished a comprehensive need analysis in which we investi-
gated the design task in depth and gained considerable knowledge
about the problem domain, functions, and various constraints. We
have also generated the design requirements, so now we are ready to
begin with the actual design of the new artifact. When we are n-
ished with the conceptual design stage (the dashed block in Fig. 5.1),
we should have one or several well-developed conceptual designs. As
you will soon learn, these conceptual designs are much more than
general ideas of how to solve the design task. In fact, they are quite
71
Conceptual Design
Parameter
from Analysis
Need to downstream
Analysis PI CS design stages
Technology
Identification
Figure 5.1 The conceptual design stage consists of technology identication

and parameter analysis. (PIParameter Identication, CSCreative Synthesis,
EEvaluation).
detailed congurations that have evolved through several cycles of

parameter analysis.
The rst block shown in the diagram is called technology identi-
cation. It refers to the process of looking into possible technologies
that can be used for the design task at hand, thus establishing several
starting points, or initial conditions, for parameter analysis. Having
selected a core technology, we can now begin the parameter analysis
process. As discussed in the previous chapter, the parameter analysis
methodology emphasizes the discovery of one or a few critical con-
ceptual issues at a time, calls for implementing these concepts as
congurations, and directs the designer to keep evaluating the
evolving design to identify new, emerging dominant issues at the
conceptual level. The process consists of going through cycles of the
following distinct steps, which correspond to the above activities:
1. Parameter Identication (PI): Recognizing or discovering

new important conceptual issues.
2. Creative Synthesis (CS): Generating a conguration, or
hardware, that realizes the previous concept.
3. Evaluation (E): Assessing the last conguration.
Parameter analysis can be applied to any design task and at any level
of innovation. It facilitates technological innovation, or technologi-
cal breakthroughs, by assisting designers during the important con-
72
ceptual design stage, where most opportunities for novel design

solutions exist.
5.2 Technology Identication
Often, several core technologies, or physical principles, can be used in

a particular design. This is especially true when designing sensors, but
also with other design tasks, as discussed in the following example.
Technology Identication for a Robot Plow
Suppose we are assigned the task of designing an autonomous

plow for agricultural use. What is needed is an automatic
devicea sort of a farm tractorthat will plow elds with a
minimum of human intervention. We perform a need analysis in
which we quantify all the relevant considerations, such as eld
sizes and topographies, required force and speed of plowing,
navigation and guidance, and cost. With our newly gained
understanding and knowledge in the area of farming, as well
as a list of design requirements, we proceed to the conceptual
design stage.
Realizing that the plowing operation itself is a well-estab-
lished technique and that existing farm tractors do a good job
of providing the necessary mobility and supporting platform for
a plow, we decide to focus our effort on the navigation and
guidance aspect of the design. Technology identicationthe
listing and assessment of possible means of guiding an
autonomous vehicle in a eldmay look like this:
Vision
Ultrasound
These are technologies for sensors that can be mounted on the

tractor and used to detect nearby objects. They seem more suit-
able for obstacle avoidance than for navigation.
73
Buried line following
This technology is widely used in guiding autonomous vehicles

on factory oors. A buried metal line, for example, could be
detected and followed with the help of magnetic sensors. Three
problems with this method are quickly recognized: The initial
cost of installing the guiding lines might be very high; the plow
would always follow the same path with no exibility; and the
plow might accidentally destroy the guiding line if the latter is
not buried deep enough.
Triangulation
The robot plow could nd its location in the eld at any time by
measuring range, direction, or a combination thereof, in rela-
tion to some xed targets. Comparing this location to a stored
map of the eld, an on-board navigation computer could guide
the vehicle. The type of measurement can vary widely between
radio signals, lasers, and more.
Global positioning
Systems that rely on signals from satellites have recently become

quite affordable. As with triangulation, the measured location of
the vehicle should be compared with a pre-stored desired loca-
tion. Whether the degree of accuracy obtainable with a com-
mercial system would satisfy the requirement of this design is
somewhat questionable, however.
This example is by no means a complete listing of pertinent

technologies, but it demonstrates several advantages of carrying out
technology identication. First, it offers additional insight into the
problem domain. For example, if we had not realized it during need
analysis, it now becomes clear that in addition to a navigation sys-
tem for locating the tractor relative to its desired path, we may also
74
need to provide a means of obstacle avoidance to increase the oper-

ating safety.
Second, technology identication offers us several possible ini-
tial conditions for parameter analysis. From the preceding list we
select the technology that seems most likely to result in a successful
design, and the rst parameter, or concept, usually becomes: use
technology X to achieve task Y. For the robot plow example, X
might be triangulation and Y, navigation.
If more than one conceptual design is to be developed, as is
often the case, different technologies identied at this stage can be
used as the starting points to yield conceptually different results.
And in case one parameter analysis attempt fails, the list of various
candidate technologies can be revisited and another selection can be
made for a second attempt.
5.3 Parameter Identication
The main concern during parameter identication is to generate a

new and improved concept using scientic and engineering princi-
ples. This is done by examining all the information about the design
task, the latest congurational solution to it, and the results of the
latest evaluation step, and generating a new design parameter.
Design parameters are not design variables. They normally are

not dimensions, material properties, or any other piece of con-
gurational information, which in other contexts may be
referred to as parameters. Design parameters in the context
of parameter analysis and this book are issues at the concep-
tual level. They may be the dominant physics governing a
problem, a new insight into critical relationships between
some characteristics, an analogy that helps shed new light on
the design task, or an idea of what should be the next focus of
the designers attention.
Dominating parameters may not be immediately apparent in a new

situation but may have more inuence on the outcome than those
75
that are obvious. The following example illustrates some character-

istics of parameters.
Critical Parameters in Electrical Fuse Design
Fuses are safety devices that prevent damage to an electrical

circuit due to excessive current under overload and short-circuit
conditions. The difference between these two conditions is the
magnitude of the current involved, with short-circuit currents
being orders of magnitude greater than overload currents.
It is quite obvious that a fuse must react faster than the event
against which it is guarding. So a critical parameter in design-
ing a fuse may initially be formulated as make the time
required for blowing the fuse shorter than the time it would take
the excessive current to damage the circuit. This parameter is
often quantied as a timecurrent graph, where the required
time to blow is plotted for various currents.
Now suppose we are designing a fuse for short-circuit pro-
tection. To determine the critical parameters, we need to examine
the underlying physical processes involved. The time required to
blow a fuse depends on the heat generated in the fuse element
(which is a function of the current owing through the fuse ele-
ment and its resistance), heat dissipated from the fuse element,
and its material properties, such as melting temperature, specic
heat, and latent heat. It seems that these energy considerations
dominate the design and are sufcient to yield a satisfactory fuse.
A closer examination, however, will reveal that arcing may
occur in the fuse during the blowing period. The arcing current
continues even when there is a discontinuity in the fuse element.
The critical parameter therefore becomes control the arcing,
which was not obvious from the initial energy considerations.
In the parameter identication stage, the designer spends con-

siderable time and effort to properly dene the current concept to
be realized next. The extent of innovation achieved is largely gov-
erned by the designers ability to recognize crucial relationships that
76
dictate possible solution paradigms. An innovative design often

results from the designer identifying parametersconcepts and
ideasthat make some relationships irrelevant and break away
from current solution paradigms. This is illustrated by the next
example.
Neutral Buoyancy: A Common Parameter in

Submarine Design
Designing a submerged structure to be neutrally buoyant is a

widely used concept. An obvious example is a ship, but the
gates of the Panama Canal were also designed to be neutrally
buoyant to relieve the hinges from carrying the weight of the
gate. The same concept, or parameter, dominates the design of
contemporary submarines.
Submarines rely on stealth for their functioning and sur-
vival, so a small submarine that can go deep is desirable.
However, deeper-going submarines experience higher hydro-
static pressures and require thicker, heavier hulls. A crucial con-
sideration in submarine design is therefore to maintain neutral
buoyancy by increasing the size of the submarine to where the
weight gain due to the larger hull thickness is balanced by the
weight of the additional water volume displaced. The result is a
direct relationship between the operating depth and the size of
the submarine.
The key to radical innovation in submarine design lies in
discovering such fundamental relationships and identifying con-
cepts that might make them irrelevant. For instance, sub-
mersibles that use hydrofoil or vector propulsion to raise and
lower themselves can have some advantages in certain naval
activities. And in this case, because the submarine does not
need to be neutrally buoyant, its size is almost independent of
the operating depth. An analogous paradigm shift occurred
when air transportation moved from the lighter-than-air air-
ships to heavier-than-air airplanes.
77
Parameter identication includes two key activities: simplication

and transformation. Simplication of a design task is aimed at making
it less complex, and it usually involves the following activities:
Stripping the task of less important considerations: Less

important is a temporal term that depends on the exact stage
in the conceptual design process. Issues that are less important
early in the process may later be identied as the critical
parameters or may only be tackled during embodiment or
detail design.
For example, the double-pendulum tiltmeter design
described in Chapter 4 would not be a viable solution without
devising a conguration for nearly frictionless hinges and an
effective means of measuring angles or lateral displacements.
However, these issues are only secondary at the beginning of the
design process, when the more critical aspects are the view of a
pendulum as a spring and obtaining a small stiffness by sub-
tracting two large stiffnesses. Moreover, the hinges and meas-
urement device cannot be identied as critical parameters at the
outset of the design process because they only emerge later, in
connection with the particular conguration developed.
Focusing on make-or-break issues: While several issues con-
front the designer at each step of the conceptual design
process, he or she must be able to identify those aspects of the
task that, if not done right, would jeopardize the whole design.
Designers focus their efforts on solving these issues rst.
Consider, for example, the robot plow discussed in the pre-
vious section. The actual plowing operation (i.e., disturbing
the soil with the plow blade), propelling, braking and steering
a vehicle, and designing a tractor-like vehicle to traverse vari-
ous terrain, are all secondary issues that we believe we know
how to handle. What turns out to be of primary signicance
at the beginning of this design process are the issues of navi-
gation and accuracy: The autonomous vehicle has to know
where it is and where it is supposed to be, and this must be
done with certain accuracy which is dened by the require-
ments. From our understanding of the state-of-the-art of tech-
78
nology, we judge that the navigation and accuracy are the

make-or-break issues in this design task.
Developing simple models to understand governing relation-
ships: It is often necessary to develop simple models that cap-
ture some essential relationships. These models can provide
deeper understanding of the underlying scientic principles
and identication of potential conicts or inconsistencies that
need to be resolved during the design process.
Understanding Essential Relationships in Car Headlights
The design of car headlights poses an interesting challenge.

Two key requirements for headlights are: provide visibility
for night driving and avoid the blinding of oncoming traffic.
In the traditional headlight design, these two requirements
can be satisfied by varying the brightness of the light. As
shown in the relationship model of Fig. 5.2, increasing the
brightness positively affects visibility and negatively affects
blinding. Thus, the model accentuates a key conflict in head-
light design.
Provide Avoid
Visibility Blinding
Impacts Impacts
Positively Negatively
Increase Brightness
Figure 5.2 A simple relationship model for car headlight design.
79
It is up to the designer to continually judge what are the critical

parameters at any given point and to differentiate them from the less
important issues. To a large extent, the quality of the whole concep-
tual design process depends on this ability of the designer. By focus-
ing only on the most dominant parameters, the overall design task
is simplied, thus increasing the chances of discovering or changing
the fundamental principles and concepts of the design.
The second important activity during parameter identication
is transformation of design parameters. This is the thought process
of gaining new understanding of the task, and it often takes place
when one concept is abandoned in favor of a new solution
approach. Searching for a new conceptual direction, the designer
identies new design parameters that may lead to the generation of
new ideas. Design parameters are often transformed as a result of
one or more of the following activities:
Rephrasing the task statement: Restating the design task in a

way that emphasizes the important issues and postpones the
consideration of secondary parameters. This often involves
replacing some key words with more appropriate ones and
allows the designer to view the design task differently.
Rephrasing the Task Statement for a CT-Scanning Fixture
A large defense manufacturer needed a universal xture for

holding assemblies of various sizes and shapes at any desired
orientation for quality control of the inner components by com-
puted tomography (CT) scanning. The original task statement
design a xture to hold objects at any orientationled the
designers to attempt a structure that would securely hold test
objects at any orientation given to them by the human operator
before placing them on the xture in the CT scanner (Fig. 5.3a).
After much struggle with this design task, a breakthrough took
place when it was realized that the object could rst be
clamped in a xture in some easy-to-clamp orientation, and then
80
Spring-
Test loaded, Test
object lockable object
pins
(a) (b)
Figure 5.3 (a) A schematic of a xture to hold a test object at any

orientation. The pins are spring-loaded and conform to the shape
of the object before being locked in place. (b) A simpler solution in
which the object is rst clamped and then oriented with the mov-
able part of the xture.
the object plus xture could be given the desired orientation as

one unit. The modied task statementdesign a xture to hold
and orient objectssoon resulted in a successful solution,
shown schematically in Fig. 5.3b.
Explaining the cause of a problem in a different way: This

activity is strongly related to the need to understand underly-
ing physics as a precondition to successful design.
Understanding the Underlying Physics of

Clear-Ice Manufacturing
Ice served on airplanes and expensive restaurants is clear,

whereas the ice cubes frozen in refrigerators are hazy or
cloudy. At a supercial level, the haziness is attributed to the
presence of dissolved air in the water. During solidication, the
81
air separates from the water and forms small air bubbles. An
obvious solution would seemingly be to eliminate dissolved air
in the water. After some design effort, however, this task turns
out to be very difcult, forcing the designer to further investigate
the root cause of haziness in the ice.
Water at room temperature contains a certain amount of
dissolved air. As the water is cooled, ice crystals start to form
on the outside surfaces and proceed to grow inwards when the
temperature is reduced further. Because the solubility of air in
ice is very small, these crystalsknown as primary icerelease
dissolved air to the liquid water, thus increasing the concentra-
tion of dissolved air in the liquid. Then, at the eutectic point, the
remaining water (with the dissolved air) freezes to form sec-
ondary ice. During this reaction, the air separates out of the ice
into small air bubbles, which produce the undesired cloudiness
in ordinary ice. Thus, the primary ice formed in the early stages
tends to be clear, whereas the secondary ice is cloudy. When
designing a device to produce clear ice, the critical parameter
becomes prevent the liquid solution from reaching the eutectic
point. This insight is used in building most of the commercial
clear-ice machines.
Stepping back: Reexamining the design task from a differ-

ent perspective.
Different Perspectives on Controlling Industrial Robots
The accuracy of industrial robots relies heavily on the rigidity of

their structure and elimination of backlash in the joints. The posi-
tion of the end-effector of a robot arm is determined by sensing
displacements at the joints. Any deection of the arm or back-
lash in the mechanism would create a discrepancy between the
calculated and actual locations. Robot designers always attempt
82
to use stiffer structures and tighter tolerances; however, a totally

different perspective on this issue is also possible: If it is the loca-
tion of the end-effector that needs to be controlled, then why not
sense this location directly? Suppose some noncontact sensor is
used to measure the location of the robot arm tip. One possi-
bility is to follow the end-effector with laser beams to determine
its location. In this case, the mechanical structure of the robot
can be made much less rigid and expensive. The resulting lower
weight and inertia will also improve dynamic performance
through higher speed and acceleration.
Parameter substitution: Replacing one design parameter with

another, or with a combination of other parameters, which is
more fundamental.
Parameter Substitution in Total Hip Replacement
Articial hip reconstruction surgery is necessary in the case of

hip joint failure. About 170,000 Total Hip Replacement (THR)
surgeries are performed each year in the United States alone.
In the THR surgery, the surgeon replaces the head of the femur
and the socket part of the pelvis bone. The femur portion of the
surgery involves removing the head of the femur, drilling a deep
hole into the medullary canal, and inserting the stem of the pros-
thesis into it. Historical data suggest that 10% of THRs fail in ten
years and 20% in twenty years.
An important issue faced by the designers of the prosthesis
is stress shielding, the phenomenon of bone loss and reduc-
tion in bone thickness due to the decrease in the load on the
bone. The stiffness of metals is much greater than that of the
bone. For instance, Youngs modulus for titanium is 150 105
psi, while that of bone is 6 105 psi. A solid implant that con-
forms to the drilled hole in the femur has a much greater stiff-
ness than the bone it replaces. When load is applied, the stiff
83
metallic stem deforms very little and therefore carries most of the
load. The bone, on the other hand, experiences very little load
due to its very small deformation. This quick look at the problem
suggests that the critical design parameter is nd a material
whose elastic modulus is identical to that of the bone.
It turns out that such matching of elastic moduli is very dif-
cult. Instead, a new key parameter is stated as follows: match
the stiffness of the implant with that of the bone. By replacing
elastic modulus with stiffness, designers can conceive dif-
ferent ideas because stiffness depends on both the material and
geometry. THR designers employ various structural principles,
such as the use of porous materials, to match the stiffnesses.
Note that matching of some property is a common and funda-
mental design parameter that plays a role almost any time indi-
vidual components are made to work together.
Analogical thinking: The thought process of recalling a famil-

iar conguration and adapting it to a new situation. The con-
gurations are analogous because they realize the same design
concept, or parameter. For example, airbags have been
recalled from the automotive eld and adapted for cushioning
the landing impact on Mars. Analogies can also occur at a
deeper, more conceptual level. For instance, stress concentra-
tions are often reduced by understanding the stress eld using
ow analogies. An analogy to electrical fuses, based on their
need to react faster than the event against which they guard,
may be useful in automotive airbag design. The role of analo-
gies is demonstrated in the following example.
Analogies with Aluminum Foil
Aluminum foil used for cooking has a shiny side and a dull side.
For heating food, the shiny side should be placed toward the
food, so it reects radiant heat back to the food while the dull
84
side absorbs the external heat. This conguration thus creates a

bias in the radiation heat transfer. The same concept has been
used in the design of chocolate wrappers and car sun shades,
where the glossy outside reects heat away to keep the choco-
late or the interior of the car cool.
The design of space suits poses an interesting challenge.
Without an appropriate space suit, an astronauts body heats
up due to solar radiation when facing the sun and loses heat
when in the shade. The temperature uctuations could reach
400F. In this particular application, the critical parameter is
block the solar radiant heat from entering the space suit while
reecting the body heat back. In other words, the design must
integrate the features of chocolate wrappers to keep the exter-
nal heat away with the features of cooking foil to retain internal
heat. NASA uses a multilayered, double-shiny foil to achieve
this task. The same principle is used for thermal shielding of
spacecraft components and for building insulation.
Every cycle of parameter analysis (Fig. 5.1) will always have a

parameter identication step. Even when it seems that a congura-
tion can be changed by mentally staying in conguration space,
this is never really the case. When the designer examines a congu-
ration and realizes that something needs to be modied, he or she
has actually performed an evaluation and then identied a new crit-
ical issuea parameterthat now becomes the conceptual basis for
the congurational change.
Although good identication of dominant parameters is the key
to successful conceptual design, not every single parameter identied
in the course of developing a design must be a breakthrough param-
eter. Some parameters are in fact quite mundane. Nevertheless, their
presence as a component of the parameter analysis process provides
the necessary conceptual justication for any physical aspect of the
evolving design.
Theoretically, the most signicant conceptual issuesthe domi-
nating parametersare identied early in the design process, and as
we proceed downstream we encounter and handle the more routine
85
issues. However, design is rarely such a linear process, and it often

takes a large effort and many cycles of parameter analysis before the
real dominant issues and concepts emerge. In this sense, the intellec-
tual effort during parameter analysis, including failed attempts and
taking wrong paths, can be regarded as a stimulant and contributor
to the eventual success in identifying the real issues of the design task.
5.4 Creative Synthesis
The new design concept, stated in previously identied parameters,

becomes the focal point for generating design congurations during
the creative synthesis step. A design artifact is proposed here to solve,
satisfy, or embody the conceptual parameters. Sketches and back-
of-the-envelope calculations are used extensively during this step to
provide enough detail about the design and its operation.
A conguration will usually be represented as a sketch and
should always be quantied. By this we mean that the sketch usually
contains some dimensions, perhaps material selections, and other
details. This will ensure that analysis can be applied and the cong-
uration can be evaluated for its extent of meeting the design require-
ments (which, of course, are quantitative, too).
Creative synthesis may be a highly divergent step in the sense
that many congurations can be proposed at any given time to sat-
isfy the same set of parameters. In this case, the next evaluation step
will be used to select among these congurations. Alternatively, if
only one conguration is proposed to embody the concept, then the
next evaluation will determine its viability, and, according to the
results, another parameter may be identied, to be followed by a dif-
ferent or modied conguration.
Skyscraper Congurations to Resist Wind Loads
Fazlur Khan designed two of the tallest buildings in the world

the John Hancock Center and the Sears Tower, both in Chicago.
In the design of high-rise buildings, one of the design challenges
86
is handling lateral wind loads. Gustave Eiffel addressed this

issue in the design of the Eiffel Tower by tapering it to reduce
the wind loads on the structure. In the John Hancock Building
(schematic shown in Fig. 5.4a), Khan used a gentle taper to
(a) (b)
Figure 5.4 Two skyscraper congurations to resist lateral wind

loads: (a) The X-braced tube of the John Hancock Center and (b)
the bundled-tube of the Sears Tower.
reduce the wind loads and 18-story tall exterior steel diago-
nalsthe famous X-bracesto carry this load. Thus, the con-
guration of the building is a tube stiffened by the oors. Later,
Khan designed the Sears Tower (Fig. 5.4b) using another inno-
vative tubular system, known as the bundled-tube approach.
Nine steel-framed tubes, each 75 feet square, are interlocked to
form a massive column whose combined thickness can easily
resist wind loads. Both of these rigid-tube congurations elim-
87
inated much of the internal bracing that could eat up costly oor
space.
As the word synthesis suggests, the creative synthesis step is

concerned with creatively integrating different constituent elements,
or building blocks, into a unied design. In the case of system
design, the building blocks are subsystems and system components.
For product design, they may be the machine and structural com-
ponents. In component design, the constituent elements are materi-
als, geometry, and manufacturing processes. Creative synthesis is
therefore about
Identifying constituent elements: The designer can choose

different building blocks to create the design, as illustrated by
the following example.
Identifying Constituent Elements in Explosives Design
The effectiveness and safety of explosives are measured in terms

of power and sensitivity to detonation. The power rating com-
pares the volume of gases produced by the explosive to that of
picric acid, which is given a rating of 100. The higher this num-
ber, the more powerful is the explosive. Sensitivity to detonation
is the tendency to explode when a force is applied. Picric acid,
with a rating of 100, is the standard of comparison for sensi-
tivity, too. Explosives with a sensitivity rating less than 100 are
considered unstable and unsafe.
Nitroglycerine is a powerful explosive that releases 12,000
times its own volume in gases when exploding. It has a high
power rating of 160 but also a sensitivity rating of 13. This
means that it is extremely sensitive and may explode even by a
mere shake. No one had been able to harness the potential of
nitroglycerine because its sensitivity made the handling and
transportation extremely dangerous. To allow practical use of
88
nitroglycerine, its sensitivity would have to be suppressed while

retaining its explosive power
Alfred Nobel succeeded in solving the problem by identify-
ing another constituent elementkieselguhr, a very porous
diatomaceous earth. In dynamite, Nobel used kieselguhr to
absorb nitroglycerine without reacting with it, thus retaining the
explosive power while reducing sensitivity. The resulting plastic
explosive was powerful and safe to handle. Dynamite has since
been used as a safe and effective explosive for military, con-
struction, and demolition applications.
Integrating constituent elements: The designer can combine

the building blocks in different arrangements, both spatial
and temporal. Even though the design may use the same com-
ponents, the level of innovation and the overall performance
depend on their arrangement. For instance, the parts of the
Sony Walkman were designed to t together from one direc-
tion, allowing the use of efcient robotic assembly. The fol-
lowing examples describe how constituent elements (geometry,
material, and manufacturing process) are integrated in a sin-
gle component.
Foil Manufacturing
The functional need to create a bias in radiation heat transfer

can be addressed by a foil with shiny and dull sides, as
described in the previous section. To manufacture this unique
combination of features, designers conceived the process of
pack rolling, where two foils are rolled together. The sides of the
foils facing the rollers become glossy, and the sides facing each
other become dull. The materialaluminumis an excellent
choice because it is very ductile and can be formed into very
thin sheets (56 mil). Thus, the foil conguration integrates
geometry (two different sides), manufacturing process (pack
rolling), and material (aluminum).
89
The Shape of Manhole Covers
The circular shape of manhole covers addresses several

design issues. A circular cover will not accidentally fall into its
hole, whereas all other shapes might, if given the right orien-
tation. A circular cover is easy to move by rolling it. It is also
stronger than other shapes because the center point is equi-
distant from the periphery, where it is supported. Such syner-
gistic integration during creative synthesis is one of the keys to
innovation.
Why Do Coins Have Rims?
Two main considerations in the design of coins are allow stack-

ability and prevent wear of the faces. These two issues are
addressed by incorporating a rim at the edge of every coin con-
figuration. The rim is higher than the face, so coins can be stacked
easily. Wear is also prevented because the coins do not touch
each other at the face.
5.5 Evaluation
This third step in parameter analysis serves two purposes: assess-

ing the design performance and providing additional design
information. The latest configuration is examined in this step in
light of the conceptual goal and overall design requirements.
Weaknesses are highlighted to help identify the next design
parameter. Three distinct evaluation methods are: (1) compari-
son of the configuration to the original requirements; (2) identi-
fication of weaknesses; and (3) comparison between several
configurations. These evaluation methods are demonstrated by
the following examples.
90
Evaluation by Comparison to the Requirements:

Battery-Powered Cars
Californias Zero Emission Vehicle (ZEV) regulations stipulate

how many cars sold in California should be ZEVs in the coming
years. Recognizing that other states and countries would soon
follow a similar course, the auto industry is focusing on devel-
oping ZEVs in general and electric battery-powered cars in par-
ticular. One of the performance criteria in using battery
technology is specic energythe amount of energy that can
be stored per unit mass for a specied rate of discharge.
Gasolines specic energy is 12,000 Wh/kg, and this gure
establishes a convenient yardstick in the design requirements for
electric vehicles.
An evaluation of existing battery technology reveals that the
theoretical maximum specic energy for lead-acid batteries is
about 170 Wh/kg, with current technology being around
3540 Wh/kg. Current lithium-ion battery technology can pro-
vide about 120 Wh/kg. Table 5.1 summarizes the characteris-
tics of three current battery technologies.
Table 5.1 Comparison among some automotive battery technologies.
Nickel-Metal
Hydride Lithium Ion
Lead-Acid (NiMH) (Li-Ion)
Advantages The rst commercially Clean, reliable, Very light-
available and mainte- weight.
rechargeable nance-free.
battery. Range of up to
Cheap to mass 150 miles per
produce. charge.
Low maintenance. Expected to last the
life of the vehicle.
Disadvantages Moderate range. More expensive More expen-
Cold weather than lead-acid sive than
reduces storage batteries. NiMH
capacity. batteries.
91
Evaluation by Identifying Weaknesses:

Fuel-Cell Powered Cars
Fuel cells are an alternative to batteries for powering cars, and

automakers are experimenting with them. A fuel cell combines
hydrogen and oxygen into water, as shown in the diagram of
Fig. 5.5. In the process, it produces heat and electricity. The cat-
alyst helps hydrogen to separate into a proton and an electron.
The electrons constitute the electric current, while the protons
pass through the electrolyte and combine with oxygen to form
water.
The hydrogen can be either stored on-board in a com-
pressed state, in an absorbed state in metal hydride, or
obtained from methanol. Fuel cells that use methanol have a
Figure 5.5 Hydrogen and oxygen combine in a fuel cell to produce

electricity and water.
high specic energy of 2000 Wh/kg. Fuel cells with Proton

Exchange Membrane (PEM) electrolytes are being developed
92
for cars. They are very expensive because they use platinum.
New technological breakthroughs, which will reduce the
amount of platinum needed, are required before this technology
can be used commercially.
Evaluation by Comparing Several Congurations:

Hybrid-Powered Cars
As the name suggests, a hybrid vehicle has two energy sources:

an internal combustion engine and an electric battery. These
can be connected in parallel or series, as shown in Fig. 5.6.
In the parallel conguration, the power from the engine is
combined with the power from the electric motor. Maximum
Motor/
Batteries
Transmission
Controller
system
Fuel Engine
(a)
Fuel Engine Generator
Motor/ Transmission
Batteries Controller system
(b)
Figure 5.6 (a) A parallel arrangement and (b) a series arrange-

ment for the power train of a hybrid vehicle.
93
power is therefore high, and a generator is not needed.

However, this arrangement requires the difcult job of joining
the torque from two power sources. On the other hand, the
series conguration reduces emissions by eliminating the idling
of the engine and greatly simplies the transmission system.
Both these arrangements minimize energy losses and emissions,
and reduce the fuel consumption by 50% for the same power
output.
To determine how well a design measures up to its expected per-

formance, it is necessary to evaluate the conguration quantita-
tively. This requires that the conguration contains enough detail,
such as materials information and dimensions, and that approxi-
mate calculations be carried out. In addition to gaining more insight
into the design task, the evaluation step helps to determine whether
(and how) the current design concept can be improved to meet the
requirements, or there should be a more radical conceptual change.
The following example shows how life-cycle concernsmanufac-
turing in this caseneed to be considered during the evaluation
stage in addition to functional issues.
Evaluation of Connecting-Rod Manufacturing Details
Connecting rods are forged to their nal shape. In the forging

process, a small amount of ash extends at the parting line. This
ash is removed by grinding. The two methods of grinding are
represented by the two grinding wheels in Fig. 5.7. In the rst
method, the grinding wheel rotates in the plane of the connect-
ing rod, whereas in the second method the grinding wheel is per-
pendicular to this plane. Although both methods seem equally
good in terms of removing ash, the second method creates
transverse surface cracks, which may lead to what is known in
fracture mechanics as Failure Mode I, which is very dangerous.
94
In-plane
grinding wheel
Flash
Transverse
grinding wheel
Figure 5.7 The two grinding wheels represent two methods of ash
removal from a forged part. The top one is preferable from the frac-
ture mechanics viewpoint.
The evaluation is by no means a mere ltering mechanism. The

main purpose is not nding fault but, rather, constructive criticism.
A well-balanced observation of both good and bad aspects of the
design is crucial to pointing out possible venues of improvement in
the next design cycle. In terms of the designers thought process, the
evaluation step belongs in both conguration and concept spaces. It
involves an assessment of the latest embodiment of the design
(clearly, a member of conguration space) in light of the design
parameters (the conceptual goals and critical issues that belong in
concept space).
5.6 Parameter Analysis Revisited
The solution to a design task consists of two components: a physical

conguration and a concept, that is, the operational principles
behind the conguration. Using parameter analysis, we nd that the
design solution evolves through multiple cycles of triplets: parame-
ter identication to generate a concept, creative synthesis to embody
the concept in virtual hardware, and evaluation to show the direc-
tion for the next cycle.
95
If we were to rank the three repeating steps in parameter analysis

according to signicance in the overall conceptual design process, then
parameter identication would turn out to be the most important, fol-
lowed by evaluation, and creative synthesis as the least inuential. The
reason for this ranking is that the ability to conceptualize novel ideas
and new ways of looking at problems is the key to true innovation.
Evaluation is the driving force behind the identication of new critical
issues and concepts, while creative synthesis, resulting in a new cong-
uration, is merely an embodiment of previously generated ideas. If a
bad conguration is generated in creative synthesis, its shortcomings
will easily be pointed by the next evaluation, and a better concept will
emerge in parameter identication. On the other hand, a bad concept
cannot be xed by good hardware design, no matter how skilled the
designer is while working in conguration space.
The cyclic character of parameter analysis, that is, repeatedly
rening the design through cycles of analysis and synthesis, com-
bines two important characteristics of any design process: divergent
and convergent thinking. The mental processes in concept space (see
Fig. 4.9), namely, parameter identication and part of evaluation,
are convergent in the sense that they focus the design progression by
narrowing down the choices. On the other hand, activities in con-
guration spacecreative synthesis and the other part of evalua-
tiontend to be more divergent. When we are at the parameter
identication step, we concentrate on one or a few critical issues,
and we set aside the rest. Then, in creative synthesis, we may exam-
ine many ways of realizing that idea, but we use the evaluation step
to perform an abstraction that focuses us again on the most critical
parameters. This repeated combination of divergent and convergent
thinking is depicted in Fig. 5.8.
While Fig. 5.8 may suggest that parameter analysis is a very lin-
ear process, with a distinct beginning and ending, this is usually not
the case. Recall that the process generally begins by selecting a single
technology from several possible ones recognized during the tech-
nology identication stage and by using that technology as a start-
ing point. It may well happen that a long parameter analysis process
will culminate in a dead-end, meaning that at some point it becomes
apparent that the fundamental concepts are erroneous or that the
96
Divergent Convergent
thinking Configuration 1 thinking Configuration m+1
Configuration 2 Configuration m+2
Configuration 3 Configuration m+3

Parameter 1 Parameter 2
. .
. .
. .
Configuration m Configuration n
Figure 5.8 Going through the steps of parameter analysis corresponds to

alternating between divergent and convergent thinking modes.
concepts are good but their technological realization is infeasible. A

characteristic of a good designer is his or her ability to recognize
such situations and instead of spending more effort and resources,
backtrack and start over. The designer may use a different technol-
ogy for the second attempt at conceptual design, or a less radical
approach by choosing a different conguration among several pos-
sibilities at any earlier stage of parameter analysis. The many paths
from the initial technology identication to nal solutions are
shown schematically in Fig. 5.9.
Suppose point A is the starting point for conceptual design.
Several technologies are identied here as feasible approaches to the
design task. The designer chooses one of these technologies and ini-
tiates the parameter analysis process along path I in Fig. 5.9.
(Parameter analysis paths are shown as curly lines to symbolize the
many cycles of parameter identication, creative synthesis, and eval-
uation, which constitute each path.) The process reaches a dead-end
at point B, at which the designer decides to attempt a different con-
guration instead of the one used earlier in the process, say, at point
C. The designer then discards everything that took place between
points C and B, and moves in a new path (II ) from C to D, where a
satisfactory solution is reached.
Path III in Fig. 5.9 shows another situation. A dead-end is
reached at E, and the designer decides to try a totally different
approach, or technology, for the conceptual design. Backtracking to
the starting point (A) takes place, another technology is chosen, and
a new path (IV ) is taken until a new solution is created at point F.
97
Backtracking
Dead-end
III E
A
I Backtracking
C
Dead-end
IV
B
II
Solution Solution
F D
Figure 5.9 A schematic of the branching character of parameter analysis.
The decisions described here, namely, to backtrack partially but

stay on the same path versus complete backtracking and starting
over, are not easy to make. On the one hand, a good concept should
not be abandoned too early in the face of difculty; rather, such dif-
culties are often the driving forces that lead to valuable new
insights in concept space. On the other hand, there is not much
point in insisting on developing a fundamentally bad design, espe-
cially when realistic time constraints and limited resources are pres-
ent. In the end, it is up to the skillful designer to make the correct
decision. The following example demonstrates some of these issues.
Backtracking in Magnetic Seal Design
Centrifugal pumps used with chemicals usually employ a

dynamic seal between the shaft and the casing. This seal is
often a major source of leakage. In an attempt to overcome this
problem, engineers came up with a novel idea: transmit the
motor torque by a set of permanent magnets placed across a
98
physical barrier, thus eliminating the dynamic seal altogether.

Let us follow several steps in the parameter analysis process,
starting with this idea:
PI: Use magnets across a physical barrier to transmit torque
while sealing.
CS: This idea is realized in magnetic drives, shown schemat-
ically in Fig. 5.10. Permanent magnets are installed on
both the motor and the pump shafts and are separated by
a static barrier. The barrier needs to resist corrosion from
the pumped chemicals and withstand high tensile stresses
(from the uid pressure), so it is made of steel.
E: A major side effect of metallic barriers is the generation
of eddy currents due to the relative motion between the
conducting barrier and the magnets, and the electrical
resistivity of the barrier material. This causes hysteresis
losses that could be greater than 50% of the total power
transmitted.
PI: Eliminate eddy current losses by using a nonmetallic barrier.
CS: Use a ceramic barrier.
Barrier
Pump
Motor
shaft Pump
shaft
Motor
Permanent
magnets
Figure 5.10 Schematic of a typical magnetic drive.
99
E: The uid pressure in the pump causes tensile stresses (cir-

cumferential and longitudinal) in the barrier, and ceramics
are not good at carrying tensile loads.
PI: There are at least two choices now. One is to pursue the
ceramic barrier solution further, which may involve intro-
ducing ber reinforcement to carry tensile loads. The sec-
ond is to look at alternative methods for reducing eddy
currents. The following insights can be gained from study-
ing the underlying physics:
Eddy-current losses are
directly proportional to the square of the speed;

directly proportional to the square of the radius of the drive;
directly proportional to the square of the ux density; and
inversely proportional to the resistivity of the barrier
material.
To improve cost effectiveness, the amount of magnetic

material required can be reduced by decreasing the distance
between the two magnets. However, the increase in flux den-
sity between the magnets will increase the eddy current losses
and result in a lower efciency. The diameter of the drive can
also be reduced. However, both approaches can improve the
efciency only to a certain level. Fundamentally, the eddy cur-
rents would still persist.
We can look at other situations where designers deal with
eddy current losses, such as electric transformers and motors.
The transformer core and the motor rotor use laminar con-
struction to minimize eddy currents. Eddy currents decrease
linearly with the number of layers because the flux driving
them in each layer is reduced. The I2R losses decrease with the
square of the number of layers. If the barrier is made of two
pieces, the eddy current losses will decrease to 25%. Four
pieces will decrease the losses to 6.25%. The new parameter
is therefore: decrease the eddy currents by using a layered
construction.
100
Figure 5.11 A seal design that signicantly reduces eddy currents

while adequately carrying the pressure-induced loads. (Courtesy of
Nova Magnetics Limited.)
CS: Nova Magnetics Limited, a Canadian company, came up

with the innovative solution shown in Fig. 5.11, which
uses laminar construction. The inner shell consists of rings
that enclose the fluid and carry the pressure-induced cir-
cumferential load. Because the thickness of each ring is
small, the eddy current losses in this design are negligi-
ble. High-pressure gaskets are provided between the
rings to prevent leakage of chemicals. The outer shell car-
ries the longitudinal load due to the fluid pressure, and
the rings and gaskets are live-loaded and held in place
by a large wave spring and retaining ring. This outer shell
is also slotted longitudinally to help minimize eddy current
losses in it. In addition, the magnets are assembled in
such a way as to help reduce eddy current losses. The
design uses an alternating magnet pole configuration
down the length of the coupling. This causes the direction
of the eddy current generated by each magnet set to
reverse when compared to the adjacent magnet set.
Following Kirchhoffs Laws, this effectively cancels the
movement of eddy currents from one magnet area to the
other, thereby reducing resistive losses.
101
Technology Stripping of less important issues

Identification Simplification Focusing on make-or-break issues
Conceptual Developing simple models
Design Parameter
Identification Rephrasing the task statement
Parameter (PI)
Analysis Explaining causes differently
Transformation "Stepping back"
Substituting parameters
Thinking analogically
Creative Identifying constituent elements
Synthesis
(CS) Integrating constituent elements
Comparing to the requirements

Evaluation
(E) Identifying weaknesses
Comparing among configurations
Figure 5.12 Summary of the activities in conceptual design.
Conceptual design was shown to be composed of two stages: tech-

nology identication and parameter analysis. Core technical con-
cepts that may be used for the design task at hand are listed rst, and
the one with the highest success potential, in the designers opinion,
is chosen as a starting point for parameter analysis. Parameter analy-
sis consists of repeatedly cycling through three distinct steps:
parameter identication, where concepts and ideas are generated;
creative synthesis, where a conguration is devised to embody and
realize conceptual insights; and evaluation, where the design is
assessed constructively. Many activities and mental mechanisms that
are associated with these steps were described in the chapter and are
summarized in Fig. 5.12.
A basic question associated with parameter analysis is: how do we
know that we are done? In other words, at what point should param-
eter analysis stop and the current conguration be declared a nished
conceptual design? The answer is quite simple to state: parameter
analysis should continue as long as there is doubt about the success of
the conguration in meeting all the design requirements. Parameter
analysis can be terminated only when no unresolved issues remain
and all foreseeable questions regarding the operation of the design
have been answered. The evaluation step is the point at which such
determination is made, so the last step in parameter analysis is usually
102
evaluation. This guideline implies that conceptual designs developed

by parameter analysis are much more detailed than what many prac-
tices and textbooks call conceptual design. The latter usually consists
of a very general idea or just a rough sketch, as opposed to the well-
developed conguration of the former.
1. Suppose that your task is to design paper currency for use by

blind people. Perform technology identication, listing as many
core technologies as possible to help the visually impaired sense
the value of bills. Critically assess each technology for its poten-
tial to develop into a viable solution. Keep in mind economical
and usage constraints. For example, making notes of different
sizes may be expensive, and the effect of impregnating them with
different odors may not last very long or may be ineffective due
to contamination by other odors.
2. Perform technology identication for the task of providing light-

ing in undersea applications. Remotely operated vehicles used in
offshore oil drilling need the lighting to provide clear video images.
Manned deep sea submarines used in rescue missions need to
provide adequate visibility to their crew.
3. Identify the underlying physics, which may be the critical param-

eters, in the design of the following devices:
(a) Different types of positive displacement pumps
(b) Heat pipes
(c) Piezoelectric actuators
4. Two important performance criteria with ink-jet printers are res-

olution and speed. Speed is usually measured as pages per
minute, or ppm. Investigate the physics of injecting small ink
droplets to understand what might limit the maximum operating
speed of single-head printers.
103
5. Several generic types of bridges are in common use. Study the con-
gurations of beam, arch, suspension, cantilever, and cable-stayed
bridges. How does each conguration handle vertical loads, such as
the weight of cars or trains crossing the bridge?
6. Most coins are round. Write some of the design requirements for
coins, and evaluate round, square, and hexagonal coins against
the requirements and against each other.
7. Some applications require a spring that applies a small but uni-

form force over a large displacement. A good example is the
spring supporting the tailgate lid in hatchback cars. Identify dif-
ferent congurations that can be used for this spring design.
8. Different vacuum cleaning technologies are currently in use.

Identify the technologies and discuss the advantages and disad-
vantages of each one.
104
6
Conceptual Design Case Study:
HVAC Airflow Sensor
As the rst detailed case study of conceptual design using the

parameter analysis methodology, this chapter presents a relatively
simple design example. Only a partial need analysis, highlighting the
most important issues, is given here. The discussion under the value
category uncovers the importance of keeping the cost down, and
this issue also proves to be the dominant parameter in the develop-
ment of the concept. This case study demonstrates the sequential
evolution of the design throughout the parameter analysis process.
Heating, ventilating, and air conditioning (HVAC) systems in build-

ings consist of a heating source, a cooling source, a fan to force air-
ow, and a closed-loop temperature control system. The last-named
component usually includes a computer to compare actual tempera-
ture readings to set, or desired, values and activate the system accord-
ingly. Such a system can handle the temperature well, but it does not
account for the freshness of the indoor air. If conditions happen to
be such that the temperature does not need to be actively changed by
the system for a relatively long period of time, the air may become
uncomfortably stale. The need is therefore to design a sensor to
measure the airow in HVAC ducts to allow the control system to
monitor the number of air-volume exchanges in the rooms.
105
6.2 Abbreviated Need Analysis
Performance
The sensor needs to tell the computer how much volume of air
per unit time passes through the duct. An obvious way would be
to measure the flow velocity in the duct and to let the computer
multiply the reading by the cross-sectional area of the duct to pro-
duce a volume-per-time reading. Alternatively, the sensor could be
calibrated to suit the intended duct during manufacture or instal-
lation. In the latter case, readings of volume-per-time can be pro-
vided directly. The question then is, what range of velocities
should the sensor be capable of measuring? The answer can be
found by looking up information for typical HVAC systems,
revealing that flow velocities of 50 to 1000 ft/min cover a wide
range of duct sizes.
Every sensor has some properties associated with its measure-
ment. Accuracy, defined as the deviation of the reading from the
actual quantity being measured, is the first that comes to mind.
For the HVAC sensor, it does not seem that a very high absolute
accuracy is needed in general. Without presenting a detailed dis-
cussion of error analysis (this can be found in many books on
measurement), we set the desired sensor accuracy to 3% of full
scale, or 30 ft/min.
Linearity is another property that may be relevant. Linearity was
important in the precomputer days, when Laplace Transforms were
used extensively in the design of control systems. However, a com-
puter can easily accommodate a nonlinear output from a sensor, so
the potentially demanding requirement of linearity should not be
posed for the HVAC sensor.
Next, we examine the environment in which the sensor func-
tions. HVAC ducts typically accumulate considerable amounts of
dirt and dust, and the functioning of the sensor should not deterio-
rate noticeably over time. It is difcult to quantify this item with
regard to type and amount of contamination; however, the designer
should be aware of this issue. If the sensor is relatively sensitive to
the effects of contamination, more frequent maintenance will be
required. This matter is also addressed in the next section.
106
Conceptual Design Case Study: HVAC Airow Sensor
Value
In order to establish target gures for manufacturing, installation,
operating, and maintenance costs of the sensors, we need to fully under-
stand the savings that may be realized by using the airow measure-
ment scheme. Considering a temperature-control-only HVAC system
may be misleading because such a system does not address the ques-
tion of air freshness at all. The real contender is an open-loop system,
which provides a change of air volume every preset time. This overly
simplistic method would probably result in too many air changes with
the accompanied waste of energy. (Energy is required to force the new
volume of air through the system, and even more energy may be con-
sumed in bringing that volume of air to the desired temperature.)
A study of many ofce buildings, which typically have hundreds
of sensor-mounting locations, would show that the sensors must be
very inexpensive, selling for about $10 to $15 each. Manufactured
cost therefore must only be at most a few dollars. Implied by this cost
ceiling is the realization that electronics would probably have to be
shared: instead of a signal-processing unit in each sensor, the signals
would be sent to a central computer. Keeping the overall cost down
also entails easy installation and low maintenance requirements.
Climbing into HVAC ducts routinely to clean sensors should be
avoided, so a requirement of at least ve years of maintenance-free
operation is generated.
Size and Safety

We shall not elaborate on these categories because the needs are
quite obvious. The sensor should t within existing ducts without
disturbing the normal operation of the system, and it should com-
ply with all building safety codes. The sensor should not in any way
reduce the quality of air circulating in the system. Moreover, exam-
ining layout drawings of typical HVAC systems shows that air is usu-
ally fed to rooms and ofces through horizontal or vertical ducts, so
the sensor should function in both types of ducts.
The need analysis will be concluded at this point, although it is

still incomplete. Other relevant issues remain to be identied and
studied. For example, we did not describe the requirements on the
107
signal generated by the sensor or the constraints regarding the phys-

ical installation in the ducts. Some of the issues mentioned, such as
size and compliance with building codes, need to be quantied. The
need analysis presented so far is summarized in the following
requirements list, which also contains some hypothetical numbers.
6.3 Some Design Requirements
1. Measurement of airow velocities in the range of 50 to 1000

ft/min.
2. Measurement accuracy: 30 ft/min.
3. Maximum manufacturing cost: $3 at tens of thousands of
units per year.
4. Maintenance-free continuous operation for at least ve years.
5. Minimum life: twenty years.
6. Installation in horizontal or vertical ducts.
Many more requirements, covering size, weight, installation,

and other aspects, would be listed here if a complete need analysis
were performed. Note again that the required manufacturing cost is
considerably lower than the target selling price.
Looking at existing ways of measuring air velocity and ow rate may

provide considerable insight into the physics of the problem and
establish various starting points for parameter analysis. Here are
several existing devices and methods for measuring ow velocity,
presented in no particular order:
Hot-Wire and Thin-Film Anemometers

The basic physics is that of convective heat transfer and electri-
cal-resistive energy dissipation. An electric current through a wire or
lm placed in the ow is controlled to keep the temperature con-
stant. The faster the ow, the more cooling action is applied to the
108
conductor so that more current is required to produce resistive heat-

ing. The current change thus corresponds to the ow velocity. A
variation of this implementation is to apply constant current and
measure temperature changes.
Two well-known properties of anemometers are their fragility and
sensitivity to contamination. The latter weakness is recognized as a
potential problem in light of the previous need analysis. Although we
could still apply parameter analysis to this physical principle and
attempt to overcome the difculties associated with dust accumula-
tion, we may conclude that such a device would probably turn out to be
too expensive and we may decide not to develop this concept further.
Pitot-Static Tube
The basic physics is that of measuring the difference between
the static and stagnation (or total) pressures of the ow to deduce its
velocity. At rst glance, it seems that the need to utilize one or more
pressure gages in the design would render it too expensive.
Venturi, Flow Nozzle, and Orice Meters

Like the Pitot Tube, this is also a p device. A constriction accel-
erates the ow so that potential energy is converted to kinetic
energy. The velocity is deduced by measuring the pressure difference
between inlet and outlet.
Rotameter
This is a common device used as a ow meter. A oat sus-
pended in a vertical tapered tube locates itself at the point where
the drag force produced by the upward ow balances the weight. The
drag force is proportional to the flow velocity, so the location of
the oat can be calibrated to indicate this velocity or can directly
provide readings of the ow rate.
Turbines and Propellers

The physical principle is utilization of aerodynamic lift and drag
forces that the ow exerts on the rotor blades to produce rotational
speed, which is proportional to the ow velocity and hence to ow
rate. Several methods are available to monitor the rotational speed
109
of the turbine. For example, a permanent magnet and coil can be

mounted so that an ac voltage is induced in the coil as the blade
passes through the eld. The voltage frequency will be proportional
to the ow rate.
Particle Marking
This technique is used extensively in wind tunnels to visualize
ow. Momentum is transferred from the uid to particles through
drag force. The ow velocity can be measured by timing the motion
of a particle between two locations. Two problems associated with
applying this technique to the HVAC sensor immediately stand out:
it may be difcult and expensive to trace the motion of released par-
ticles, and they may contaminate the air fed to the rooms.
Furthermore, an appropriate supply of particles is required.
Doppler Effect
An acoustic or optical (laser) device could send a wave of sound
or light through the ow and measure the frequency shift in the
reected signal. This shift in frequency correlates with the velocity of
the ow. Any device utilizing this principle would undoubtedly cost
more than $3 to make.
Drag-Force Flow Meter

This is another aerodynamic drag device. The ow exerts a drag
force proportional to the square of the velocity. Measuring the drag
force can be used to infer the velocity.
These technologies offer a variety of initial conditions for

parameter analysis. For conceptual design purposes, the most
important aspect of such a listing is understanding the underlying
physical principles associated with each technique or device. We
now need to scan the list and select the most promising candidates.
As often happens in design, this is not a unique process. Different
solutions can be derived from different initial conditions. For the
present discussion, we will use the last technique, a drag-force
device, because its simplicity promises a better chance of satisfying
the cost constraint.
110
6.5 Parameter Analysis

PI: Immerse an object in the ow and measure the drag force
exerted on it.
CS: A circular aluminum disk, 2-in. in diameter, is a drag-produc-
ing body. This disk is rigidly connected to the top surface of the
duct through a at arm. The drag force can be measured by
attaching a strain gage force-measuring transducer, as shown in
Fig. 6.1. (Note that the dimensions and material are subject to
change later. What is important now is to have a specic con-
guration to evaluate.)
E: The drag force F is given by
F = (1/2) ACD 2 (6.1)
where is the density of air, A is the frontal area, CD is the drag

coefcient, and is the ow velocity. For a disk normal to the
ow, CD is about 1.1. The density of air is = 0.075 lbm/ft3.
Neglecting the area of the arm, we calculate the area of the 2-
in. disk to be A = 0.0218 ft2. This gives the following relation
between the drag force in lbf and the ow velocity in ft/s:
F 2.8 105 2
For the desired measurement range of 50 to 1000 ft/min, this

relation roughly translates into forces in the range 2 105 to
0.008 lbf. (Note how a 20:1 velocity range translates into a 400:1
range of forces in a drag-force device.)
To check whether these forces can be measured by the
strain gage, we perform a simple analysis of the cantilever beam
with the drag force applied at the free end. This shows that the
strains range from 0.1 to about 40 . These are relatively small
strains, so we may consider changes in the design to increase
the sensitivity. However, a more serious problem is that strain
gages are fragile, sensitive to contamination, and probably too
expensive for this design.
PI: Force cannot be measured directly, only through deection.
Strain gages measure very small deections. A large deection
111
1/2 "
Strain gage
1/32 "
2-in.
4" dia.
Flow
Figure 6.1 The drag force exerted by the ow can be measured by a strain
gage as deection.
may be easier and cheaper to measure. A pendulum can serve

as a large-deection, drag-force device.
CS: A rectangular 6-in. by 1-in., 1/32-in.-gage aluminum plate
hinged from the top can constitute the pendulum, as shown in
Fig. 6.2. The relationship between the deection of the pendu-
lum, , and the ow velocity, , should be analyzed, beginning
with the drag-force calculation according to Eq. (6.1). The rec-
tangular plate, like the disk, has an approximate drag coef-
cient of CD = 1.1. The area of the 6 1 plate is A = 0.0417
ft2. This gives the drag force in lbf as a function of the ow
velocity in ft/s:
F 5 105 2
For the desired measurement range of 50 to 1000 ft/min, this

relation translates into forces in the range 3.5 105 to 0.014 lbf.
To nd the relation between the drag force and the pendu-
lum deection angle, , we need to analyze the static equilib-
rium condition of the pendulum, as in Fig. 6.3. Taking
moments about the hinge, we nd:
W l sin = F l cos
112
1/32" v
Flow
6"
1"
Figure 6.2 A pendulum tilts in response to the drag force exerted by the ow.
Figure 6.3 Drag, F, and weight, W, are the forces acting on the pendulum
in equilibrium.
= tan1 (F/W)
The weight of the plate, W, is estimated from the weight density

of aluminum (165 lbf/ft3) and volume (6 1 1/32 = 0.1875
in3 = 1.09 104 ft3) at 0.018 lbf. Thus, for ow velocities
between 50 and 1000 ft/min and the corresponding drag forces
of 3.5 105 and 0.014 lbf found previously, the inclination angle
of the pendulum will be between min = 0.1 and max = 37.
113
E: As expected from a drag-force device, the range of angles to be

measured is quite wide. Especially troublesome is the need to
measure 0.1 cheaply, and under the dirty conditions inside the
ducts. Note also that the above calculation is approximate
because it assumes that the drag coefcient and frontal area do
not change with inclination. This degree of accuracy in calcu-
lations is sufcient at this stage. However, recognizing that both
drag coefcient and area actually decrease at higher angles and
ow velocities is important. This understanding leads to iden-
tifying the next parameter, or critical issue.
PI: Increase sensitivity at the low end only. Examining Eq. (6.1),
which governs the behavior of the pendulum, and in light of
the last evaluation step, we realize that the problem of the drag
force increasing with the square of the velocity may be circum-
vented by decreasing other variables in the equation at the same
time. The density is constant, but the drag coefcient CD and
frontal area A are sensitive to the angle of the plate, which in
turn varies with the ow velocity. Thus, in order to increase
sensitivity at the low end only, we need to amplify the effect of
the inclination angle by exposing a large area to the ow at low
velocities and a much smaller area at high velocities.
CS: A conguration that exposes different areas at low and high
velocities may use two pendulums as shown in Fig. 6.4. Note
that the downstream plate is split into two so that full exposure
to the ow is obtained.
For simplicity of calculations, let us neglect the drag force
on the plates when they are almost horizontal. Thus, at the low
velocities we make the A2 area large enough to produce a sub-
stantial deection. For example, using the relations developed
earlier, if A2 is set to about 0.2 ft2 (two 7-in. 2-in. plates), the
drag force at v = 50 ft/min will be about 18 105 lbf, which is
ve times more than with the previous rectangular single pen-
dulum. The major improvement, however, is at the high end.
Now area A1 is almost normal to the ow. If we set it to 0.05 ft2
(a 3.5-in. 2-in. plate), its drag force at v = 1000 ft/min will be
about 0.018 lbf, which is only slightly more than with the single
rectangular plate. Static equilibrium analysis is now required to
114
A1
A2
Flow
Zero flow position
Small v Large v
Figure 6.4 The pendulum at the top exposes the large area A2 at low-ow
velocities (lower left) and the small area A1 at high velocities (lower right).
nd the deections. It will show that the device is somewhat

heavy and needs to be made lighter; for example, by reducing the
sheet thickness or making them out of plastic. Once an adequate
combination of material, plate size, and angle between the plates
is found, the design process can proceed with further evaluation.
E: The calculations show that high sensitivity at low velocities and
low sensitivity at high velocities can be combined in the proposed
conguration. However, the device would function well only
when mounted in horizontal ducts. What about vertical ducts?
PI: In the horizontal ducts discussed so far, the pendulum func-
tioned by creating equilibrium between drag and gravitational
moments. For use in vertical ducts, a restoring force other than
gravity should be used.
CS: The conguration in Fig. 6.5 adds a spring, a counterbalancing
mass, and a mechanical stop. At low velocities, the larger A2 area
is almost normal to the ow so the drag force is relatively large,
115
m
Stop
Small v
A1
A2
Stop
A2
Medium v
A1
Stop
A2
Large v
A1
Figure 6.5 A spring, counterbalance, and mechanical stop are added to

facilitate operation in vertical as well as horizontal ducts.
producing large deections. At some medium velocity, the A2

plates reach the end of their sway, which is dened by the stop.
The large drag force that now acts on the smaller plate needs to
overcome the relatively high resistance of the spring.
E: With the correct choice of A1, A2, m, K (the spring constant),
and some additional geometrical characteristics such as arm
lengths and initial angular opening, the device can be made to
function properly in both horizontal and vertical ducts. The
116
next item still missing is provision for measuring the tilt of the
pendulum. In fact, only the deection of A1 needs to be meas-
ured, relative to a xed frame of reference.
PI: Measure the absolute angle of A1 relative to the mounting
bracket of the device. This angle sensor should be very inex-
pensive to keep the overall cost down.
CS: A resistive potentiometer is a very simple, low-cost angle sensor.
E: Potentiometers can easily measure the desired angle; however,
low-cost potentiometers usually have high internal friction. A
potentiometer with low friction would be too expensive. A poten-
tiometer is therefore not a good choice because it will always have
parts that move relative to each other and at the same time need
to provide good electrical contact.
PI: In addition to low cost, low friction turns out to be of primary
importance with the angle sensor. Only a noncontact device
has the potential to cost very little and have no friction.
CS: Use a capacitor-type angle sensor. It is very inexpensive and
almost frictionless.
E: We conclude the development of this design now. Several more
parameter analysis iterations need to be applied before the con-
ceptual design can be considered complete. For example, we
have not yet addressed the problem of minimizing friction in
the hinges, the installation of the sensor inside the ducts, or the
processing of the angle measurement signals.
Figure 6.6 summarizes the parameter analysis process used to design

the airow sensor. None of the identied parameters represents
breakthrough understanding in the problem domain; the designer
was only required to utilize fundamental physical principles.
The process described above is not unique. We could start the
design process with any of the other initial conditions described in
the technology identication section. At any point in the process,
different parameters could be identied in the PI step or other con-
gurations suggested in CS. For example, when we decide on a split
117
Parameters Evaluation Configurations
Measure drag force on an

object immersed in the flow
Aluminum disk with arm
and strain-gage to
Strain-gage fragile measure deflection
and sensitive to
contamination
Increase deflections by
using a pendulum
Rectangular aluminum
Wide range and high plate on a hinge
resolution difficult to
Increase low-end deflection obtain
by controlling and varying
the frontal areas
Double and split pendulum

Functioning in vertical
ducts, not just horizontal
Use a restoring force
other than gravity
Spring, counterbalance, and

mechanical stop added
Tilt measurement
still missing
Incorporate low-cost
angle sensor
Resistive potentiometer added

Low-cost potentiometers
have too much internal
friction
Use low-cost, low-friction
non-contact angle sensor
Capacitor-type angle
sensor added
Finished
Figure 6.6 Summary of the airow sensor parameter analysis process.
double-pendulum to increase low-end sensitivity through manipu-

lation of frontal areas, we could identify other parameters. If we
decide to increase the sensitivity of the pendulum by regarding it as
too stiff a spring (both pendulums and springs deect more in
response to larger forces), we may try to reduce the spring-constant
by lengthening the pendulum, reducing its weight, or counterbal-
ancing it with an added mass. This last conguration is sketched in
Fig. 6.7. The mass creates a moment about the hinge that joins the
118
Figure 6.7 Counterbalancing a pendulum increases its sensitivity as a ow-

velocity sensor.
drag force in producing a relatively large deection, thus increasing

the overall sensitivity. Note, however, that sensitivity at the high end
also increases with this arrangement.
An alternative parameter would be to decrease the sensitivity at
the high velocity while not changing sensitivity at low speeds. Still
having a spring model for the pendulum in our minds, we realize
that a spring-like device may be made stiffer by adding another
spring to it in parallel, as in Fig. 6.8.
Here, a torsion spring was added to produce the following effect:
When there is no ow and the pendulum is vertical, the spring is
unstressed. When the ow is low, the inuence of the spring is hardly
noticeable because small deections produce only small spring forces.
When the ow is high, however, the increased deection magnies the
resistance of the spring, resulting in smaller pendulum deections at
the high end of the velocity range. Now that the overall range of
deections has shrunk, we may want to continue the parameter analy-
sis process to increase overall sensitivity so that the small deections
would not be too difcult to sense.
Another conceptual design path may emerge from the low-
cost, low-friction angle measurement parameter. In the above dis-
cussion, a capacitor-based sensor is selected. But the same
parameter could result in a multitude of other congurations such
as an optical encoder (one or two pairs of LED-type emitters and
119
Figure 6.8 Adding a torsion spring reduces the deection at high velocities
more than at low velocities.
Insulating tube Metal ball
Contacts
Figure 6.9 A proposed design for an angle sensor. The metal ball short-cir-
cuits different pairs of contacts depending on the tilt angle.
receivers placed on opposite sides of a slotted disk), or a magnetic

sensor in which geometrical changes vary the intensity of a mag-
netic eld. Another possibility is to design our own low-cost, low-
friction angle sensor similar to the one shown in Fig. 6.9, where a
metal ball short-circuits pairs of electrical contacts:
1. Modern computer-controlled automotive engines employ an air-

ow sensor to match the quantity of fuel injected with the
120
amount of air. Investigate how these sensors work and whether

the same operation principle can be used for HVAC applications.
2. One ow-velocity sensing device mentioned in the technology

identication section (6.4) is the anemometer. However, we ruled
it out due to sensitivity to contamination. Apply parameter
analysis to the design of a low-cost anemometer, with emphasis
on eliminating this sensitivity.
3. Another method mentioned under technology identication is

the use of a rotameter. The design of these devices for upward
ow measurement is straightforward. Can you design a similar
device for horizontal and downward ows?
4. In the discussion and summary section (6.6), we mentioned sev-

eral alternative solutions to those adopted in the main design
process (Section 6.5). Repeat this process using the alternative
parameters and congurations to generate more airow sensor
designs based on the basic pendulum and drag-force concept.
5. As a way to practice parameter analysis, choose any of the meth-

ods listed under technology identication as a starting point for
a new airow sensor design process.
6. Locate available low-cost angle sensorssuch as potentiometers,

capacitance- and inductance-based, magnetic, and encodersin
commercial catalogs and compare them based on accuracy, dura-
bility, and cost.
6.8 Bibliography
A source of information on HVAC systems is required for a good

need analysis and to formulate the requirements. For example:
ASHRAE HandbookFundamentals. Atlanta: The American
Society of Heating, Refrigerating and Air-Conditioning
Engineers, 1997.
121
McQuiston, F. C., Spitler, J. D., and Parker, L. Heating, Ventilating,

and Air Conditioning: Analysis and Design. 5th ed. New York:
John Wiley & Sons, 2000.
Many books on measuring techniques are available and will be use-

ful to a designer working on sensor design. One of the most com-
prehensive is:
Doebelin, E. O. Measurement Systems: Application and Design.
4th ed. New York: McGraw-Hill, 1990.
A general handbook with good in-depth coverage of many topics is:

Avallone, E. A. and Baumeister, T. (eds.), Marks Standard
Handbook for Mechanical Engineers. 10th ed. New York:
McGraw-Hill, 1996.
Automotive-specic information that can help with thought ques-

tion no. 1 may be found in:
Bosch, Robert. Automotive Handbook. 4th ed. Cambridge, MA:
Bentley Publishers, 1997.
and in books published by the Society of Automotive Engineers

(SAE).
Information about commercial sensors is widely available on the

Internet, in publications such as Sensors magazine, and in catalogs of
companies such as Edmund Scientic Co. and Omega Engineering,
Inc.
122
7
Cut-Edge Sensor for
Flooring Removal
This chapter presents another case study of sensor design. The sen-
sor guides an asbestos-ooring removal machine along the cut edge
of the ooring material once the rst strip has been removed. As
with many sensor design tasks, a multitude of technologies can be
used. Noncontact imaging is selected for this example because of the
harsh operating environment and the need to avoid accidental
release of asbestos bers into the air.
An asbestos-ooring removal machine is shown in Fig. 7.1, whereas

a cross-section through a typical oor is presented in Fig. 7.2. The
machine was designed to assist in removing old ooring material
without releasing harmful asbestos bers into the air. It is equipped
with heating elements that radiate enough heat (about 1500 W) to
raise the temperature of the adhesive to around 100C. Then, the
operator makes a cut in the ooring and removes a 30-cm wide strip
by pulling on it. Some of the molten adhesive remains on the sub-
oor, but some is still attached to the removed layer, thus containing
the hazardous asbestos bers and preventing them from being dis-
persed in the air. The process of heating the adhesive layer is rela-
tively slow: the machine traverses the room at a speed of 3 m/hr. To
save operator time, it is desirable to design a sensor that will allow
123
Figure 7.1 The existing ooring removal machine.
Wear layer
1-2 mm Backing layer (asbestos)
Adhesive
Subflooring (concrete, wood)
Figure 7.2 A cross-section in typical asbestos-containing ooring.
the machine to steer itself along the cut edge of the ooring once a
rst strip has been removed.
Performance
As with any sensor design, the quantity to be measured needs to
be quantied rst. The cut edge to be followed results from a person
making a cut with a utility knife at an approximate distance of 30 cm
from a wall or a previously cut edge. The cut edge may therefore be
somewhat skewed, jagged, or wavy, as shown in Fig. 7.3. Reasonable
assumptions to quantify the cut edge are a maximum waviness of 3
cm (amplitude of 1.5 cm) and a maximum jaggedness of 0.5 cm.
The height of the cut edge is 1 to 2 mm above the suboor, but it
may be tilted sideways; that is, it may form angles other than 90
124
Conceptual Design Case Study: Cut-Edge Sensor for Flooring Removal
Top view of a wavy edge 3 cm
Top view of a jagged edge 0.5 cm
1m
Figure 7.3 Characterization of wavy and jagged cut edges in the ooring.
with the oor when viewed from the front. We set this deviation to
a maximum of 30 from the vertical.
When use of the sensor-guided machine is considered, it is
revealed that too accurate a sensor will cause the machine to closely
follow the irregularities in the cut edge, which is undesirable. What
is needed is some averaging effect where local variations and devia-
tions are ignored. Achieving this by mathematical averaging of sev-
eral sensor readings is clearly overkill. The same result can be
obtained by designing a low-sensitivity sensor in the rst place.
Thus, we set the desired resolution to 1 cm, meaning that deviations
smaller than 1 cm to either side of the machine heading should not
produce a change in the sensor reading. If the sensor functions well,
the deviation from the nominal course will never exceed about 1 cm.
However, to be on the safe side, we may want the desired range of
measurement to be 3 cm from the center.
Some aspects of the operation of the machine are relevant to the
functioning of the sensor. The low speed of 3 m/hr implies that the
sensor can have a slow response time and still provide adequate
positioning accuracy. Consider, for example, the unlikely case of the
machine deviating by a full 90 from its course. If the sensor takes
10 seconds to recognize it, the deviation will still be less than 1 cm,
or 3% of the cut-strip width.
The high temperatures involved in the ooring removal opera-
tion may pose a problem to the sensor. When the adhesive reaches
125
100C, the top surface of the wear layer may be at 220C, and the
sensor should be able to endure this condition. In addition, the sur-
faces over which the machine operates may be characterized as fol-
lows: the subooring is wood or concrete of variable color, covered
with soft and hard adhesive residue. The wear layer of the yet-to-be-
removed ooring may have any color and pattern printed or embossed
on it. Such patterns may also consist of grooves as deep as one-half
the total thickness, that is, 0.5 to 1 mm deep.
As does the machine itself, the sensor should also use regular
electrical household current for its operation. In order not to sub-
tract from the power available for the heating task, the power con-
sumption of the sensor will be limited to 100 W.
Value
Freeing the human operator from the task of continuously
monitoring and guiding the ooring removal machine is the major
benet of the cut-edge sensor. With a sensor-equipped machine, the
operator will be able to attend to other tasks or place several
machines in different locations to operate simultaneously. The
monetary value of the sensor can therefore be established by con-
sidering the increased productivity. For the present discussion, we
recognize that the machine itself is a relatively expensive device with
a retail cost of around $2000, so the target production cost for the
sensor is set at $100.
Size
Since the bottom of the machine is covered with heating ele-
ments, possible sensor locations, which offer downward accessibil-
ity, include the front, back, and sides. Installation in the front is
probably preferable in terms of the timely detection of deviations in
course. There does not seem to be a reason to severely limit the
physical size or weight of the sensor. It should not be as bulky or
heavy as to hinder the performance of the machine, however, so we
set its maximum size to a 20 20 20-cm cube, and its maximum
weight to 30 N (3 kg). We also note that the sensor may be stored
and transported separately from the machine, and mounted and
plugged in on site.
126
Safety
Besides provisions normally expected from a device used near
people, the most important safety requirement is to comply with the
overall goal of the machine; that is, the sensor should not contribute
to releasing any asbestos bers from the oor. This requirement is,
of course, of utmost importance.
1. Measurement of the location of the 12-mm high cut edge

with respect to the machine heading. The characteristics of
the cut-edge geometry, subooring, and removed layers are
as described in the need analysis.
2. Measurement resolution: no smaller than 1 cm.
3. Maximum deviation to detect: 3 cm to either side of the cut
edge.
4. Toleration of surrounding air temperature of 100C.
5. Maximum power consumption at household electricity:
100 W.
6. Maximum manufacturing cost: $100 at an expected produc-
tion volume of 500 units per year over a 10-year period.
7. Maximum size: 20 20 20 cm.
8. Maximum weight: 30 N.
9. Not causing the release of any asbestos bers during operation.
We may divide various position-sensing techniques into two broad

categories: contact and noncontact. A mechanical follower idea that
can be used as a cut-edge sensor is shown in Fig. 7.4. The problem
with using contact sensors in this application is that they need to
follow a sticky edge covered with hot adhesive and may also release
asbestos bers from the backing layer. A much more promising
venue to explore is noncontact sensors.
127
Angle sensor
Follower
Wear and
backing Subflooring
layers
Figure 7.4 A contact sensor based on a mechanical follower.
Noncontact sensing can be based on various principles. We may

detect the cut edge by performing two range measurements, one to the
wear layer and one to the suboor. The difference in range is 12 mm,
so an accurate range nder will be required. We may look into range
measurement by timing the ight of sound waves, as may be done in
an auto-focus camera, for example. The required sensor resolution (1
mm corresponds to about 3.6 107 second) is much higher than that
in cameras, however, and may be expensive to implement.
A different noncontact technique is imaging, where the intensity
of light reected from the surfaces is measured. It is unclear at this
stage whether a distinct image of the cut edge can be created.
Nevertheless, this principle seems more promising than range sens-
ing, and it is therefore selected as a starting point for parameter
analysis.
PI: Measure the different light intensities bouncing back from the
wear layer and subooring.
CS: The conguration in Fig. 7.5 uses a light bulb and two one-
dimensional arrays (rows) of photosensitive elements. The bulb
is shielded so that its light does not reach the optical detectors
directly. Each array senses a different light intensity because the
light reects from different surfaces. if the machine deviates
from its intended course, the variation in light intensity will
produce an error signal.
128
Photodetectors
Figure 7.5 A conguration to sense the intensity of light reected from dif-
ferent surfaces.
E: The reected light may not vary enough to determine the loca-
tion of the edge because of the presence of adhesive residue and
dirt and the possibility of different material and color combi-
nations for the surfaces.
PI: Create a greater difference in light intensities. If the images of the
wear layer and subooring are indistinguishable from each other,
create and sense a third imagethe shadow of the cut edge.
CS: Position the bulb away from the cut edge and let its light shine
at an angle across the edge so that a shadow is cast, as shown in
Fig. 7.6. The shadow will be detected as a dark line on a single
optical detector array.
E: Now that the image of the shadow is the quantity to be sensed,
we should try to make it large enough for easy detection. Also,
the cut edge may be inclined at up to 30, so the angle of inci-
dence of the light (the angle between an incident ray and the
normal to the surface) should be greater than 30 to produce a
shadow.
PI: The size of the shadow depends on the angle of incidence.
Increasing this angle will make the shadow easy to detect.
CS: The conguration of Fig. 7.7 moves the bulb further away from
the cut edge to cast a large shadow. The detector array is moved
and tilted accordingly. At the nominal position, with the image
of the shadow at the center of the detector, the angle of inci-
dence of a light ray hitting the cut edge is about 60.
E: The shadow will not be clearly seen on the photodetector
because the light bouncing from the oor will diffuse and scat-
ter in all directions.
129
Shadow
Figure 7.6 Lighting the cut edge at an angle creates a detectable shadow.
Figure 7.7 Increasing the angle of incidence of the light enlarges the image
of the shadow.
PI: Focus the reected light by using a lens.

CS: A lens is added in front of the detector to focus the image of the
cast shadow. For the required performance, a low-cost plastic
lens, as used on single-use cameras, may sufce. The congura-
tion in Fig. 7.8 integrates the lens with the detector in one hous-
ing. A separate shield between the bulb and detector is still
required to prevent direct illumination of the detector. The
focusing lens (convergent, or convex, lens) allows the use of a
smaller detector array.
At this point we need to calculate some dimensions to
match the performance of the design with the required resolu-
tion and range of detection. Figure 7.9 shows the information
pertinent to using a convergent lens.
The fact that the image is inverted relative to the object
should not pose a problem because this can easily be accounted
for in the signal-interpretation step. The magnication of the
lens, that is, the ratio of image size to object size, is given by b/g.
Another important relationship is the following approximation:
130
Lens
Figure 7.8 Adding a lens in front of the photodetector to create a sharp

image of the shadow.
g b
f
Object
Image
Lens
Figure 7.9 Basic geometrical properties of a convergent lens (g = object dis-

tance, b = image distance, f = focal length).
1 1 1
= +
f g b
There are two issues to consider when determining dimen-

sions for our design: the spacing between the sensing elements
and the thickness of the image of the shadow on the detector. If
the actual detection range (3 cm = 60 mm) is converted to an
imaging range of 20 mm, the magnication is set at one-third.
From the constraints on the size of the sensor, we decide to choose
g = 75 mm and b = 25 mm, giving the desired magnication. The
focal length of the lens is thus determined to be 18.75 mm.
The height of the cut edge varies between 1 and 2 mm, and
the shadow is cast by shining light at 30 to the horizon. The
thickness of the object to be sensed is therefore twice the
height, or 2 to 4 mm. With the one-third magnication, this
131
3.33 mm 3 mm
Figure 7.10 The seven-photodiode linear array used as detector.
translates into an image 0.67- to 1.33-mm wide. In order to keep

the resolution low enough, that is, not detect deviations smaller
than 1 cm, we should use six sensors in the array. For reasons of
symmetry, however, seven sensing elements are a better choice.
We also decide now to use 3-mm diameter photodiodes as the
sensing elements. The layout of the detector is shown in Fig. 7.10.
E: The signal-to-noise ratio may be too small due to the presence
of ambient light. The signal is the image of the shadow, created
by different reected light intensities. Light from other sources
in the room, especially the heating elements on the machine,
constitutes the noise.
PI: A larger signal-to-noise ratio can be obtained either by decreas-
ing the noise or increasing the signal. To decrease the noise,
additional physical shielding could be used to further enclose
the bulb, cut edge, and detector. A simpler solution, however,
may be to increase the signal by using a light source whose
wavelength is different from that of the noise.
CS: The heating elements emit light whose infrared content is rela-
tively high. Operating the bulbdetector pair at the other end of
the spectrumthe ultraviolet (UV) wavelengthwould help
separate the signal from the noise. Specifying a UV bulb and
detector, or a lens-mounted lter that blocks all but the UV
waves, is straightforward.
E: The conguration developed so far seems viable; however, what
would guarantee detection of the cut-edge shadow and not
shadows produced by grooves in the wear layer?
PI: The depth of the wear-layer grooves is about one-half the
height of the cut edge. We may try to use this difference to dis-
tinguish between the two, but adhesive residue on the sub-
132
Groove Cut edge
Figure 7.11 The difference between a groove (left) and the cut edge (right).
Sensor housing
Detector 2 Detector 1
Shield
Bulb 1
Bulb 2
Figure 7.12 A design that comprises two multiplexed bulbdetector pairs.
ooring may cause the cut edge to appear shallower than it

really is. A better concept is based on the realization that
grooves are symmetrical while the cut edge is one-sided, as
shown in Fig. 7.11. This property can be used to cast shadows
by shedding light from different directions.
CS: If the design is doubled and its operation made to alternate
(multiplex) between the bulbdetector pairs, then the groove
will always produce a shadow while the cut edge will be detected
on alternate readings only. A conguration with this arrange-
ment is shown in Fig. 7.12. A groove on the wear layer would
produce a signal (shadow) on both detectors, while the cut edge
would show on detector 1 only.
E: Most major issues have been dealt with and the solution seems
feasible, so we conclude the conceptual design development at
this point. The sensor will be mounted on one of two brackets
provided on the front of the ooring-removal machine. One
bracket will align the center of the sensor with the left side
panel of the machine and the other, with the right side panel.
133
Measure different light

intensities bounced from the
wear layer and subflooring
Light bulb with 2 rows of
photosensitive detectors
Not enough variation
in reflected light
Create and sense the
shadow of the cut edge
Position the bulb so that it

casts a shadow
The cut edge may
be tilted
Increase angle of incidence
to make the shadow larger
Bulb and detectors positioned

for nominal angle of
Bounced light incidence of 60
diffuses
Use a lens to focus
the reflected light
A low-cost focusing lens

is added to the detectors
Ambient light lowers
signal-to-noise ratio
Increase signal intensity by
using a different wavelength
for the light source
Use UV light bulb and

Grooves in the wear- detectors
layer may obscure the
cut edge
Shedding light from different
directions
The design is "doubled"

and operation multiplexed
Finished
Figure 7.13 Summary of the cut-edge sensor parameter analysis process.
The parameter analysis process of designing the sensor is summa-

rized in Fig. 7.13. It shows how the simple concept of measuring the
intensity of light bounced from the oor has evolved into a device
that actually creates a new imagea shadow of the cut edgeand
senses it. The quantity to be measured by the sensor can thus be
enhanced and distinguished from the surrounding noise, which is in
the form of grooves and other surface features and ambient light.
134
(a) (b)
Figure 7.14 Images of the cut-edge shadow on a two-dimensional photode-

tector array: (a) parallel course deviation, (b) angular deviation.
Several secondary issues regarding the design should still be

addressed. A decision needs to be made whether to use one-dimen-
sional detector arrays (a single row of photodiodes) or two-dimen-
sional grids. The signals from the detectors should be conditioned
(amplied, ltered, and possibly digitized) and fed to a controller
for the steering motor. A control strategy needs also to be developed,
specifying how error signals will be converted to course-correction
instructions. For example, Fig. 7.14 shows two possible signals
recorded on a two-dimensional detector array. The one in (a) repre-
sents a parallel deviation, while the signal in (b) shows an angular
deviation. It may also be desirable to calculate the velocity of devi-
ating from the nominal course by taking two successive position
readings over a given time period.
1. Examine noncontact range-sensing technologies and compare

them in terms of their ability to provide a 1- to 2-mm measure-
ment resolution at a reasonable cost.
2. Design a noncontact sensor for the application of this chapter

based on a range-measuring principle.
3. In the technology identication section (7.4), we mentioned that

contact sensors had two major problems: they would be touching
135
a sticky and hot cut edge, and they might cause the release of
asbestos bers. Design a contact sensor for this application that
would overcome these two difculties.
7.8 Bibliography
As in the previous chapter, general books on measuring techniques

will come in handy while designing a sensor. The following book is
again suggested:
Doebelin, E. O. Measurement Systems: Application and Design.
4th ed. New York: McGraw-Hill, 1990.
When the choice was made to use imaging, books on optics proved
useful. Many such books are available, so none in particular will be
mentioned here. A manufacturer or retailer catalog, such as the one
below, should be consulted for availability of low-cost lenses and
photodetectors:
Edmund Scientic Co. Edmund Industrial Optics.
136
8
Low-Cost Industrial Indexing
Systems
This chapter describes low cost and exibility as two key require-
ments in the design of a new drive system for a mechanical indexer.
Need analysis helps to generate two key insights into the design task.
Based on these two insights, two competing designs are developed
using the parameter analysis methodology to address the require-
ments. A detailed discussion of technical issues involved in realizing
these concepts is presented. Both of the designs were built and tested
for performance.
Indexing systems are often used in industry to incrementally move

products along an automated line and stop them at predetermined
workstations. At these workstations, people or machines perform
assigned tasks, such as welding or assembly of components. Two
common technologies for producing incremental motion are servo
drives and mechanical indexers. Servo drives employ a dc motor,
speed reducer, clutchbrake system, sensors, and electrical controls.
They are highly exible in the sense that they can accommodate
changes in the product line by adjusting the timing sequence for
various activities. In contrast, mechanical indexers constitute rigid
automation. They use a continuously running electric motor, speed
reducer, clutchbrake system, and indexer unit. The indexer unit
137
converts the constant input motion of the electric motor into inter-
mittent output motion (linear or rotary) by using a camfollower
mechanism. Both of these indexing technologies are expensive.
Market research has indicated that there is a need for exible, yet
moderately priced, indexing systems. The mechanical indexer tech-
nology would capture a large segment of the low-end market if it
could satisfy cost and exibility requirements. In addition, typical
applications in this market segment are usually not very sensitive to
the acceleration prole of the load. The goal of this design task is
therefore to develop a low-cost drive system for mechanical indexers.
The black-box methodology with inputs and outputs (see Section

2.2) can be used to precisely identify the need. In this particular
case, the input is ill-dened because several types of energy inputs
are possible. The need is thus formulated by looking at the output
of the black box and requiring that it provide energy to the shaft
driving the mechanical indexer. Several questions come to mind in
this context:
1. What is the energy requirement for a typical mechanical

indexer? Is it constant, or does it vary with time?
2. How much energy or power is required?
3. What are the characteristics of an ideal solution?
4. What are the possible energy sources?
5. Where is the cost locked up? Or what are the primary cost
drivers in the current design?
Performance
The primary function of the drive system is to supply energy
to the indexer. Because the indexer continuously cycles through
acceleration, deceleration, and dwell (stop) periods, the required
power and torque do not stay constant. To determine them, an ideal
acceleration prole for the output load must be determined. This
138
Conceptual Design Case Study: Low-Cost Industrial Indexing Systems
prole can then be used to nd the velocity, displacement, and

torque proles.
Several acceleration proles (constant, trapezoidal, modied
trapezoidal, and modied sinusoidal) can be used to accelerate and
decelerate the load. Most indexers use the modied sinusoidal accel-
eration prole since it reduces jerk, peak acceleration, and peak
velocity. By decreasing the peak velocity, this prole reduces the
amount of kinetic energy that has to be dissipated each time the
assembly line is brought to a stop. Note that although the nal solu-
tion may not produce the modied sine acceleration prole, this
assumption provides an initial starting point for determining torque
requirements and, later, sizing parts. It also helps in the evaluation
of competing conceptual solutions when they satisfy all other
requirements to the same extent.
Based on the equations of a modied sinusoidal acceleration, a
plot of the angular acceleration and velocity of the load for different
positions of the input shaft is shown in Figs. 8.1 and 8.2. It is assumed
in these calculations that the speed of the input shaft is constant.
The output torque of the drive system is the product of the
angular acceleration () and inertia (I) of the load. However, to
design the drive system, the input torque required by the indexer
must be characterized. The indexer can be regarded as a speed
reducer with variable gear ratio, where the instantaneous gear ratio
is the output shaft speed divided by the input shaft speed. Based on
this insight, the input torque to the indexer can be obtained by mul-
tiplying the output torque of the drive system (I) by the gear ratio
calculated from Figs. 8.1 and 8.2. The resulting input torque prole
for the indexer is shown in Fig. 8.3. An ideal drive system must
closely emulate this torque prole.
Let us now estimate the energy requirement of the indexer. At
an abstract level, the indexer system accelerates an object from a
state of rest to the desired velocity and then decelerates it back to
rest. Neglecting frictional losses, we nd that the net energy require-
ment is therefore zero. Close examination of the input torque pro-
le (Fig. 8.3) reveals this fact: The area under the curve during the
acceleration period is numerically equal to the area during the
deceleration period, and these areas represent energy. Energy supply
139
0.025
Acceleration Decelaration Dwell

0.02
0.015
Output shaft acceleration (degrees/s 2 )
0.01
0.005
0
0 10 20 30 40 50 60 70 80 90 100 110 120 130 140 150 160 170 180
-0.005
-0.01
-0.015
-0.02
-0.025
Angular position of input shaft (degrees)
Figure 8.1 The modied sinusoidal acceleration prole of the output shaft.
0.9
0.8
Output shaft velocity (degrees/s2)
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
0 10 20 30 40 50 60 70 80 90 100 110 120 130 140 150 160 170 180
Angular position of input shaft (degrees)
Figure 8.2 The output shaft velocity profile obtained by integrating the
acceleration.
140
200
150
Required input torque (in-lb)
100
Energy
input
50
0
0 10 20 30 40 50 60 70 80 90 100 110 120 130 140 150 160 170 180
-50
Energy
removal
-100
-150
-200
Angular position of the input shaft (degrees)
Figure 8.3 The required input torque for the indexer.
is needed to drive the load in the acceleration phase, but this energy
must be removed from the system during deceleration.
Value
Cost analysis reveals that over 50% of the product cost comes
from the electric motor and the speed reducer. For the new drive sys-
tem to compete on the basis of cost effectiveness, it must eliminate
these components. Power consumption is relatively insignicant for
these drive systems. To reduce the required initial investment by the
customers, the manufacturing cost should not exceed $400.
In this example, we are not generating a range of performance
versus value estimates since the target performance of the device is
somewhat narrowly dened (essentially reducing cost). In a more
complete need analysis, we would estimate the value to the user as it
varies over a range of performance specications, depending on the
intended application of the indexer.
Size
Existing indexers have proved to be extremely reliable, with a
long life. Many units, however, may require replacement of their
141
drive system. It is therefore decided that the new drive system must
interface with the existing indexers. This constraint imposes geo-
metric requirements, such as the height of the drive shaft from the
base and the maximum available space.
1. Power an indexer with a peak torque of 180 in-lb.

2. Provide the ability to adjust the timing sequence for various
activities.
3. Maximum manufacturing cost: $400 at low production
volume.
These are three key requirements. Additional requirements concern-

ing other aspects, such as size, should be included in the complete
list of requirements.
Two ideas, based on the insights gained in the need analysis phase, are:
1. Pneumatic actuation. Electric motors typically run at high

speeds, thus requiring a step-down speed reducer. Linear
actuators, particularly pneumatic cylinders, may be a low-
cost alternative as a power source. Since they are widely used
in industry, air supply is readily available. They can be actu-
ated on-demand and are also relatively slow. Power from a
pneumatic actuator may be used to continuously rotate the
drive shaft of the indexer.
2. Spring energy storage. The net energy supply required to
drive the indexer is very smallit equals only the frictional
losses. A smart approach would be to store the dissipated
energy during the deceleration phase and release it for accel-
erating the object. The actuator will then need to supply only
a small amount of energy to make up for the frictional losses.
A spring may be used for such energy storage.
142
These two ideas serve as initial conditions for the two independent
parameter analysis processes described in the next two sections.
8.5 Parameter Analysis of Conceptual Design I
PI: The rst conceptual design process starts with the pneumatic
actuator idea. For the actuator to drive the shaft, its linear
motion should be converted to the rotary motion of the shaft.
CS: The conguration in Fig. 8.4 uses a pneumatic actuator, cable,
and pulley mounted on the drive shaft to convert motion from
linear to rotary. During the actuation stroke, the piston moves
from right to left while pulling the cable. The cable in turn
rotates the pulley and the drive shaft in the counterclockwise
direction. Note that at this stage in parameter analysis, the con-
guration is not fully functional. This is an acceptable practice
because the goal is to have an initial conguration that can be
evaluated and improved.
E: During the return stroke the piston tries to push the cable.
Therefore, the cable does not return to its initial position.
PI: Tension must be maintained in the cable during the return
stroke.
CS: A tension spring can return the cable to its original position.
The new conguration incorporating the spring is shown in
Fig. 8.5. The cable moves from right to left during the actuation
stroke while rotating the shaft counterclockwise and extending
the spring. The tension spring pulls the cable back to its origi-
nal position during the return stroke.
E: This conguration produces an oscillatory motion of the shaft.
During the actuation stroke (the piston moves left), the drive
shaft rotates counterclockwise, but it reverses direction during
the return stroke (indicated by dashed arrows). This is obvi-
ously undesirable because the drive shaft needs to continuously
rotate in one direction.
PI: Eliminate the clockwise rotation of the shaft during the return
stroke of the actuator.
CS: A one-way clutch can be used to transmit torque in one direc-
tion while allowing free running in the opposite direction. Two
143
Direction of the actuation stroke

Piston
Direction of motion during
the actuation stroke
Rigid link
Cable
Drive shaft
Pulley
Figure 8.4 The linear motion of the actuator is converted to shaft rotation by
a cable and pulley.
Piston
Rigid link
Pulley
Spring
Drive shaft
Figure 8.5 The spring restores the cable to its original position during the
return stroke of the actuator.
popular embodiments of a one-way clutch are the sprag and

roller-ramp types shown in Fig. 8.6. When the driving outer
race rotates in the locking direction, sprags or rollers wedge
between the inner and outer races, thus transmitting the
torque. When the direction of rotation is reversed, the sprags or
rollers disengage and allow free relative motion between the
races.
The conguration is now modied to incorporate a one-
way clutch between the pulley and the drive shaft, as shown in
Fig. 8.7. The clutch will transmit motion and torque between
144
Lock direction Lock direction

Sprag
Outer race Roller
Inner race Ramp
(a) (b)
Figure 8.6 (a) A sprag one-way clutch and (b) roller-ramp one-way clutch.
Piston
Rigid link
Pulley
Drive shaft
One-way clutch
Figure 8.7 A one-way clutch is added to eliminate clockwise rotation of the

drive shaft.
the pulley and the shaft in the counterclockwise direction and

free wheel in the clockwise direction. This will ensure unidirec-
tional rotation of the drive shaft.
E: The new conguration does not utilize the return stroke of the
actuator. This creates unnecessary idle time and limits the over-
all speed.
PI: Use the return stroke for powering the drive shaft continuously.
CS: A system of two one-way clutches can be employed as shown in
Fig. 8.8. When the pneumatic actuator moves from right to left
(marked by solid arrows), pulley #1 rotates in the counter-
clockwise direction and transmits the torque to the drive shaft.
At the same time, pulley #2 free wheels as it rotates clockwise.
145
Pulley #1
Springs
Drive shaft
Bearing
One-way
clutch
Pulley #2
Idler pulley
Figure 8.8 Two pulleys in parallel continuously power the drive shaft.
During the reverse stroke of the actuator (dashed arrows), pul-

ley #2 transmits the torque while pulley #1 rotates freely. Two
springs are used to maintain tension in the cable.
E: The springs add complexity to the design as additional parts.
They also introduce continuously changing forces, which
depend on the instantaneous deection of the springs. These
variable forces subtract from each other to cause continuously
decreasing torque on the drive shaft. The resulting torque pro-
le is not the desired one.
PI: The cable moves back and forth in accordance with the move-
ment of the actuator, and the springs always deect opposite
each other (one shortens as the other elongates). A xed-
length, closed-loop cable would do exactly the same while elim-
inating the springs.
CS: A conguration that employs a closed-loop cable is shown in
Fig. 8.9. The operation of this design is identical to that of the
previous conguration except for the tension springs.
E: This design eliminates the need for springs, but it violates size
constraints. It is also more expensive because it requires an
additional idler pulley and associated shaft and bearings.
PI: Improve overall compactness and reduce the number of com-
ponents.
146
Bearing
Pulley #1
Drive shaft
Idler pulley
Bearing
One-way
clutch
Pulley #2
Idler pulley
Figure 8.9 The springs are replaced by a closed-loop cable running

between two idler pulleys.
CS: A single shaft can be used to mount both idler pulleys by reposi-
tioning them, as shown in Fig. 8.10. If the design is made compact
and uses conventional chains or belts as cable, considerable twist-
ing of the latter may produce undesirable effects and short life. We
therefore decide to use a 3-D chain (see detail in Fig. 8.10), which
is very versatile and can be arranged between pulleys at different
angles. The chain teeth positively engage the sprockets, thereby
eliminating any possibility of slip. Furthermore, the overall size of
the design is now greatly reduced.
E: This conguration succeeds in replacing the costly electric
motor and the speed reducer. However, it still requires a brake
to bring the load to rest. A commercially available brake will
need to be added. Otherwise, this conceptual design process,
summarized in Fig. 8.11, is complete.
8.6 Parameter Analysis of Conceptual Design II
PI: The second conceptual design is based on the spring energy

storage idea recognized in the technology identication step:
Use a spring to accelerate and decelerate the drive shaft.
147
Pulley #1
Drive shaft
Pulley #2
Bearing
Idler pulleys
Figure 8.10 Rearranging the idler pulleys eliminates one shaft. A 3-D chain
(see detail) is specied.
Convert the linear motion of a

pneumatic actuator to the Pneumatic actuator -
rotary motion of the drive shaft pulley arrangement
Cable does not return
to the original position
Maintain tension in the cable Tensioning spring
added
Produces a back-and-forth
oscillatory motion
A one-way clutch added
Eliminate the clockwise rotation
between the pulley and
of the shaft
Does not use the drive shaft
the return stroke
Use the return stroke to Parallel arrangement of
continuously power the shaft the pulleys
Increased design
complexity
Eliminate springs by using
a closed-loop cable Closed-loop cable added
Violates size constraints.
Increased part count
Reduce the overall size
and part count 3-D arrangement
Need to remove energy

in the deceleration stage.
Commercially available brake
can be used. Finished.
Figure 8.11 Summary of the parameter analysis process for conceptual

design I.
148
Barrel cam
Cam follower
Spring
(a) (b)
Figure 8.12 A barrel cam mechanism shown in (a) the fully compressed and
(b) extended positions of the spring.
CS: A barrel cam can convert the force of a helical compression spring
into rotary motion of the drive shaft, as shown in Fig. 8.12. The
spring is initially compressed (Fig. 8.12a) and therefore stores
potential energy. This stored energy is released during the acceler-
ation phase of the modied sine prole (Fig. 8.1) until the spring
reaches its fully extended position (Fig. 8.12b). Next, the rotation
of the drive shaft compresses the spring, thereby transferring the
energy back to it, and the cycle repeats itself.
E: Barrel cams are difficult to manufacture and may be too
expensive.
PI: Plate cams are cheaper to manufacture than barrel cams, so use
a plate cam for the motion conversion.
CS: Figure 8.13 shows a plate cam mechanism. The operation is
identical to that of the barrel cam mechanism. The cam can be
initially sized by performing an energy balance (neglecting fric-
tional losses):
Fspring dr = T d
where Fspring is the force exerted by the spring (a function of its

displacement), dr is the incremental change in radius, T is the
149
Drive shaft Drive shaft
Cam
Spring Cam Spring
Follower
Follower
(a) (b)
Figure 8.13 A plate cam conguration shown in (a) the fully compressed
and (b) extended positions of the spring.
time-dependent torque required to drive the indexer, and d is the

incremental change in the angle of the drive shaft. Assuming ini-
tial values for the spring rate, initial compression, and base radius
of the cam, the cam prole can be determined. The preliminary
calculations show that a small elliptical cam of about 4 major
diameter and a spring with a 75 lb/in. rate can drive the indexer.
E: The sizing calculations prove the viability of the design.
However, frictional losses were neglected. Because there is no
external energy input into the system, the friction forces will
prevent the cam from reaching its original position; that is, the
fully compressed position of the spring.
PI: Add an external energy source to make up for frictional losses.
CS: A pneumatic actuator would be a cost-effective energy source.
A modified configuration incorporating a pneumatic actua-
tor is shown in Fig. 8.14. The operation of this design consists
of three distinct phases. The initially compressed spring (Fig.
8.14a) drives the cam attached to the drive shaft through 90,
reaching its fully extended position (Fig. 8.14b). The cam pro-
file can be determined by treating an infinitesimally small sec-
tion of it as an inclined plane and resolving the spring and
frictional forces. Since there are no impact loads, the profile
can be made to match the modified sinusoidal acceleration.
Next, the torque from the inertial load compresses the spring
by rotating the shaft through another 30 (Fig. 8.14c). The
cam profile can be determined as before to match the modi-
fied sinusoidal acceleration profile. Finally, the pneumatic
150
Pneumatic
actuator
Drive shaft
Cam
Follower
Spring
(a) (b) (c) (d)
Figure 8.14 The spring is in the fully compressed position (a) when it starts
rotating the cam and reaches its fully extended position (b). The load inertia
compresses the spring further during the deceleration period (c), when the
pneumatic actuator kicks in (d).
actuator rotates the shaft by 60 (Fig. 8.14d), bringing the

spring back to its fully compressed position. A simple
mechanical switch triggered by the rotation of the drive shaft
times the actuator.
E: The spring supplies a significant portion of the energy during
the acceleration phase, so it is important to stop the drive sys-
tem in the fully compressed position of the spring. Otherwise,
the actuator will not be able to provide the required energy to
accelerate the system. When the spring starts from rest in the
fully compressed position, the direction of cam rotation is
unpredictable.
PI: Stop the cam slightly after the fully compressed position of the
spring. This will ensure rotation in the correct direction.
CS: A solenoid-released latching mechanism can be used to stop
the follower arm, as shown in Fig. 8.15. The latch is spring-
loaded and therefore locks the follower when the groove in the
follower arm aligns with the latch. When the solenoid is actu-
ated, it pulls the latch back, thus releasing the spring.
E: Locking the follower will not stop the rotation of the drive shaft
since the inertial load can still rotate the cam, as shown in Fig. 8.16.
PI: To stop the system, lock the cam or drive shaft directly.
151
Pneumatic
actuator
Drive shaft
Cam
Latch
Follower
Spring
Solenoid
Figure 8.15 The latch mechanism in the locked position.
Pneumatic
actuator
Drive shaft
Cam
Follower
Latch
Spring Solenoid
Figure 8.16 Drive shaft rotation due to the inertial load is possible even
when the follower is locked.
CS: The latch mechanism is positioned to lock the drive shaft

directly.
E: All critical parameters have been addressed, so the parameter
analysis process is concluded and summarized in Fig. 8.17.
152
Use the spring to

accelerate and decelerate Barrel cam configuration
the drive shaft Barrel cams are difficult to
manufacture
Use a plate cam for Barrel cam is replaced
motion conversion by a plate cam
Frictional losses prevent
the cam from reaching
the fully compressed position
Add an external energy source Pneumatic actuator
to make up for frictional losses added
The drive should be stopped
in the fully compressed position
Stop the cam after the of the spring
fully compressed position Latching mechanism on
of the spring the follower arm
Inertial loads can
rotate the cam
Stop the cam or drive shaft
Latch mechanism on
the drive shaft
Finished

design II.
This case study demonstrates the strong link between need analysis
and conceptual design. A good need analysis provides thorough
understanding of the design task and valuable insights. The need for
a exible yet inexpensive drive system dominated the rst concep-
tual design, which replaced the electric motor and speed reducer of
conventional indexers with a pneumatic cylinder. Realizing that the
net energy required was very small became the dominant parameter
in the second conceptual design. Another critical issue identied in
the need analysis was cost, and both designs have been evaluated
and modied several times to address this requirement.
A prototype of the rst design was built using two solenoids to
simulate the forward and return strokes of a pneumatic actuator, as
shown in Fig. 8.18. If the force from the actuator is assumed to be
constant, then the torque on the drive shaft is also constant. The
constant torque prole does not result in the modied sinusoidal
153
Figure 8.18 Prototype of the drive system with solenoids instead of a pneu-
matic actuator. (A 12-in. ruler is shown for scale.)
Figure 8.19 Prototype of the drive system in the fully extended position of the
actuator. (The spring cannot be seen in the gure.)
154
acceleration prole. The constant torque causes a jerk during the

initial motion of the load. However, the chain acts as a damper and
reduces this jerk considerably. Even though the design does not fully
emulate the modied sinusoidal acceleration prole, it satises the
key requirements of cost and exibility.
A prototype of the second design is shown in Fig. 8.19. It was
tested to determine how closely it resembled the modified sinu-
soidal profile. Based on the results, the cam profile was slightly
modified to accommodate the force profile of a typical pneumatic
actuator.
1. Critically evaluate the two conceptual designs presented in this

chapter to identify their weaknesses. Then, apply parameter
analysis to address these weaknesses and improve the designs.
2. Develop a conceptual design that incorporates a small electric

motor instead of the pneumatic actuator in conceptual design II.
Discuss the advantages and disadvantages of such a system.
3. Develop a design for powering a small boat. The user pushes

two pedals alternately with his or her feet. This motion must be
converted to continuous, unidirectional rotation of a paddle
wheel. Perform detailed calculations to study the feasibility of
the solution. Examine if a one-way clutch can be used to solve
this problem.
8.9 Bibliography
An overview of cam design can be found in:

Norton, R. L. Design of Machinery: An Introduction to the Synthesis
and Analysis of Mechanisms of Machines. 2nd ed. New York:
McGraw-Hill, 1999.
155
The following book by the founder of Ferguson Machine Company

describes various cam mechanisms, different acceleration proles,
and their advantages:
Neklutin, C. N. Mechanisms and Cams for Automatic Machines:
Inuence of Dynamic Forces on High-Speed Machine Compo-
nents. New York: American Elsevier Publishing Company, 1969.
156
9
Equal-Channel-Angular-Extrusion
Metalworking
A novel extrusion process with its unique machinery is the topic of

this relatively complex case study. It begins with a general need for a
method to improve material properties by deformation processing.
The end result is quite a detailed design of a machine to carry out
Equal-Channel-Angular-Extrusion metalworking. Metal bars are
forced in this process through a bent channel and thus undergo
strain hardening while preserving their size and shape. The benets
of this process are also explained.
Plastic deformation in metal forming is used for two purposes: to

change size and shape, and to improve material properties such as
ductility, strength, and toughness. Forging, rolling, and extrusion
are examples of forming and shaping processes in which adding
strain to a workpiece results in rening the grain size, making the
microstructure more uniform, and work hardening of the metal.
Work hardening and ner grains are responsible for an increase in
the shear strength, and hence the overall strength, of the material.
This increase results from entanglements and impediments of dislo-
cations by shear deformation, which usually takes place at 45 to the
direction of the applied load. The greater the strain induced in the
material during processing, the more the entanglements, and hence
157
a higher strength is obtained. Metalworking can also remove metal-

lurgical casting defects, such as cavities and voids, and produce
desired textures.
Adding strain is used extensively in metalworking processes
for example, strengthening wire by reducing its cross-section by
drawing it through a die, forging the head of bolts, and producing
sheet metal for automobile bodies and aircraft fuselages by rolling.
Some of the common deciencies of conventional processes can be
demonstrated by examining the process of extrusion. In extrusion,
material is forced through a die, which is similar to squeezing tooth-
paste from a tube. The three main problems are:
1. Material properties may not be homogeneous throughout

the cross-section, due mostly to the nonuniform ow of the
metal through the die.
2. Very large reductions of size are required to obtain signi-
cant work hardening or very ne and uniform grain size. If
the dimensions of the nal workpiece are to be relatively
large, then the initial billet must be huge.
3. To produce the large changes in cross-section, high pressures
need to be applied to the billet, which in turn result in high
overall loads requiring large and expensive machines.
Figure 9.1 is a schematic of a conventional extrusion process. The

force F exerted on the punch translates into pressure p (p = F/A0) on
the billet and forces it to change its cross-sectional area from A0 to Af.
The true strain, , experienced by the material is dened as:
= ln(A0 /Af ) (9.1)
The punch pressure required to produce while overcoming the

friction between the billet and the die is estimated in many text-
books from graphs and equations such as:
p=k (9.2)
where k is an extrusion constant that depends on the work material,

the temperature, the die shape, and other factors. For example, to
158
Conceptual Design Case Study: Equal-Channel-Angular-Extrusion Metalworking
A0 Punch
Af F
Billet
p
Die
Figure 9.1 Schematic of conventional extrusion.
guarantee high-quality products from aluminum-based alloys, the

ratio of initial to nal cross-sectional areas (also called the extrusion
ratio) should reach about 50 to 1 or, from Eq. (9.1), 4. If a 100-
mm diameter rod is desired, then the initial diameter should be
about 700 mm. The required punch pressure (using a typical hot-
deformation extrusion constant of k 75 MPa at 350C for alu-
minum alloys) will be p = 300 MPa from Eq. (9.2), and the
corresponding punch load can reach about 12,000 tons. A press of
such capacity will be large and expensive.
It follows that when improved properties through high strains
are sought, the size of the nished product may be limited. The
problem with the conventional processes is therefore their need to
reduce the workpiece section considerably to produce a signicant
level of strain. What is sought in the present design is a new method
to deform workpieces and subject them to high strain intensities
without inducing dimensional changes.
Performance
It is desired to process relatively small workpieces, such as cir-
cular- and rectangular-section rods with diameter or side of up to
100 mm, and length-to-diameter or side ratio of up to 10 to 1. The
workpieces will typically be cast ingots, so high dimensional accu-
racy should not be expected. The machine should be capable of pro-
159
cessing various metals, such as aluminum, copper, titanium, and

their alloys, producing same-size products. The speed of processing
is not critical because the process is intended for limited production
volumes of selected alloys. From typical strain rates for the relevant
materials, it seems that a production rate of 10,000 cm3/min. would
be adequate. (This corresponds to one large 10 10 100-cm billet
every minute.)
The design should facilitate processing at various elevated tem-
peratures, adjustable up to about 1000C. To prevent the chilling
effect (cooling of the workpiece upon contact with the tool), billet
and die temperatures should be maintained to within 50C.
Loading and unloading the machine can be done manually, due to
both the limited size and weight of the workpieces and the low pro-
duction rate.
Value
The value of the design is that of producing high-quality work-
pieces while eliminating the problem of size reduction associated
with conventional processes. This means that the mechanical prop-
erties attainable with the new machine can be acquired by work-
pieces that are much larger than before. Eliminating the need for
size reduction also has the potential of utilizing a much smaller and
cheaper machine and saving energy in the processing of the materi-
als. The value of the design will be even higher if it can utilize stan-
dard metalworking equipment.
Size
The machine may be quite large and heavy, comparable to high-
capacity industrial presses used in processes such as forging and
extrusion. The only size/weight limitations may be related to allow-
able loads on reinforced-concrete oors and tting through large
doors. Transportation constraints may also dictate a modular
design.
Safety
The same precautions used with industrial presses apply to the
current design. Hot parts should be shielded, and access to moving
160
parts should be prevented during operation. Conventional safety

techniques, such as dead-man controls, should be used.
1. Deform metallic workpieces to induce high strains while main-

taining the original shape. Typical materials to be processed are
aluminum-based alloys such as 2090 aluminumlithium. These
have a ow stress f of about 50 MPa at 350C and require
working to effective strains of about 4.
2. Accept mainly rods with rectangular cross-sections. Circular
and other sections may be desirable, too.
3. Accept rod diameter or sides of up to 100 mm, and length to
diameter or side ratio of up to 10 to 1.
4. Adjust processing speed up to 100 mm/s.
5. Use hot deformation at variable temperatures up to 1000C.
Workpiece and tool temperature should be maintained
within 50C of the set temperature.
6. Loading/unloading may be manual.
The processes mentioned beforenamely, extrusion, rolling, and

forgingrepresent different forming methods that use plastic defor-
mation to produce shape changes and incorporate strain for work
hardening or grain renement. Any one of these processes can serve
as a starting point for conceptual design, with the objective of keep-
ing the strain-inducing aspect while preserving the workpiece size.
For the present example, extrusion is chosen as the starting point.
PI: The main deformation mechanism in plastic forming is shear.

Try to accomplish shear deformation without reduction in the
161
Punch
Die
First channel

Billet

Shearing plane
Second
channel
Figure 9.2 Extruding a billet through two intersecting channels of equal cross-
section.
cross-sectional area by forcing a billet through a bent channel.

The resulting simple shearing action will produce large strain,
but the overall billet size and shape will not change.
CS: Figure 9.2 is a schematic of the process. The tool is a block with
two intersecting channels of identical cross-section with an
obtuse angle 2 between them. The billet is a bar or rod, and it
ts snugly in the channels. The process begins with the well-
lubricated billet being placed in the rst channel. A punch forces
it around the corner and into the second channel. The deforma-
tion is realized by simple shearing in thin layers at the crossing
plane of the channels. In this way, all sections of the billet (except
small end parts) are deformed in the same manner. Finally, the
punch moves back, and the deformed billet is withdrawn from
the second channel.
Analysis of the velocity vectors of the material elements
while they go through the intersecting channels shows that the
effective strain experienced by the billet depends mainly on the
bend angle and is given by:

(2/3) cot (9.3)
162
which translates into the following sample values:
Bend angle, 2 Effective strain,

180 90 0
150 75 0.31
120 60 0.67
90 45 1.15
The required punch pressure (neglecting friction between work-

piece and die) can be found by energy-conservation analysis to be:

p = f = (2/3)f cot (9.4)
and the punch force F is found from the punch pressure times the
(constant) billet cross-sectional area, A:
F=pA (9.5)
For the case of extruding the aluminum-based alloy dis-

cussed in the need analysis, if the desired effective strain is about
= 4 (corresponding to a 50-fold area reduction in conventional
extrusion), we need to pass the billet four times through a 90
bend, or six times through a 120 bend, and so on. Because the
shape and dimensions of the workpiece do not change, multiple
passes can be done on the same machine. Moreover, if the extru-
sion takes place at an elevated temperature of 350C, the billet
ow stress, and thus the punch load will remain almost
unchanged even after several passes.
The required punch pressure at each pass through a 90
bend (without accounting for friction) is found from Eq. (9.4) to
be about 60 MPa, and the corresponding force on a 100-mm
diameter billet is about 50 tons from Eq. (9.5), well within the
capacity of small presses. Even though we neglected friction in
the last calculation, it is interesting to note that our earlier calcu-
lation showed that to produce the same level of strain with con-
ventional extrusion, 12,000 tons would be required!
E: The multiple-pass concept has another advantage: billet orienta-
tion can be controlled between passes. For some applications
requiring very homogeneous properties, the billet should be
163
rotated 180 after each pass, and the total number of passes
should be even. Other applications may benet from specic
microstructures and textures. Laminar and brous structures
would result from maintaining the same billet orientation dur-
ing consecutive passes.
As implied by the table above, there is a trade-off between
the number of passes required to obtain a desired effective
strain and the angle 2 between the channels. What is the opti-
mal combination?
PI: The smallest possible angle between the channels is 90. (Note
that this is the largest possible angle through which the mate-
rial can be deformed.) This angle would minimize the number
of passes required to obtain the desired effective strain and sim-
plify the design of the machine: the dies may even be installed
on a conventional vertical press. On the other hand, it would
require a larger punch force. Estimate the punch force includ-
ing friction, and if it is not prohibitively high, choose 90.
CS: Figure 9.3 shows a 90 setup. The length of the rst channel is
L, and the billet diameter is d. The punch pressure p in Eq. (9.4)
is what is required for plastic deformation. But due to Poissons
effect, a billet under longitudinal compression would also push
sideways, producing normal forces on the walls of the rst
channel, which result in a friction force. To overcome the effect
of friction, the pressure exerted by the punch should be
increased by p. For static equilibrium, the added punch pres-
sure times the cross-sectional area of the billet should equal the
friction force on the walls of the rst channel. The friction force
is not uniform along the length of the rst channel, but as a rst
approximation we shall assume that it is. The friction force is
then found from the coefcient of plastic friction () times the
normal force. The normal force can be expressed as the hydro-
static ow stress in the billet times the area of contact between
billet and channel:
p(d 2/4) = f dL (9.6)
or rearranging,
164
p+p
L
Friction
force
Figure 9.3 A right-angle extrusion die. The friction between billet and die
requires additional punch pressure p.
p = 4f (L/d) (9.7)
A similar derivation can be done for other sections. For exam-

ple, the added pressure to overcome friction with a square sec-
tion of side b is
p = 4f (L/b) (9.8)
Using a good extrusion lubricant such as graphite, with 0.1,

and assuming an L/d or L/b of 10, we get for aluminum-based
alloys (f = 50 MPa at 350C) an added pressure of 200 MPa.
Adding this to the 60 MPa calculated previously, we nd that the
total punch pressure comes up to 260 MPa. For a 100-mm diam-
eter billet this translates to a punch force of about 200 tons, and
for a 100-mm square billet the punch force is about 260 tons.
E: The previous calculation was pessimistic: if d = 100 mm and
L/d = 10, it is implied that the length of the rst channel is L =
1 m. Clearly, friction can be reduced considerably by making L
shorter, for example, two to three times the diameter d. But
even with the worst-case assumption, the 200- to 260-ton total
punch force is still reasonable. Therefore, the 90 design is kept.
165
Shortening L to reduce friction seems like a good idea, but

it presents two problems: possible buckling of the billet and the
large compressive stress in it. The latter is very serious: as the
preceding calculation shows, the punch pressure and the result-
ing compressive stress in the billet can be as much as ve times
the hot-deformation ow stress of the material. Even if friction
is reduced signicantly, the compressive stress will always
exceed the ow stress and can cause the billet material to
behave plastically outside the die.
PI: The billet must be fully contained in the die to prevent plastic
deformation of the workpiece before entering the rst channel.
Obviously this would also prevent the billet from buckling.
CS: The structure of Fig. 9.3 is made to contain the full length of a
billet whose length-to-diameter ratio is 10 to 1. The rst chan-
nel needs to be about 1-m long to accommodate the longest bil-
lets. The second channel need not be that long; two to three
times the billet diameter or side dimension seems sufcient to
assure that the billet emerges straight from the die and does not
curl upward. The punch, however, also needs to be about 1-m
long so that it will reach the bottom of the rst channel.
E: The punch is slender (with a length-to-diameter ratio as high
as 10 to 1) and may buckle. But the punch could easily be sup-
ported along its length by some external, possibly retractable,
braces, thus minimizing the free length and the danger of buck-
ling. We shall ignore the issue of punch stability for now
because the details of the die and punch design are still unclear.
The die in Fig. 9.3 seems difcult to manufacture because
of the internally intersecting channels, which may have various
cross-sections. Moreover, the design needs to accommodate
different size and shape billets, so making new dies every time
may be expensive.
PI: Reduce turnaround time and cost and simplify manufacturing
by using interchangeable inserts for the channels, and one uni-
versal die housing.
CS: Figure 9.4 shows a design consisting of a split housing in which
two inserts are tted. The inserts are square outside and have
the desired channel/billet cross-section inside (round section is
166
First channel insert Split die housing
Clamping bolt
Second
channel
insert
Figure 9.4 Extrusion channel inserts clamped between two die halves.
shown). Several bolts can be used to clamp together the hous-

ing halves and secure the inserts in place.
E: Machining of the housing halves with their sharp internal cor-
ners may still be quite expensive. Moreover, if the design of Fig.
9.4 is sized to accommodate up to 1-m long billets but is used
with much shorter workpieces, the process (placing the billet in
the rst channel, operating the press, and extracting the billet
from the second channel) will be inconvenient.
PI: Extend the modular structuring principle so that simple build-
ing blocks can be combined to form a die for billets of different
lengths.
CS: Figure 9.5 shows a modular structure made of ve plates that
can be bolted together. Channel inserts are no longer necessary
because the sides of the plates form the extrusion channels. The
design shown can accommodate billets with rectangular sec-
tions. The internal plates can be modied to t billets with
round or other sections.
E: The structure is now relatively inexpensive to make. It will also
allow disassembly and extraction of a billet in case something
167
Figure 9.5 A modular die consisting of ve plates and no inserts.
goes wrong during extrusion and a workpiece becomes stuck

inside the die. Another advantage of the modular plates struc-
ture is the ease of adapting it to extrusion through angles other
than 90 between the channels. Figure 9.6 shows how such a
change takes place by modifying two of the three internal
plates.
Because the billet is fully contained in the rst channel,
friction forces are very high (as calculated earlier). This will
require a higher-capacity press but, even more important, may
cause nonuniform deformation across the billet section. The
friction may result in different strains being produced at the
surface than at the center of the workpiece. This is a well-
known problem with extrusion and is present even when the
best available lubricants are used.
PI: As established earlier, the billet must be fully contained within
the rst channel. However, it may not have to rub against all the
surfaces of that channel, if the channel is made to move with
the billet. In this way, bad friction between the billet and
channel may be replaced by friction between the moving chan-
168
Figure 9.6 Two internal plates can be modied to produce angles other than
90 between the channels.
nel and the die housing which would not damage the work
metal.
CS: Figure 9.7 shows a design in which the rst channel consists of
three walls that move with the square-section billet, and only
one wall is stationary. (Having all four walls move with the bil-
let would make the transition into the second channel difcult,
perhaps impossible, to implement.) The sliding walls all belong
to the same partthe sliderwhich can translate downward
through an opening in the bottom of the die assembly. This
slider rubs against stationary components of the die, but
because the compressibly stressed billet is contained within it,
the normal forces exerted on the two side plates cancel each
other. The friction at the third rubbing surface (opposite the
stationary wall) is between the slider and the die, so it can be
controlled easily. (The large rubbing surface is better at pre-
venting lubrication breakdown.) Only one surface experiences
the moving billetstationary wall friction.
An implementation for nonrectangular billet sections may
be similar in nature, though different in some detail. As an
important by-product, the punch can now be made much larger
than before, having a section identical to that of the combined
169
Side plates
Stationary wall Billet Slider Punch
Billet
Opening
Figure 9.7 The slider constitutes three out of the four walls of the rst channel.
billet and slider. This would eliminate the risk of punch buck-
ling, which was identied earlier.
E: Although the total punch pressure required to overcome friction
may not have changed much, the last conguration succeeds in
signicantly reducing the direct friction on the billet, thus greatly
improving the uniformity of deformation. However, another
known problem with extrusion is that brittle materials tend to
suffer from defects and even fractures if no back pressure, or
counterpressure, is applied during the process. The back pressure
required to avoid these problems may reach about one-half the
ow stress when extruding difcult-to-deform materials.
PI: Back pressure can be applied by introducing resistance in the
outlet of the second channel. This would increase the required
punch pressure: rst, to overcome the back pressure, and sec-
ond, because of the increased friction in the second channel
due to the compressive loading on the billet. Experience with
conventional extrusion shows that the back pressure needs to
be proportional to the applied punch load but should not
170
Hydraulic Ram
Channel
cylinder insert
Throttle valve
3-position, 4-way, solenoid-actuated

directional control valve
Pump
Reservoir
Figure 9.8 Back pressure in the second extrusion channel results from oil
being forced out of the cylinder through an orice.
exceed the ow stress of the work metal. It also needs to be eas-

ily adjustable for different materials.
CS: A single-acting vented hydraulic cylinder with a ram connected
to an insert that ts in the second channel may be used for back
pressure, as shown schematically in Fig. 9.8. The amount of back
pressure can be regulated by using the proper valve and control
system. The gure shows a variable-resistance orice used to
drain the oil from the high-pressure side of the cylinder.
To estimate the size of the required cylinder, let us assume
a back pressure of one-half the ow stress, that is, 25 MPa for
aluminum-based alloys at 350C. When this pressure is applied
by a 100 100-mm channel insert, it translates to a force of 250
kN. Producing this force at the other end of the ram using a
standard 3000-psi hydraulic system will require a cylinder with
an internal diameter of 125 mm.
E: A 125-mm diameter cylinder, its associated hydraulic components,
and the control system will be bulky and expensive. A larger
171
hydraulic cylinder may be required for materials that are more dif-
cult to deform. An even more serious problem with the arrange-
ment of Fig. 9.8 is its inexibility: dimensional changes in the die
(for example, to extrude shorter workpieces) may entail reposition-
ing of the cylinder, such as mounting it higher or lower. Another,
simpler method of providing back pressure may be benecial.
PI: One of the simplest ways to produce proportional perpendicu-
lar forces is pressing down on an object that rests on an inclined
plane. If the motion of the object were constrained to be verti-
cal, the inclined base would move sideways. The downward
force exerted by the punch may be converted to a horizontally
acting back pressure by a wedge-like mechanical arrangement.
CS: Figure 9.9 shows an implementation wherein the bottom of the
second channel is dened by the top surface of a platethe
horizontal sliderwhich moves with the billet. The slider rests
on a stationary base tilted at an angle to the horizon. During
extrusion, the slider is pushed to the left by the action of the bil-
let on the obstructing step, which is built into the end of the
slider. The leftward motion of the slider is restricted by the
inclined base, so back pressure is generated. The more punch
pressure is applied and transmitted to the obstructing step by
the billet, the greater the normal force on the inclined base. A
greater normal force causes a larger friction force, resulting in
the back pressure increasing in proportion to the punch pres-
sure. The alignment of the slider is maintained by two sets of
horizontal tongue-and-groove connections to the side plates.
During operation, the whole die assembly also moves slightly
upwarda distance marked by x in Fig. 9.9on four guide
pins provided at the bottom of the assembly.
Because the horizontal slider needs to accommodate the full
length of the billet, a change is made to the structure of the rst
channel. The rst channel now consists of two (instead of three)
sliding walls and two stationary walls. Accordingly, the single
vertical slider of Fig. 9.7 is now split into two rectangular-sec-
tion sliders, each constituting one channel wall. The punch now
assumes an H-shaped cross-section, as shown in Fig. 9.10. The
reduced friction in the second channel due to the horizontal
slider partly makes up for the added friction in the rst channel.
172
Punch
press
Punch
section
Punch
Horizontal
Tongue slider Step
and
groove

Guide
pin x
Inclined base Opening
Figure 9.9 Back pressure is induced in the second channel by the wedge
effect of the horizontal slider on the inclined base.
Punch
First
channel
Stationary
wall
Vertical
slider
Figure 9.10 The walls of the rst channel are created by the two vertical slid-
ers and two stationary walls. The H-shaped punch is also shown.
173
The base of the die assembly still contains an opening to allow

the vertical sliders to descend during extrusion.
E: This seems like a good arrangement, but how will the horizon-
tal slider return to its place at the end of each extrusion process?
Gravity alone may not do the job if the current die assembly is
to operate with existing horizontal presses; that is, the extrusion
is done in a horizontal plane.
PI: Use springs to return components to their initial position.
CS: The die assembly, which rises during extrusion (through distance
x in Fig. 9.9), could be returned by springs. Figure 9.11 shows how
each of the four vertical guide pins of Fig. 9.9 is replaced by a spe-
cial spring-loaded screw, which in turn requires a cavity in the
punch-press bed. During the upward motion of the die assembly,
the springs will compress. When the punch force is removed, the
springs will extend and pull the die assembly back down. This
arrangement will work even if the die is placed on its side and
gravity does not assist the return motion.
E: Friction present between the moving parts may interfere with
the spring action and prevent the die parts from being reposi-
tioned accurately. Moreover, the vertical sliders also need to be
repositioned at the end of their stroke.
PI: Use hydraulic actuators to reposition parts. The horizontal
slider will be returned by new cylinders. The vertical sliders can
use the punch press itself for the return motion if a joint is
designed to allow pulling the slider up by the punch.
CS: To return the horizontal slider sideways and downward along
the inclined plane of the base, a pair of hydraulic cylinders is
added with a cross-bar connected to the slider, as shown in Fig.
9.12. These cylinders can be quite small and can be incorpo-
rated into the main hydraulic system of the press. In contrast to
the hydraulic cylinder of Fig. 9.8, which was mounted inde-
pendently of the die, the current pair of cylinders is attached to
the die, so they all move up and down in unison during extru-
sion. The springs proposed for incorporation into the base of
the die assembly (Fig. 9.11) are no longer required, so the orig-
inal guide pins (Fig. 9.9) can be used instead.
To return the vertical sliders upward, a hook-type connection
is implemented between the punch and the sliders. As shown in
174
Die
assembly
Figure 9.11 Spring-loaded screws replace the guide pins to reposition the
rising die assembly.
Punch
Hydraulic
cylinder
(1 of 2)
Cross-bar
Vertical
slider
(1 of 2)
Figure 9.12 Hydraulic cylinders and tongue-and-groove connections at both

ends of the punch are added to facilitate repositioning.
175
Billet Punch
(a)
(b) (c)
Figure 9.13 (a) The billet is inserted sideways and down; (b) the punch is
moved left and extrusion begins; (c) about two-thirds of the stroke is completed.
Fig. 9.13, the operator moves the punch sideways on its tongue-
and-groove connection to the punch press to allow placement of
the billet in the rst channel. The punch is then returned while
engaging it with the vertical sliders. During extrusion, the punch
176
presses down on the sliders and billet, and pulls the sliders up at
the end of the stroke.
E: The horizontal cylinders can serve two additional purposes: they
can be used for supplemental back pressure, and they can help
retract the billet at the end of the extrusion process. They only
need to be of the double-acting variety to perform all three tasks.
Billet extraction, however, may still be a problem. Even if the hor-
izontal slider is made quite long, there is no guarantee that the
billet would emerge from the second channel when the cylinders
are made to pull on the slider. In fact, the billet may remain
stuck between the three stationary walls of the second channel.
PI: Create an attachment between the billet and the horizontal
slider so that when the slider is pulled by the hydraulic cylin-
ders, the billet will follow it.
CS: A small cavity created in the horizontal slider will be lled by the
work metal when it is in a plastic state, thus creating a hook
for the slider to pull the billet. This is shown in Fig. 9.14.
E: The billet will have a small protrusion at its end, created by the
cavity in the horizontal slider. This protrusion will have to be
ground off before inserting the billet for another extrusion pass.
This solution is acceptable because the very ends of the billet
anyway do not undergo the shearing action of angular extrusion
and do not possess the improved material properties of the rest
of the billet. Still missing from the design is a provision for
maintaining the billet at constant temperature during extrusion.
PI: The extrusion process is quite fast: extrusion speeds of up to
100 mm/s can be applied by the current design when coupled
with conventional presses. The billet needs to be preheated to
the desired temperature before entering the die. The die should
also be heated to prevent the well-known chilling effect of the
billet material when contacting the die.
CS: An induction furnace to heat billets will be included in the system.
This is a standard, off-the-shelf item and need not be designed. A
heater will be added around the die, and the die assembly may
also be separated from the punch-press bed by an insulating layer.
Figure 9.15 offers an overview of the design with the added heater.
A control system will also be included to monitor the die temper-
ature and regulate the heater operation accordingly.
177
Horizontal slider Billet
Cavity/protrusion (size exaggerated)
Figure 9.14 A cavity at the front end of the horizontal slider creates a
hook in the billet to help with its extraction.
Punch
Heater
Hydraulic
cylinder
Figure 9.15 A heater is mounted around the die assembly to maintain con-
stant temperature.
E: All major issues seem to have been addressed. The device has
been designed to operate with vertical presses; however, it can
be adapted to horizontal operation. The nal conceptual design
applies to processing of rectangular-section billets. Some adap-
178
Extrude billet through a bent

channel to produce simple
shear
Block with two channels
intersecting at angle 2
What is the
Smallest angle (90) minimizes optimal angle?
number of passes but requires
more punch force
90 setup; estimated friction

High compressive is not too high
stress in billet
Fully contain billet in first outside die
channel to prevent plastic
deformation outside the die
First channel and punch made

as long as the longest billet
Die difficult to make
and not versatile
Interchangeable inserts in a
universal die housing are easy
to make and versatile
Split, bolted housing with

two channel inserts
Machining still difficult;
operation inconvenient
Modular structure from
clamped-together plates is
easy to make and operate
5 plates bolted together and no

High friction causes separate inserts
non-uniform
Replace "bad" billetdie deformation
friction with "better" friction
between die parts
Vertical slider provides 3

moving walls in first channel
Back pressure required
in second channel
Figure 9.16 Summary of the ECAE parameter analysis process. (Continues

on page 180.)
tation may be required to accommodate circular and other sec-

tions, although the overall concept is still valid.
This chapter presented a relatively complex design example whose

parameter analysis process is summarized in Fig. 9.16. It begins with
179
Introduce easily-adjustable,
load-proportional resistance
in the second channel
Hydraulic cylinder with orifice
Large, expensive valve produces back pressure
cylinder; inflexible
Inclined plane can convert mounting
punch pressure to
proportional back pressure
Horizontal slider on tilted base;

die assembly moves vertically
How is the horizontal during extrusion
slider returned?
Springs are a gravity-
independent means of
returning components
Spring-loaded screws replace

Friction during return vertical guide pins
motion; vertical sliders
Hydraulically return horizontal need to be returned too
slider; punch returns vertical
sliders
2 die-mounted actuators
connected to the horizontal
slider; hook-type connection
How is the billet between the punch and
extracted? vertical sliders
Pull billet out by the horizontal
slider
Small cavity in horizontal slider

creates "hook" in billet
Maintain desired
Preheat billet before extrusion; billet temperature
heat die to prevent chilling
effect
Induction furnace heats billet;

another heater mounted on die
Finished
Figure 9.16 Continued
an abstract need to devise a method for improving material proper-

ties by deformation processing, and it ends with a relatively detailed
design of an Equal-Channel-Angular-Extrusion (ECAE) metal-
working machine. Accordingly, some of the identied parameters
represent innovative breakthroughs, while others deal mainly with
practical aspects of structuring a viable machine.
The next design phase will add a control system for the extrusion
machine, including means of activating the horizontal cylinders and
180
sensors (e.g., micro-switches) to signal the end of the vertical and

horizontal strokes. Some heat-transfer calculations are required to
size the induction furnace and the die heater. Shields may also be
added around the die assembly to prevent the operator from acci-
dentally touching the die heater or any of the moving parts.
1. The provision made in the design of this chapter to extract the

billet from the second channel (i.e., the cavity in the horizontal
slider) is not a very elegant solution. Design a better method for
billet extraction to be incorporated in the design.
2. The advantages of using sliders, or channel walls that move with

the billet, became clear during the design process. Design a die
assembly that has at least three moving walls in the rst chan-
nel, similar to the concept of Fig. 9.7. Can a die be designed to
have four moving walls in the rst channel?
3. How would you turn angular extrusion into a continuous process,

suitable for handling very long workpieces (e.g., metal wire)? It
would be impractical to enclose very long billets in the rst chan-
nel, so the conceptual solution should be very different from the
design presented above. One direction worth exploring is an
adaptation of the Conform extrusion process, where the billet is
pulled, rather then pushed, into the die by a rotating wheel.
4. One common practice with conventional extrusion is using a

dummy billet to help extract the workpiece from the die. Use
this concept for billet extraction and modify the design of this
chapter accordingly.
5. Instead of using a dummy billet, workpieces could be fed contin-

uously so that each billet pushes its predecessor out of the second
channel. However, the design needs to be modied; for example,
another horizontal slider should take the place of the rst hori-
181
zontal slider when the latter has been displaced by a preceding

billet. Make the necessary provisions to the design to accommo-
date this semicontinuous process.
9.8 Bibliography
Some reference material may be useful in better understanding this

chapter. General books on manufacturing processes usually have
chapters on material behavior and fundamentals of extrusion. Two
such books are:
Groover, M. P. Fundamentals of Modern Manufacturing: Materials,
Processes, and Systems. Upper Saddle River, NJ: Prentice-Hall,
1996.
Kalpakjian, S. Manufacturing Engineering and Technology. 2nd
ed. Reading, MA: Addison-Wesley, 1992.
Numerous books are available on more specic topics related to this

chapter. Chapter 4 in the following book contains information on
the Conform extrusion process, which may be useful for thought
question number 3:
Blazynski, T. Z. (ed.) Design of Tools for Deformation Processes.
New York: Elsevier Applied Science Publishers, 1986.
Harris presents a more theoretical treatment of the behavior of met-

als in several manufacturing processes:
Harris, J. N. Mechanical Working of Metals: Theory and Practice.
Elmsford, NY: Pergamon Press, 1983.
The design presented in this chapter is based to a large extent on the

following patent:
Segal, V., Goforth, R. E., and Hartwig, K. T. Apparatus and
Method for Deformation Processing of Metals, Ceramics, Plastics
and Other Materials. U.S. Patent 5,400,633, 1995.
Several papers describing Equal-Channel-Angular-Extrusion follow.

The rst is an overview of the process, and the others mostly
demonstrate its benets when applied to specic materials:
182
Segal, V. M. Materials Processing by Simple Shear. Materials

Science and Engineering A197, (1995): 157164.
Segal, V. M., Goforth, R. E., and Hartwig, K. T. The Application
of Equal Channel Angular Extrusion to Produce Extraordinary
Properties in Advanced Metallic Materials. Proceedings of First
International Conference on Processing Materials for Properties,
Hawaii, November 710, 1993, pp. 971974.
Goforth, R. E., Segal, V. M., Hartwig, K. T., and Ferrasse, S.
Production of Submicron-Grained Structure in Aluminum
3003 by ECAE. Proceedings of Superplasticity and Superplastic
Forming Conference, Las Vegas, February 1315, 1995, pp. 2532.
Semiatin, S. L., Segal, V. M., Goetz, R. L., Goforth, R. E., and
Hartwig, K. T. Workability of a Gamma Titanium Aluminide
Alloy During Equal Channel Angular Extrusion. Scripta
Metallurgica et Materialia 33, No. 4 (1995): 535540.
183
10
Need Analysis and Conceptual
Design Case Study: Ball Mover
A group of second-year engineering students applied the need analy-

sis and parameter analysis methodologies to design an entry in the
1997 American Society of Mechanical Engineers (ASME) Student
Design Contest. The objective of the competition was to automati-
cally transport two ping-pong balls and one golf ball between two
locations. This case study describes the valuable insights gained by
these students during the need analysis stage and the implication of
these insights for the conceptual design process. It also highlights the
natural tendency of designers to subconsciously make implicit
assumptions and not recognize them throughout the design process.
These assumptions often lead to conceptual oversights and may cost
the design its competitive edge.
Most mechanical designers handle the challenge of transporting

objects from one location to another quite easily. The 1997 ASME
Student Design Contest provided an opportunity for student mem-
bers to learn some of the intricacies of designing material handling
systems. The objective of the competition was to design and build a
device to safely transport two ping-pong balls and one golf ball from
the top of a platform to a receiving box separated by a given dis-
tance, as shown in Fig. 10.1. Contest regulations allowed the com-
184
Need Analysis and Conceptual Design Case Study: Ball Mover
Golf ball
Ping-pong balls
ce
tan
Dis Platform
Contest surface
Receiving box
Figure 10.1 Setup for the 1997 ASME Student Design Contest.
petitors to choose their own distance anywhere between 50 and

211.8 cm. The score would be computed as
Score = Average velocity = d/T
where d is the distance in cm between the platform and the receiving

box and T is the total time in seconds from the start signal to when the
balls come to complete rest in the receiving box. Note that the goal
was not to transport the balls in the shortest time over a set distance,
but rather to transport balls at the highest average velocity. This means
that the competitors could choose the desired distance (within the
specied limits) for which the ratio of distance to time should be
maximized. Key competition guidelines are summarized as follows:
1. The system should be powered by a single 1.5 V size AA alka-

line battery and a single specied dc motor. The motor can-
not be modied, and additional energy sources or motors are
not permitted.
2. In all energy storage devices except for the battery, the energy
stored at the end of the run should be greater than or equal
to the energy at the start.
185
3. After the start of the run, any external communication,

interaction, or inuence is not allowed; that is, the system
must be autonomous.
4. The complete system must t in a 16 16 32-cm box
before assembly or placement. If assembly is required, the
system must be put together and ready for the run in ve
minutes. The assembled system can only contact the contest
surface, the receiving box, and the platform.
5. The outside dimensions of both the platform and the box are
16 16 32 cm. They are constructed of 1/4 plywood. The
balls initially rest in 1/2-diameter holes on the top of the
platform, with 8 centimeters from center to center.
6. The system must deliver the three balls from the platform
into the receiving box within one minute of the start signal.
Designing a device to transport the balls was relatively easy.

However, the following need identication and analysis indicate that
the real challenge lay in doing so at the highest average velocity.
10.2 Need Identication
The students used the black-box technique to precisely identify the

need, as shown in Fig. 10.2. The balls on the platform, the electrical
energy from the battery, and the start signal from the user were the
three inputs to the black box. The desired output was the balls rest-
ing in the receiving box. The designed device had to transform the
three inputs to the required output. The need was restated as
transport the balls from the platform to the box using the energy
from the battery at the highest average velocity. Note that the stu-
dents consciously avoided the statement move the balls at the
maximum velocity because it related to only one part of the total
time (total time = initial time before the motion starts + accelera-
tion time + time of motion at constant velocity + deceleration time
to full stop).
Some of the questions that came to the students minds while
dening the need were:
186
Balls on the platform
Electrical energy Design Balls in the

from the battery Artifact receiving box
Start signal
Figure 10.2 Black-box description of the design artifact.
What is the ideal distance between the platform and the

receiving box?
Is it advantageous to go for the full distance?
What is the ideal or theoretically highest possible score?
What are the characteristics of an ideal solution?
What are the important hurdles to achieving the highest
score? What is preventing us from creating an ideal solution?
The students performed the following need analysis to answer these

and other questions, gain deeper understanding of the design task,
and generate the design requirements.
10.3 Need Analysis
Performance
The primary objective is to create a design that gives the high-
est score or performance. The score is a function of the distance (a
variable that can be chosen to optimize the score) and the total
time to transport the balls. As mentioned earlier, the total time can
be divided into initial time before the motion starts, acceleration
time, time of motion at constant velocity, and deceleration time to
full stop.
For establishing targets, it is important to get a rough idea of the
ideal score. To simplify the calculations at this stage, the deceleration
time to full stop, and the distance traveled by the balls during decel-
eration are assumed to be zero. The transportation of the balls
involves energy, so the initial calculation can use energy balance.
187
Energy from the battery can be converted to potential and kinetic

energies of various components:
EBattery = PE + KE
where is the combined efciency of energy conversion (from elec-

trical to mechanical) and transfer (from the battery to the motor
shaft and from there to the balls and other moving parts), EBattery is
the energy output from the battery, PE is the potential energy
increase in various components, and KE is the kinetic energy given to
any moving object. The energy output from the battery is given by
EBattery = P t
where P is the power output from the battery and t is the duration
of energy output. The kinetic energy gained by all the moving com-
ponents is:
KE = (1/2)mMovingMassv 2
where mMovingMass is the total mass of all moving objects and v is

their instantaneous velocity. (At this point, it is assumed that all
moving objects have the same velocity.) Combining the last three
equations gives
P t = PE + (1/2)mMovingMassv 2
and rearranging the terms shows the instantaneous velocity to be
v = 2 ( P t PE)/mMovingMass
The last expression means that the instantaneous velocity can be

increased by:
Increasing the duration, t, of power output from the battery.

Note that the maximum duration is equal to the total time for
the run, T.
188
Eliminating any potential energy increase. This can be done

by not lifting any component from its original position.
Decreasing the moving mass, mMovingMass, by eliminating any
motion except for that of the balls. In this ideal situation, the
moving mass will be equal to the mass of the balls only.
Increasing the efciency of energy conversion and transfer,
mostly by reducing frictional losses.
The instantaneous velocity is the velocity achieved by the balls

after the energy is supplied for a certain duration of time. However,
our interest is in determining the average velocity (d/T) and the
optimal value for the distance (d). The average velocity will depend
on whether the energy output from the battery is for a brief period
at the start of motion or continuously throughout the run. Solutions
that employ the former strategy will convert the electrical energy to
mechanical energy and store it for a short period of time during
which the balls do not move. The stored energy will then be released
to increase the velocity of the balls.
The velocity prole in Fig. 10.3a shows the short idle period at
the beginning of the run, the sharp increase in velocity as energy is
supplied to the balls, and the gradual drop in velocity due to fric-
tional losses. The area under this curve represents the distance trav-
eled. As can be seen, increasing the time beyond the start of motion
increases the distance covered. The average velocity prole (Fig.
10.3b) is calculated as the ratio of this distance to the total time. It
shows that when the distance is short, the initial idle time con-
tributes signicantly to the total time, thus reducing the average
velocity. Increasing the distance initially increases the average veloc-
ity, but beyond a certain point, frictional losses reduce it.
Increasing the initial idle time before the motion starts increases
not only the peak velocity, but also the total time. Therefore, it
adversely affects the nal score. On the other hand, decreasing the
initial time reduces the velocity (because there is less time for the
electrical-mechanical energy conversion), so it also has a negative
effect on the nal score. Any solution that employs this strategy
entails a trade-off between the initial idle time and the peak veloc-
ity, so this solution strategy is discarded.
189
Average Velocity
Velocity
Time Distance
(a) (b)
Figure 10.3 (a) Instantaneous velocity prole and (b) average velocity pro-
le for the initial energy output strategy.
The velocity prole for designs that employ the second solution
strategythat of continuously accelerating the ballscan be
obtained by plotting the instantaneous velocity at the end of differ-
ent time periods of energy transfer, as shown in Fig. 10.4a. The area
under the curve is equal to the distance traveled by the balls, and this
is shown in Fig. 10.4b. Because the balls cover greater distance in a
shorter time interval during the later part of the run, increasing the
distance between the platform and the receiving box will increase
the average velocity. This insight is depicted graphically in Fig. 10.4c.
Additional calculations also show that using this strategy, the balls
can be moved from the platform to the receiving box within one sec-
ond. A target of one second is therefore set for transporting the balls.
The understanding gained so far from the above analysis can be
summarized as the following conclusions:
Continuously accelerate the balls for an optimal score. In

other words, keep drawing energy from the battery for the
entire duration of the run.
Use the maximum allowable distance.
Reduce the moving mass.
Avoid increasing potential energy of components.
Increase the efciency of energy conversion and transfer.
Value
The value of a successful design cannot be estimated, in this par-
ticular case, in monetary terms. The competition is a valuable experi-
190
Displacement
Velocity
Time Time
(a) (b)
Average Velocity
Distance
(c)
Figure 10.4 (a) Velocity, (b) displacement, and (c) average velocity proles
for the continuous acceleration strategy.
ence for students because it exposes them to various facets of the

design process. Students can also evaluate their design skills when
competing at the regional and national levels, and they take great pride
in their achievements. From the cost viewpoint, a budget of $100 was
allocated to purchase off-the-shelf components and materials.
Size
The competition rules specify that the complete system must t
in a 16 16 32-cm box. The dimensions of the platform and the
receiving box, the initial ball locations, and other geometrical data
are as described in Section 10.1.
Safety
Although the design uses electrical energy, there is no electrical
hazard. While the balls are moving, they might bounce uncontrol-
lably onto the contest surface, resulting in an ineligible run. This
rule simulates a drop from the material-handling system that would
191
damage the product in real-world situations and might also be

unsafe. Competition guidelines explicitly state that the balls cannot
be damaged during the run.
Special
The device must meet the ASME Student Design Competition
rules and regulations. Some requirements in this category include:
Setup (assembly) time should be under 5 minutes, and all the

team members may participate in it.
The device can contact only the platform, the receiving box,
and the contest surfacebut not the ballsbefore the start of
the run.
There should be no residue left on the balls at the end of the
run.
10.4 Design Requirements
Table 10.1 summarizes the requirements generated in each of the

need analysis categories. The students not only identied the real
need but also made important decisions. For instance, based on the
calculations in the performance category, the target time for the
complete run is set to 1 second, and the distance between the plat-
form and the receiving box is set at 211.8 cm.
In the need analysis stage, the designers identied the need for con-
tinuously moving the balls to achieve a high score. Three prelimi-
nary ideas for designs that continuously accelerate the balls are as
follows:
1. Drawbridge. A drawbridge that connects the platform and

receiving box can be raised at the platform end. The balls will
roll down the drawbridge and into the receiving box due to
gravity.
192
Table 10.1 The ball-mover requirements list.
Category Design Requirements

Performance 1. The device must move two ping-pong balls and one golf
ball in less than 1 minute. Target time is 1 second.
2. The target distance between the platform and the box is
211.8 cm.
3. Power source: One size AA alkaline battery (rated volt-
age: 1.5 V, rated current: 1 A).
4. Motor: Radio Shack dc motor model 273223 (no load
speed: 5,700 rpm, max. efciency: 45% at 4100 rpm,
power used at max. efciency: 0.39 W).
5. Ping-pong ball: Mass: 2.5 g. Diameter: 38 mm (1.5 in.).
6. Golf ball: Mass: 45 g. Diameter: 41 mm (1.6 in.).
7. The energy stored in all energy storage devices (except
for the battery) at the end of the run should be greater
than or equal to the energy at the start.
8. The system must be autonomous after the start; that is,
external communication, interaction, or inuence of any
kind is not permitted.
Value 9. Cost of materials and purchased items should be under
$100.
Size 10. The complete system must t in a 16 16 32 cm box.
11. Platform dimensions: 16 16 32 cm.
12. Receiving box dimensions: 16 16 32 cm.
13. The balls initially rest in 1/2 holes centered on the top
of the platform 8 cm apart from center to center.
Safety 14. Balls or any other part of the mechanism should not
bounce in an uncontrollable manner.
15. Balls should not be damaged in any way.
Special 16. Setup (assembly) time should be under 5 minutes.
17. Number of people available for setup: 4.
18. The device can only contact the platform, the box, and
the contest surface at the start of the run.
19. The device should meet the 1997 ASME Student Design
Contest rules and regulations.
20. There should be no residue on the balls at the end of
the run.
2. Puller rod. A xed bridge can connect the platform with the
receiving box. A cross-bar, or puller rod, moves the balls
along the bridge and into the receiving box.
3. Cable car. A transport device will pick up the balls, carry
them on a cable running overhead from the platform, and
release the balls into the receiving box.
193
The rst two concepts served as initial conditions for the two
parameter analysis processes described below. Note how the draw-
bridge concept seems to contradict the eliminate any potential
energy increase conclusion from the need analysis.
The nal score in the competition also depends on the stopping
time. Two concepts for bringing the balls to rest are:
1. Damping. The kinetic energy of the balls is gradually dissi-

pated until they stop completely.
2. Capturing. The balls are arrested in a contraption that pre-
vents their motion.
10.6 Parameter Analysis of Conceptual Design I
PI: Continuously move the balls using the drawbridge idea.

CS: An initial sketch for this concept is shown in Fig. 10.5. The
motor mounted on a support tower lifts the drawbridge using a
cable (or a shline). The balls roll down the drawbridge into the
receiving box. Holes or slots (not shown in the gure) can be
machined in the drawbridge so that the balls do not contact it at
the start of the run, as required by the contest regulations. The
raised longitudinal edges of the drawbridge prevent the balls
from falling off.
E: Several key problems with this design are:
A signicant portion of the energy goes into lifting the

weight of the bridge.
Because the support tower and the drawbridge must t in
the box, they should be assembled from smaller compo-
nents. This problem seems to be easy to solve by modify-
ing the design.
There is no stopping mechanism for the balls.
The rst problem is the most difcult because it is inherent to

the drawbridge concept and can potentially render the whole
idea nonviable.
194
Cable
Balls
Motor
Bottom support
Motor
Cable Drawbridge
Support
tower
Platform Contest surface Receiving box
Figure 10.5 The initial conguration for the drawbridge idea to move the balls.
PI: The concept of counterbalancing is used in some designs, such

as elevators, to reduce the required lifting force. Apply this idea
to the drawbridge.
CS: The cable used to raise the drawbridge (see Fig. 10.5) is made
to go over a pulley mounted on the motor shaft and attach to a
counterweight at the other end. The weight of the drawbridge
can thus be balanced, and the motor would not need to lift the
weight of the drawbridge.
E: The motor must now overcome the inertia of not only the
drawbridge, but also the counterweight. This will result in
slower acceleration at the beginning of the run and adversely
affect the total time and score. Based on this insight, the con-
cept of counterbalancing the drawbridge is discarded.
PI: Decrease the energy required for lifting the drawbridge by
reducing its weight.
CS: The drawbridge can be constructed of berglass fabric stiffened
with epoxy. The composite sheet can be selectively stiffened so
that it can be folded to meet the size requirement. It can also be
shaped like a converging channel, or trough, for the balls to roll in.
195
E: The students built several models and realized that the design
modications were difcult and expensive to incorporate due
to the required curing time of the composites and the cost of
the materials. Even though the concept seemed very attractive,
it was abandoned due to the implementation difculties.
PI: Reduce the weight of the bridge by using a plain fabric.
CS: A lightweight fabric can be used to connect the platform and the
receiving box. When the cloth is lifted by a shline connected to
its two corners at the platform end, the balls will roll down the
slope. Because the fabric is exible, it automatically forms a chan-
nel to guide the balls. A sketch of this idea is shown in Fig. 10.6.
E: This design effectively reduces the weight. The students per-
formed a simple experiment wherein they lifted the fabric at
the platform end using the motor. The balls rolled down the
fabric and moved the full distance in about 8 seconds. The
experiment revealed an important problem: The fabric sagged
due to the weight of the golf ball, as shown in Fig. 10.7. As a
result, the slope ahead of the balls sometimes became negative,
and the balls decelerated in the nal stages of the run.
PI: Prevent sagging by reducing the stretch of the fabric.
CS: The following modications are incorporated into the design:
The fabric is changed to a thin perforated plastic sheet.

The plastic sheet stretches to a lesser extent under the
golfball weight. An additional advantage of the perforated
plastic sheet is that it reduces the rolling resistance and the
air drag associated with the lifting.
A thick plastic backup sheet is added. It connects the plat-
form with the receiving box and is not raised during the
run. It is rmly attached to prevent sagging around the
receiving box end of the run, as shown in Fig. 10.8.
E: There is no mechanism to stop the balls when they arrive in the

receiving box.
PI: Stop the balls by using the damping idea listed in the technol-
ogy identication section.
CS: Several damping techniques to remove energy from the balls
were tested. Students found that a tray containing very ne
196
Cable
Balls
Motor
Bottom support
Motor
Section A-A
Cable Cloth
A B
Support C Receiving box
tower Section B-B
A B
C
Section C-C
Platform Contest surface
Figure 10.6 A cloth connects the platform with the receiving box, naturally
forming a channel to guide the balls.
Direction of lift
Direction of motion
Figure 10.7 The sagging due to the golf ball creates a negative slope in
front of the balls.
powder packed in thin plastic wrap was the most effective

method for damping the motion. The plastic wrap prevented
the balls from picking up any residue, which was one of the
design requirements.
E: Several problems persisted throughout the development of this
concept:
The balls moved unpredictably in the initial stages of the

run, when they rst touched each other in the converging
channel. Occasionally, they collided with each other and
jumped out of the channel onto the contest surface.
197
Thin perforated
plastic sheet
Ball
Receiving
Thick backup
box
plastic sheet
Figure 10.8 A plastic backup sheet reduces the sagging problem.
The sagging problem still persisted around the middle of

the span.
To continuously accelerate the balls, the required height
was more than 8 ft. The assembly time requirement and
stability considerations limited the maximum height to
about 6 ft.
The balls took approximately 2 seconds to stop in the
damping tray due primarily to their rolling motion.
Depending on the material in the wrap, either the ping-
pong balls or the golf ball took more time to come to
complete stop.
Looking at these problems, it was decided to abandon the draw-

bridge concept entirely. Important insights gained during the con-
cept development are:
Moving the balls individually provides better control of the

balls. For instance, different damping methods can be applied
to different balls.
The rolling motion of the balls contributes signicantly to the
stopping time and thus to the total time.
Figure 10.9 shows a summary of this parameter analysis process.

The decision to relinquish this concept, shown as a dead-end in the
last evaluation step, led to backtracking all the way to the technology
identication stage. Starting the second parameter analysis process
with a different initial concept (but with some new insights) is
described in the next section.
198
Continuously move the balls Drawbridge lifted up by

using the drawbridge idea a motor which is mounted
on a support tower
Energy required for
lifting the drawbridge
Reduce the required force
by using the concept of Drawbridge with
counterbalance counterweight
Increased inertia
Reduce the weight of

the drawbridge Composite materials
Difficulty in incorporating
design modifications
Reduce the weight by Cloth connects
using a plain fabric the platform with the
receiving box
Sag due to the weight
of the balls
Prevent sag by Thin perforated plastic sheet
reducing the stretch of to move the balls. Thick
the fabric backup sheet to
reduce the sag
No stopping mechanism
Damp the motion of Damping tray with very fine

the balls powder packed in
thin plastic sheet
1. Unpredictable motion of the balls.

2. Persistence of sag.
3. Required height of the support tower.
4. Rolling of the balls in the tray.
DEAD-END

design I.
10.7 Parameter Analysis of Conceptual Design II
PI: Move the balls individually from the platform to the receiving
box using a puller rod.
CS: Three sets of tracks, one for each ball, are shown in Fig. 10.10.
The tracks are rmly supported. The motor is mounted on the
receiving box, and a shline runs from the motor pulley along
the middle tracks to the puller rod. The guides on the puller rod
199
Puller rod
Guide
Tracks
Track support
Motor Fishline
Figure 10.10 The initial conguration for the puller-rod concept.
prevent it from inadvertently leaving the tracks. When the

motor turns, it rotates the pulley and winds the shline. The
shline pulls the puller rod toward the receiving box and, in the
process, rolls the balls along the track. To minimize the moving
mass, the puller rod is made of thin aluminum tubing.
E: No stopping mechanism exists.
PI: One of the insights gained in the previous conceptual design
process is that the rolling motion during the stopping phase
contributes signicantly to the total time. Based on this insight,
rolling motion during the stopping of the balls should be pre-
vented.
CS: A rather simple solution to address this parameter is shown in
Fig. 10.11. At the end of the run, each ball falls into its own cav-
ity. The shape of the cavities is determined by using the princi-
ple of kinematic constraints. It requires three noncollinear
point contacts to fully constrain the ball from rolling. A trihe-
dral cavity formed by three planar surfaces provides such con-
tacts and constrains the ball. The balls come to a complete rest
in less than 0.1 second.
E: The layout of the overall system indicates that it would function
well. However, the system does not meet the size requirement.
200
Surface 2
Ball
Surface 1 Surface 3
Figure 10.11 A ball rests in the trihedral cavity formed by three planar
surfaces.
PI: Create modular subassemblies that t in the 16 16 32-cm

box.
CS: Each track is assembled from 31-cm long, 1/4-diameter alu-
minum tubing. Dowel pins are used to connect the sections
together. The track supports are assembled from three wooden
blocks. The trihedral cavity is created on-site by folding a card-
board sheet and xing it in the wooden holder. All these mod-
ications are shown in Fig. 10.12.
E: A small section of the track was built to identify potential prob-
lems. During the testing, the students noticed that the track rods
moved in the transverse direction due to the weight of the golf ball.
PI: Prevent the transverse motion of the track rods.
CS: Slots are cut into the track rods, and dowel pins are inserted on
top of the horizontal support. When assembled, the dowel pins
t into the slots in the tracks and prevent the transverse motion
of the tracks, as shown in Fig. 10.13.
E: The puller rod does not stop at the end of the run.
PI: Stop the puller rod at the end of the run using mechanical stops.
CS: Two rods can be incorporated into the holder of the trihedral
cavities. The rods protrude from the top of the receiving box
and stop the puller rod.
E: When the motion was taped with a high-speed video camera,
the puller rod was observed to lag behind the balls in the nal
201
Track rod Folds
Dowel pin
Track rod
Cardboard sheet
(a) (c)
Horizontal
support block
Dowel (d)
pins
Vertical
support
blocks
(b) (e)
Figure 10.12 (a) Track rod sections are connected by dowel pins. (b) Track
supports consist of one horizontal and two vertical blocks. (c) The cardboard
sheet is folded into a trihedral cavity. (d) The cavities holder. (e) The arrange-
ment of tetrahedral cavities when assembled in the holder.
stages of the run. In other words, the acceleration prole did

not match the ideal case: the puller rod accelerated at the begin-
ning only instead of continuously throughout the run.
PI: The speed of the puller rod is determined by the motor speed.
The puller rod stops accelerating when the motor reaches its
rated speed. The puller rod should be made to keep accelerating.
CS: A new custom pulley, shown in Fig. 10.14, was designed and
built. The effective diameter of the pulley increases as the shline
is wound around it, resulting in ever-increasing tangential veloc-
ity, that is, continuous acceleration of the puller rod.
202
Dowel pin
Slot
Track rod
Horizontal
support block
Figure 10.13 When assembled, the dowel pins prevent the transverse
motion of the tracks.
Figure 10.14 Modied pulley geometry.
E: Further review of the videotape revealed that the initial accelera-

tion was rather low due to the inertia of the moving components.
PI: Improve initial acceleration by increasing the initial torque.
CS: A small capacitor was added in parallel with the battery. Initially,
the motor draws its energy from both the battery and the capac-
itor. The capacitor charges very quickly during the stopping
motion of the balls.
E: The conceptual design was considered complete at this point,
since the major issues had been addressed and initial tests
showed that the total time was less than 1 second.
The overall assembly developed in this conceptual design process

is shown in Fig. 10.15. Figure 10.16 is a summary of the parameter
analysis process.
203
Mechanical stops
for the puller rod
Figure 10.15 Overview of the puller-rod system shown at the end of the run.
Need analysis had a profound impact on the nal design. The students
systematically incorporated the characteristics of an ideal solution into
their conceptual design using parameter analysis. The need analysis
calculations provided the team with the ideal or theoretical scores.
During the entire design process, they continuously checked the true
performance against the ideal performance and strove to match the
two. Continually evaluating the current score against the ideal one dur-
ing the design process helped the team to assess their progress. The stu-
dent designers entered the ASME Region VII Competition and won
rst place by transporting the balls over the full distance in under 1 sec-
ond. Later they nished fourth at the ASME Winter Annual Meeting
where thirteen ASME regional winners competed.
Let us look at one error committed by these students. In the
kinetic energy calculation, they inadvertently ignored the angular
velocity term. The balls roll when they are dragged from the plat-
form, and they gain not only linear momentum but also angular
momentum. Therefore, part of the energy output from the battery
is wasted. An ideal solution should transport the balls linearly with-
out spinning. This conceptual mistake probably cost the students
rst place. It could have been avoided by careful examination of
204
Move the balls individually Puller rod drags the balls on

using a puller rod individual tracks
No stopping mechanism
Prevent the rolling motion
of the balls Trihedral cavities
Does not meet

size constraints
Create modular Modular components for
subassemblies tracks and track supports.
Foldable trihedral cavity
Transverse movement of
the tracks
Prevent the transverse Connect the tracks to
motion of the track the track support using
dowel pins
Puller rod does not
stop at the end of the run
Stop the puller rod Mechanical stops are
using mechanical stops incorporated into the
trihedral cavity holder
Puller rod accelerates only
in the initial stages
Continuously accelerate
the puller rod Custom pulley
Low initial acceleration

Improve the initial
acceleration by increasing Capacitor in parallel
the initial torque with the battery
Finished

design II.
implicit assumptions and development of more different conceptual

designs to gain additional insights.
1. Estimate the theoretical or ideal shortest time for moving the

balls over the full 211.8-cm distance.
205
2. Estimate the percentage of energy lost due to the rolling motion

of the balls.
3. Improve the design of the puller-rod ball mover by eliminating

the spinning of the balls during transportation.
4. One possible initial idea for a ball mover, listed in the technology
identication section, was a cable car. Develop this concept using
parameter analysis.
5. Design a new ball mover device. Incorporate the new insights

gained in this case study.
206
11
Technology Observation
This chapter proposes and demonstrates a particular approach to

observing technology as a means of enhancing the conceptual
design abilities of engineers. Designers can improve their skills by
continuously understanding the underlying concepts in existing
designs. For this process to be benecial, however, it must be carried
out at the appropriate level. The designer who does that will be able
to develop a bag of tricks for future use.
11.1 Improving Design Abilities
Parameter analysis is not only a framework for carrying out innova-

tive design, but also a method of developing skills in designers. In
the conceptual design process, nothing can replace that spark of cre-
ativity which ignites good ideas. Some people are naturally blessed
with more creative capabilities than others, but all designers need to
practice the skill of conceptual design in order to develop it to its full
potential. This is analogous to a naturally able athlete persevering
through many hours of practice to attain a competitively high level
of performance or to a famous actor who keeps attending drama
classes throughout his professional life. In other words, a good
designer must practice continuously to develop his or her skills.
Every coach has a method by which he trains his athletes, and every
drama teacher uses a particular technique of drama instruction. In
207
a similar way, parameter analysis forms a coaching method that can

sharpen the designers inventive talents and ignite the spark of cre-
ativity more regularly.
The base on which the design methodology rests is an approach
we simply call technology observation to indicate its purpose. This is
not a specic step to be applied at a particular instance during
design, but rather the continuous process of observing and analyz-
ing existing technological products to study what others are doing.
The important aspect of this observation is understanding how and
why rather than merely what has been done. By observing technol-
ogy in a particular way, over the years the designer will accumulate
a knowledge base, or bag of tricks, that consists of understanding the
underlying features and concepts of congurations and phenom-
ena, as opposed to details of particular designs. When observed
technology is applied to creating a new design, these underlying
principles, in contrast to design specics, will allow the designer to
draw useful analogies and gain insight into the task at hand.
Technology observation is similar to reverse engineering, but
it needs to be done at a level commensurate with conceptual design
by parameter analysis. This means that just as parametersthe con-
cepts and underlying principlesdrive the creation of congura-
tions during design, so they also are the targets when the process is
reversed, and existing congurations are observed. It is not enough
to study products at the level of what they do, but rather to identify
their key parameters by analyzing what is critical to their successful
operation. In other words, we should try to gain the same insight
into the design task of the observed product as the original designer
had. An example of technology observation as applied to gas water
heaters is presented next.
11.2 Domestic Gas Water Heaters Example
An inventor proposes a low-cost modication to present gas hot

water heater congurations, which will give signicant increases in
the overall efciency of operation. An increase on the order of 10%
is claimed, with very strong emphasis on the low-cost specication
208
and its implied simplicity. Let us perform technology observation

on conventional gas water heaters and the modied version in order
to demonstrate the level of understanding at which such process
should take place.
Figure 11.1a is a schematic of a typical natural gas hot water
heater. Combustion takes place in a burner at the bottom of the
storage tank. Air is drawn in around the burner, and the combustion
products, as well as excess air, ow up the stack. The stack begins as
a tube passing through the middle of the tank. A typical unit will
have a sheet metal insert in this tube which reaches down to about
the beginning of the tubular section in the tank. The tank is sur-
rounded by berglass insulation and a sheet metal cover over the
insulation. Inlet and outlet pipes for cold and hot water are also
shown on the schematic. Control of the burner is provided by a
thermostatic system that acts to keep the water in the tank at a spec-
ied temperature.
The schematic in Fig. 11.1b describes the new design to be stud-
ied. The conguration created by the inventor consists of an
expanded stack near the top of the water heater, lled with a large
number of ns, each in intimate contact with the wall of the tank.
These metal ns provide a large increase in the heat transfer surface
in contact with the hot ue gases. Except for a slight decrease in the
volume of water stored and the fact that the inlet and outlet pipes
were moved during the design change, the new water heater cong-
uration is quite similar to the present system.
In this example, the recognition of the primary feature of the
design task, which the inventor is attempting to take advantage of, is
relatively straightforward. By increasing the heat transfer surface by
a very signicant percentage, the inventor hopes that more of the
heat that is contained in the stack gases will be transferred to the
enlarged surface area of the stack, thereby increasing the efciency
of the water heater. The principal mechanism by which this is to be
accomplished is, of course, convective heat transfer.
In order to gain a better understanding of the invention, we will
investigate briey the convective heat transfer that takes place in the
unmodied water heater (Fig. 11.1a). For purposes of approxima-
tion, we will consider a water heater of the following size. The natu-
209
Figure 11.1 Schematics of (a) a conventional gas water heater and (b) the
new design.
ral gas is owing at a rate of 50,000 Btu equivalent heating value per
hour, with the nominal efciency of the water heater at 70%. The
stack is about 4 feet high and 4 inches in diameter, and the tank
holds about 100 gallons of water. We will assume that combustion is
taking place with about 100% excess air.
The amount of heat transferred convectively is determined by
the following equation:

Q = h A Tm

where Q is heat ow in Btu per hour, h is the heat transfer coef-
cient in Btu per square foot hour degree F, A is the surface area of
the stack, and Tm is the log mean temperature difference. Tm can
be found from
T1 T2
Tm =
ln(T1/T2)
210
where T1 is the temperature difference between the hottest part of

the stack and its corresponding wall temperature, and T2 is the
temperature difference between the top of the stack and its corre-
sponding wall temperature. Using typical values for temperatures,
which are about 2500F at the bottom of the stack and 500F at the
top of the stack, we nd that in order to transfer 35,000 Btu per hour
to the water, we require that h take on a value of over 6 Btu per
square foot hour degree F. Either from experience or a little investi-
gation, we know that such a value for h is unrealistically high.
However, to make sure of this assertion, we can estimate what h
might be. The following equation is an approximate relationship for
h as a function of the velocity of the gases and the stack diameter:
V 0.75
h = 0.43
D 0.25
V in this equation is the velocity of the free air volumethat is, the
velocity if the stack gases were at standard pressure and tempera-
ture. The units for V are feet per second and for D, feet. Using this
approximation, for a water heater sized as we assumed, we nd that
h is a little over 1.5. Thus, we nd a very signicant difference
between the amount of heat transferred convectively and the total
amount of heat transferred to the water.
A quick estimate tells us that there is enough energy in the stack
gases leaving the water heater to more than yield the efciency
increase required, while keeping the ue gases above the condensa-
tion temperature of the water vapor produced during combustion
in order to avoid corrosion within the system. In briey investigat-
ing the principal mechanism of the new invention, we have been led
to the conclusion that the majority of the heat transferred to the water
is done so by some other mechanism. This may be a surprising result
in light of the general shape of the water heater conguration and
our normal intuition.
Three other mechanisms may account for this heat transfer. The
rst is radiation from solid parts near the burner; the second is radi-
ation from the gases themselves; and the third is conduction. We
know that the conduction phenomenon is small compared to other
211
mechanisms in this case, and thus we are led to investigate radiative

heat transfer. The following equation is an approximate relationship
for the heat transfer per unit area due to radiation at temperature T1
being absorbed at temperature T2, where T1 and T2 are in degrees
Rankine:

[( ) ( ) ]
Q T1 4 T2 4
= 0.15
A 100 100
A quick calculation shows us that a tremendous amount of heat can

be transferred radiatively, particularly at the bottom of the tank,

owing to the dependence of Q on the fourth power of temperature.
(In fact, visual observation of the water heater in operation indicates
that the principal function of the stack insert is to provide a sheet
metal area that can glow at a high temperature and thus produce
radiative heat transfer.) Finally, information on the amount of
energy transferred radiatively by the combustion products can be
obtained in handbooks. For our problem, this will be relatively con-
stant regardless of conguration and so will not be pursued further
here.
Quite obviously, the mechanisms of operation in a gas hot water
heater are interrelated, thus producing a rather complex situation.
During technology observation, we would like to gain further infor-
mation on the relative magnitude of these interrelationships.
Among the results pertinent to our example, we nd the following:
1. An estimate of the area necessary for the additional convec-

tive heat transfer indicates that a nearly 200% increase in the
total area is required. The temperature of the stack gases still
remains above the condensation point.
2. An additional 1.5 inches of berglass insulation can yield an
additional 5% in overall efciency. Additions of even greater
thickness of insulation, of course, yield diminishing returns.
Any changes in the thickness of insulation require the mod-
ication of the outer shell of the hot water heater and must
be evaluated carefully.
212
3. We recall that extremely sharp points in a ame glow at very

high temperatures, and this phenomenon leads us to investi-
gate further the possibilities of radiation enhancement as an
alternative to the basic concept represented in Fig. 11.1b. The
addition of radiating material into the burner area is a very
minor modication, and thus such an alternative is a factor
in determining the potential of this invention. For a wire to
radiate at a particular rate, it must also receive heat at the
same rate. Thus, there must be equilibrium between heat
transferred convectively to the wire and that radiated by the
wire. The following is an approximate relationship for the
heat transferred convectively per unit area to a wire:

Q
A
= (Ts T1) [0.63 + 0.266
(Ts 460)
1000 ] V 0.56
D 0.44
Ts is the temperature of the hot gas glowing over the wire,

and T1 is the temperature attained by the wire, both in
degrees Rankine. The diameter D of the wire is in feet, and V
is the free air velocity in feet per second. For the equilibrium
conditions, the last two equations must yield the same value

for Q /A. The reader can quickly determine that wires on the
order of a thousandth of a foot in diameter can be very effec-
tive additions near the burner in order to enhance radiation.
To complete the assessment of radiation enhancement, we need

to remember that an increase in radiative heat transfer at the bottom
of the tank will lower the temperature of the stack gases that begin
to ow up the stack, thereby reducing the amount of heat trans-
ferred convectively. An approximate calculation of this interrela-
tionship indicates that a 10% increase in radiative heat transfer will
yield an overall increase of about 5.6%. Modications required for
radiation enhancement are minimal and will require little, if any,
change to the basic conguration of the water heater.
The inventor has developed a low-cost, simple conguration
aimed at improving the overall efciency of a gas hot water heater.
Based on the information generated and on our knowledge and
213
experience, technology observation has shown that the invention

will work, provided enough area is added without occluding the
natural convection of the stack. However, adding sufcient area may
be impractical. More importantly, our evaluation has identied
other signicant areas that also may lead to new congurations that
could meet the requirements set forth.
Although this example is far from a complete analysis of water

heater operation, we have uncovered the essential elements to the
extent that we could use the conguration or draw on our under-
standing of the phenomena in application to other situations. We
can identify several concepts that we should put in our bag of tricks,
or our mental knowledge base of building-block parameters.
The example illustrates several interesting features of the tech-
nology observation methodology. Most importantly, we have seen
how an investigation of the operation of the water heater and an
identication of the principal parameters involved have led us to
some possibly unexpected conclusions. If our evaluation had just
determined whether or not the new concept would work, we would
have fallen short of uncovering signicant information about radia-
tive heat transfer in a hot water heater. Second, our study led us to
an alternative scheme, the enhancement of radiative heat transfer,
illustrating how technology observation can build up ones inventive
skills. In fact, this understanding of radiative versus convective heat
transfer becomes an innovation building block. Third, we identied
several other parameters in the overall operation of the water heater
which may be stored in our bag of tricks.
Very often the step of identifying concepts or parameters is not
as straightforward as it is in this example, and signicant mental
effort may be required before the true focus of a technological prod-
uct or invention is recognized. However, it is difcult to perform a
good technology observation unless the principal parameters are
truly understood.
214
Designers should actively and continually be engaged in tech-

nology observation to learn in this fashion and depth what the sur-
rounding technology teaches. This will sharpen the designers skill
and thought process in identifying critical issues in all phases of the
design process. Moreover, such systematically cataloged knowledge,
in the form of general principles and concepts, becomes invaluable
for later use in conceptual design.
The foregoing approach to understanding the critical concepts

behind the output of other innovators should be a life-long activity.
The reader is encouraged to practice this approach on a regular basis
as a means of continuously adding to his or her expertise.
11.5 Bibliography
Information about gas water heaters that may be useful to under-

standing the example can be found in:
Avallone, E. A. and Baumeister, T. Marks Standard Handbook for
Mechanical Engineers. 10th ed. New York: McGraw-Hill, 1996.
Incropera, F. P. and DeWitt, D. P. Fundamentals of Heat and Mass
Transfer. 4th ed. New York: John Wiley & Sons, 1996.
215
12
Conclusion
This chapter summarizes the approach of this book to the initial

stages of the design process, namely, need identication and analy-
sis and conceptual design. Cognitive aspects of parameter analysis
are discussed next, followed by a description of its relationships to
other design methodologies. The chapter concludes with a brief
overview of how the design process should proceed through concept
selection, embodiment design, and detail design.
12.1 The Essence of the Methodology
The ultimate goal of the approach to conceptual design presented in

this book is to help designers to think better. In this context think-
ing better means being more creative and innovative, and doing it
faster. In todays competitive market, a successful product needs to
perform better than its rivals and must also be introduced earlier.
The key elements of the early design stages introduced through-
out the book are:
Need Identication: Clarication of what should and what

should not be designed. The main obstacle to avoid here is fol-
lowing customers congurational task statements.
Need Analysis: The thorough study of everything that may be
related to the design task. We recommend organizing this
216
Conclusion
stage according to the following ve general categories: per-

formance, value, size, safety, and special.
Design Requirements: The functions and constraints gener-
ated during need analysis are summarized as a set of concise
and quantitative specications.
Technology Identication: Conceptual design begins by nam-
ing as many general approaches for addressing the task as pos-
sible. These core technologies are assessed for their chances of
success and serve as starting points for concept development.
Parameter Analysis: Conceptual design is carried out by con-
tinuously cycling through PICSE (parameter identication
creative synthesisevaluation) triplets. Every congurational
step must be conceptually backed by a parameteridea, phys-
ical insight, relationship, justication, and so onand evalu-
ated critically, yet constructively. This iterative process
continues until all the design requirements have been satised
and no unresolved issues remain.
12.2 Cognitive Aspects of Parameter Analysis
As we design, we spend time and resources to make progress toward

creating a solution. The key to successful design is not just making
progress but making progress at the desired rate. Thus, the designer is
faced with the challenge of controlling the design process to achieve
fast progress. As with any process, we must understand the process
in order to control it. In design, if we know:
How we design when faced with a design task, and

What the pitfalls of the natural thought process are,
then we can:
Be conscious of our thought process.

Identify when progress is not being made at the desired rate.
Take corrective actions to become more effective.
217
Let us try to understand the natural thought processthat is, the

course followed by humans when faced with a design taskand
some of the potential pitfalls. We can divide this thought process
into two distinct stages:
1. Recognition of the design task.

2. Development of the design solution.
Recognition of the Design Task

Humans recognize objects by their resemblance to the proto-
types representing various knowledge categories. For instance, we
identify a robin as a bird because it closely resembles the prototype
of the bird category. The prototypes closely resemble the typical
examples in their category. A bird prototype resembles a robin more
closely than a penguin. This fact is reected in the increased time to
recognize penguins as birds. A similar recognition process occurs
when a designer is presented with a design task.
During the initial stages, the designer categorizes the task based
on the prototypes and not on concepts or scientic principles. The
recognition process is very fast and may occur even after reading
only a small portion of the problem statement. It is based on the
configurational issues, since prototypes are configurational in
nature. At the end of the recognition process, the designer would
therefore have activated a prototype that could be either concrete or
abstract depending on the scope of the problem.
Development of the Design Solution

The development process involves modifying, adapting, or siz-
ing the prototype to satisfy the new conditions, constraints, and
requirements of the design task. A phenomenon known as design
xation plays an important role in the development stage. It pre-
vents the designer from considering alternative solutions.
In most cases, the core concept behind the prototype remains
unchanged during the development process because of the difculty
in accessing the concepts behind the prototype. For instance, for
most people the prototype for manholes is circular and that of air-
plane wings is swept backward. However, the fundamental concepts
218
Conclusion
that dictate the geometry may neither be easily accessible nor exis-
tent in the designers mind. Thus, the designer changes small con-
gurational details without gaining deeper understanding.
Pitfalls of the Natural Thought Process

Let us look at a typical thought process during the design of a toy
for blind children. The design task automatically invokes the proto-
type for the toy category. This prototype may be different for various
people depending on their background and gender. The designer
then modies the toy to satisfy the needs of blind children. It is usual
for designers who do not follow a systematic methodology to ignore
issues such as gender, age, sensory mechanisms other than touch/feel,
size of the toy, ability to nd the toy, and color. This example illus-
trates some of the key pitfalls of the natural thought process:
1. The designer develops the solution without proper under-

standing of customer needs. Such activity results in inferior
products that do not appeal to the customers.
2. Due to design xation, designers do not consider alternative
solutions and develop the prototype into the nal solution.
Since the prototype resembles the typical examples in the
category, the nal solution is most likely not innovative.
3. The attention is on congurational issues and details. Key
issues emerge as late as during the testing phase and result in
long product realization times.
4. The development activity revolves around sizing, congura-
tional modication, and adaptation. There is no real synthe-
sis to create the solution.
5. There is a natural tendency to drift from the conceptual
domain to the congurational domain. In contrast, the
movement from the configurational to the conceptual
domain requires considerable cognitive effort. As a result, the
designer may ignore critical problem areas.
The systematic methodologies of this book, namely, need iden-

tication and analysis and conceptual design by parameter analysis,
attempt to overcome the potential pitfalls of the natural thought
process, as summarized in Table 12.1.
219
Table 12.1 How the methodologies of this book overcome pitfalls of the
natural thought process.
Potential Pitfalls of the Benets of the Systematic
Natural Thought Process Methodologies of This Book
Natural tendency to design without The need identication and analysis
proper understanding of the methodology suppresses this ten-
design task. dency and denes the need in
solution-neutral terms.
Due to design xation, only the The technology identication step
prototype is developed into the provides a breadth of ideas and
nal solution. The solution is not several starting points. It reduces
innovative since it closely the design xation and thereby
resembles the typical examples. increases the possibility of creat-
ing an innovative solution.
Focus is on the detailed issues. Parameter identication (PI) focuses
attention on the key conceptual
issues in the design.
Development involves sizing, Creative synthesis (CS) increases
congurational modications, and innovative content and establishes
adaptations. synergy between basic compo-
nents of the design.
Movement from conguration to Evaluation (E) provides an in-depth
concept space is difcult. understanding and aids in mov-
ing from conguration to concept
space.
12.3 Relation to Other Design Methodologies
Several existing design methods and practices have parameter analy-

sis avor. Suhs axiomatic design (see Suh, 1990) states a model of
the design process wherein mapping takes place from the functional
space to the physical space. The functional space consists of func-
tional requirements (FRs), and the physical space contains design
parameters (DPs). Note that while this terminology is very different
from what we use in parameter analysis (we use concept and con-
guration spaces, and the elements in concept space are the ones
dubbed parameters), the essence is similar. Suh also presents design
as a sequential process, in which the most important functional
requirements should be identied rst, a physical solution created,
leading to new, lower-level functional requirements, and so on.
220
Conclusion
The principles that the mapping technique must satisfy to pro-

duce a good design are the following two axioms:
1. Maintain the independence of functional requirements.

2. Minimize the information content of the design.
Many corollaries can be derived from these two axioms to offer the
designer assistance in synthesizing physical solutions to satisfy the
functional requirements. The similarity between axiomatic design
and parameter analysis implies that the two methodologies could
perhaps be combined.
Dixon and Poli (1995) use a design methodology dubbed guided
iteration, which consists of four general steps: formulating the prob-
lem, generating alternative solutions, evaluating the alternatives, and
if needed, redesigning. This four-step process is applied at the con-
ceptual design level, as well as at the later stages of conguration and
parametric design. (These correspond to our embodiment and
detail design, respectively.) Two approaches are offered for generat-
ing the conceptual design alternatives. The rst is based on func-
tional decomposition, and the second calls for searching for suitable
physical laws and effects. The latter emphasis on underlying princi-
ples is reminiscent of our parameters. The last step in the methodol-
ogy, redesign, is guided by the results of previous evaluations, by
qualitative physical reasoning, and by knowledge of manufacturing
processes. As such, it too resembles the iterative nature of parameter
analysis.
French (1999) presents numerous thoughtful examples of con-
ceptual design principles and their application to different prob-
lems. Many of those so-called synthesizing principles and design
rationales can be shown to be similar to what this book calls design
parameters. Also resembling parameter analysis are Frenchs listed
objectives of the design methods, which are:
1. Increasing insight into design tasks and acquiring the insight

fast.
2. Diversifying the approach to the design task.
3. Reducing the size of the mental steps during design.
221
4. Prompting inventive steps and reducing the chances of over-

looking them.
5. Generating problem-specic design philosophies.
Several other design books, mentioned in the bibliography at the

end of this chapter, emphasize the systematization of the design
process. Such systematization is different from the approach of this
book, which attempts to unlock the cognitive processes that take
place in successful conceptual design.
12.4 Whats Next?
This book purposefully focuses on that portion of the design process

that begins with an initial need and ends with conceptual design.
Clearly, much work is still needed to turn a conceptual design into a
product. The downstream stages are outlined below.
Concept Evaluation and Selection

An important characteristic of parameter analysisthe concep-
tual design methodology of this bookis that evaluation is an
inseparable component of the concept development process. This
means that by the time conceptual design is nished, many evalua-
tion steps have been completed. As a result, the conceptual design
has already been judged to meet all the design requirements and to
constitute a feasible solution that is based on ready technology.
Common concept evaluation techniques, which compare the design
with the requirements, study its feasibility, or assess its technology
readiness, are therefore all unnecessary when conceptual design is
done by parameter analysis. However, these approaches may con-
tribute to the evaluation steps along the way.
The conceptual design stage often involves developing several
alternative designs. This is a common practice when a completely
new product is designed. Several designers or design teams are
assigned the same task, each designer or team performs an inde-
pendent conceptual design, and the results are evaluated during a
design review. Realistic constraints on development cost, time, and
222
Conclusion
resources usually preclude the possibility of nalizing the alternative

designs. Instead, the choice among the competing designs needs to
be made earlier, at the end of the conceptual design stage.
Concept selection is not an easy task. The main difculty lies in
basing the evaluation and decisions on order-of-magnitude infor-
mation associated with conceptual designs, as opposed to the com-
plete quantitative data available for nished designs. This may be
even more challenging in situations where the conceptual design
stage terminates after listing only general ideas and solution
approaches. Fortunately, parameter analysis requires that all the
major issues and aspects of a design be addressed, thus resulting in
a relatively detailed description of the design. Not only are the
designs quite well developed at the end of our conceptual design
stage, but the development process itself is so elaborate that much
insight into the task and the solution has been gained by this time.
The most suitable selection method to apply to conceptual
designs generated by parameter analysis is Pughs method, which is
based on a comparison matrix (Pugh, 1991). Concepts are listed as
columns, and evaluation criteria as rows. One concept is chosen as
datum, and every other concept is assigned a + if it is better, if
it is worse, and an S if it is the same as the datum, for each crite-
rion. Next, the pluses and minuses are summed, and the winning
concept is selected. There are some variations on the basic theme,
such as scoring on a numerical scale with and without relative
weights, but we shall not describe this method in detail because it is
well documented in many books.
However, a word is in order regarding the choice of the evalua-
tion criteria. The design requirements are usually not suitable as cri-
teria because all the concepts that reach this stage have been shown
to satisfy the requirements. This means that a set of criteria needs to
be developed especially for selecting the concept. With the added
insight gained during conceptual design, and keeping in mind the
original need, criteria such as the ease with which property X can
be achieved and the robustness of the design should be used. In
any case, the designer should apply Pughs method for concept selec-
tion as a tool and not let the method make the decision for him or
her. It is recommended that the total number of pluses, minuses, or
223
the difference between them not be used blindly. Instead, the

designer should use the comparison matrix to examine the specic
strengths and weaknesses of the concepts.
Embodiment Design
Having selected the most promising concept for further devel-
opment, the design process progresses to the embodiment design
stage. This stage is also known as preliminary design, layout design,
or conguration design. When conceptual design ends with vague
and general ideas, embodiment design becomes a major develop-
ment stage. However, with the detailed conceptual designs that
result from parameter analysis, not that much is left to do here.
The major task during embodiment design is rming up the
product structure. Separate components and subassemblies need to
be dened now, and a layout drawing should be produced. Issues
that were judged noncritical during conceptual design should be
addressed at this stage. For example, a specic electric motor
required by the design, the type of gears to use for speed reduction,
or the class of material (e.g., steel, plastic) can be specied.
Preliminary calculations may also be required to size components.
The conceptual design methodology of parameter analysis may
also be applied to embodiment design. The philosophy of handling
one design aspect at a time, focusing on the more important issues
rst, and supporting congurational decisions with concepts is still
useful. Principles commonly used in embodiment design, such as
use short and direct transmission path for forces, match manu-
facturing process to production volume, and reduce part count to
improve ease of assembly, are all concepts or parameters that affect
the conguration.
Detail Design
Comprehensive calculations of strength and other properties,
precise dimensioning and tolerancing, material specication, and
many other details are produced during this nal stage of design.
The remainder of the product realization process, which may
involve prototyping, testing, and the like, is beyond the scope of this
book.
224
Conclusion
12.5 Bibliography
Some of the cognitive aspects of parameter analysis are studied in

the following articles:
Jansson, D. G., Condoor, S. S., and Brock, H. R. Cognition in
Design: Viewing the Hidden Side of the Design Process.
Environment and Planning B: Planning and Design 19 (1992):
257271.
Jansson, D. G. and Smith, S. M. Design Fixation. Design Studies
12, No. 1 (1991): 311.
Axiomatic design is the subject of:

Suh, N. P. The Principles of Design. New York: Oxford University
Press, 1990.
The guided iteration design methodology is used in the following

textbook:
Dixon, J. R. and Poli, C. Engineering Design and Design for
Manufacturing: A Structured Approach. Conway, MA: Field
Stone Publishers, 1995.
Many examples with parameter analysis avor can be found in:

French, M. Conceptual Design for Engineers. 3rd ed. London:
Springer-Verlag, 1999.
Some books that emphasize systematic design processes are:

Cross, N. Engineering Design Methods. Chichester: John Wiley &
Sons, 1989.
Pahl, G. and Beitz, W. Engineering Design: A Systematic
Approach. London: The Design Council, 1988.
Stoll, H. W. Product Design Methods and Practices. New York:
Marcel Dekker, 1999.
Ullman, D. G. The Mechanical Design Process. 2nd ed. New York:
McGraw-Hill, 1997.
Ulrich, K. T. and Eppinger, S. D. Product Design and
Development. 2nd ed. New York: McGraw-Hill, 2000.
Our recommended procedure for concept selection is described in:

Pugh, S. Total Design. Wokingham, England: Addison-Wesley,
1991.
225
Index
abstraction, 61 cam
acceleration prole, indexing barrel, 149
system, 1389, 150 plate, 149
airbag, 29, 33 camfollower, 138
airow sensor, see sensor, airow can, aluminum, 356
airplane wing, 218 canal, 17
aluminum can, see can, aluminum canal, ship
aluminum foil, see foil, aluminum Panama, 178, 77
aluminum-lithium, 161 Suez, 17
analogical thinking, 84 capital cost, see cost, capital
analogy, 75, 84, 208 Carmichael, A. D., xiii
anemometer, 108 Cassini, 34
asbestos, 123 checklist for requirements, 38, 40
ASME, 184, 204 chilling effect, 160, 177
axiomatic design, 220 clear-ice, see ice, clear
clutch
backtracking, 978 one-way, 143
bag of tricks, 207, 214 roller-ramp, 144
ball mover, case study, 184206 sprag, 144
ballpoint pen, see pen, ballpoint cognitive aspects of parameter
battery, 91 analysis, see parameter
billet, 158 analysis, cognitive aspects of
black box, 19, 38, 43, 186 coin, 90
brakes, disk, 29 competition, ASME, 184, 204
buckling, 166 concept selection, 6, 222
cable car, 193 concept space, 613
227
Index
conceptual design, 13, 5, 6, 58, 102 design optimization, 578

activities in, 102 design parameter, see parameter
methodology, 64 design requirements, 34, 49, 108,
nature of, 56 127, 142, 161, , 192, 217
theoretical model of, 61 detail design, see design, detail
conguration space, 613 disk, computer oppy, 26
conict, 29 disk brakes, see brakes, disk
connecting rod, 945 divergent thinking, 96
constraints, 9, 202 Dixon, J. R., 221
constraints Doppler effect, 110
explicit, 22 drag force, 1107
implicit, 22 drawbridge, 192
temporal, 29 dynamite, 89
contest, student design, see ASME
convective heat transfer, see heat eddy current, 99
transfer, convective Eiffel Tower, 30, 87
convergent thinking, 96 electrical fuse, see fuse, electrical
core business, 145 embodiment design, see design,
core technical concept, 1, 58 embodiment
cost, capital, 30 engine
creative synthesis, 667, 72, 75, 86 reciprocating, 58
CT scanning, 801 rotary, 5860, 689
cut-edge sensor, case study, engineering design process, 4
12336 equal-channel-angular-extrusion
case study, 15782
decomposition functional, 23 evaluation, 667, 72, 90, 95
deformation explosives, 889
plastic, 157 extrusion, 157
shear, 157, 161 Conform process, 1812
design constant, 158
detail, 7, 224 ratio, 159
embodiment, 7, 224
layout, 224 xation, see design xation
preliminary, 224 ooring removal, 123
process overview, 45 oppy disk, see disk, computer
requirements, 9, 34, 49, 108, oppy
127, 142, , 161, 192, 217 ow nozzle, 109
specications, 34 foil, 845, 89
design contest, see ASME manufacturing, 89
design xation, 21820 fountain pen, see pen, fountain
228
Index
fracture mechanics, 94 initial need

French, M. J., 221 airow sensor, 105
fuel cell, 28, 923 ball mover, 184
functional decomposition, see cut-edge sensor, 123
decomposition, functional electrical fuse, 42
functional requirements, 220 equal-channel-angular-
functions, 9, 201, 256 extrusion, 157
primary, 25 indexing system, 137
secondary, 25 innovation, 1
fuse, electrical John Hancock Center, 868
case study, 4152
critical parameters keyboard, 104
in design, 76 Khan, F., 86
gas water heater, see water heater lens, 130

generalization, 61 Li, Y. T., xiii, 70
grinding, 94 light intensity, 128
guided iteration , 221
magnetic drive, 98
Hancock Center, see John magnetic seal, 98101
Hancock Center manhole cover, 90, 218
headlight, 79 market need, 10
heat transfer Mars sample return mission, 16,
coefcient, 210 36
convective, 209 matching, 84
radiative, 2113 metalworking, 157
surface, 209 methodology, 3, 65
hip replacement, see THR methodology vs. method, 3
hot water heater, see water heater Moores law, 28
HVAC airow sensor, see sensor,
airow NASA, 16, 34
hybrid-powered car, 934 NASA Mars sample return
mission, see Mars sample
ice, clear, 812 return mission
imaging, 128 natural thought process
incidence, angle of, 129 development of the design
inclinometer, see tiltmeter solution, 218
indexer, industrial, 137 methodologies to overcome
indexing system pitfalls, 220
case study, 137156 pitfalls, 219
229
Index
natural thought process (continued) as design methodology, 64

recognition of the design task, cognitive aspects of, 21720
218 overview, 71
need analysis, 6, 19, 22, 44, 106, relationship to other
124, , 138, 159, 187, 216 methodologies, 2202
airow sensor, 106 parameter analysis case studies
ball mover, 187 airow sensor, 111
case study, 41, 184 ball mover conceptual design
categories, 9, 23, 323 I, 194
cut-edge sensor, 124 ball mover conceptual design
electrical fuse, 44 II, 199
equal-channel-angular- cut-edge sensor, 128
extrusion, 159 equal-channel-angular-
indexing system, 138 extrusion, 161
methodology, 19 indexing system conceptual
need denition, 10, 45 design I, 143
need identication, 5, 6, 9, 14, 43, indexing system conceptual
186, 216 design II, 147
case study, 41 parameter analysis process
importance, 9 summary
in practice, 14 airow sensor, 118
need statement, 189, 44 ball mover conceptual design
neutral buoyancy, 77 I, 199
nitroglycerine, 88 ball mover conceptual design
Nobel, A., 89 II, 205
nozzle, see ow nozzle cut-edge sensor, 134
equal-channel-angular-
optimization, 57 extrusion, 179
orice meter, 109 indexing system conceptual
design I, 148
pack rolling, 89 indexing system conceptual
packing factor, fuse, 41, 43, 45 design II, 153
Panama Canal, see canal, ship parameter identication, 657, 72,
parameter, 65, 75, 217 75, 78, 95
building-block, 214 particle making, 110
dened, 75 particularization, 61
substitution, 83 payback period, 301
parameter analysis, 3, 53, 95, 111, pen
128, 143, , 147, 161, 194, 199, ballpoint, 58
217 fountain, 57
230
Index
pendulum, 53 radiative heat transfer, see heat

inverted, 54 transfer, radiative
performance, 23 radioisotope thermoelectric
airow sensor, 106 generator (RTG), 33
ball mover, 187 range measurement, 128
cut-edge sensor, 124 realization, 5,7, 61
electrical fuse, 44 recycling, 356
equal-channel-angular- remotely operated vehicle, see
extrusion, 159 undersea remotely operated
in need analysis, 9, 23, 323, vehicle
106, 124, 138, , 159, 187 requirements, see design
indexing system, 138 requirements
performance-value relationship, 31 return-on-investment, 301
photodetector, 129 reverse engineering, 208
photodiode, 132 robot, 734, 823
photosensitive element, 128 robot control, 82
Pitot-static tube, 109 robot plow, 73
plow, 73, 78 rotameter, 109
pneumatic actuator, 142, 150 rotary engine, see engine, rotary
Poissons effect, 164
Poli, C., 221 safety, 32
power source for space airow sensor, 107
exploration, 33 ball mover, 191
printer, computer, 1 cars, 33
product specications, 34 cut-edge sensor, 127
production volume, 31 electrical fuse, 49
propeller, 109 equal-channel-angular-
prototype, 218 extrusion, 160
prototyping, 7 in need analysis, 9, 23, 32, 127,
Pugh, S., 223 160, 191
Pughs method, 223 seal, magnetic. See magnetic seal
puller rod, 193, 199 Sears Tower, 868
punch, 158, 162 sensor
airow, 10522
QFD (quality function cut-edge, 12336
deployment), 389 shadow, 129
QLF (quality loss function), 46 ship canal, see canal
quality function deployment, see simplication, 62, 66, 78
QFD size, 32
quality loss coefcient, 47 airow sensor, 107
quality loss function, see QLF ball mover, 191
231
Index
size (continued) indexing system, 142

cut-edge sensor, 126 robot plow, 73
electrical fuse, 48 technology observation, 4, 2078
equal-channel-angular- temperature, condensation, 211
extrusion, 160 temperature difference, log mean,
in need analysis, 9, 23, 32, 106, 210
126, 138, , 160, 191 tether, 24
indexing system, 141 THR (total hip replacement),
skyscraper, 86 834
solution space, 1921, 50 tiltmeter, 536, 62, 70, 78
space suits, 85 total hip replacement, see THR
special considerations toy, for blind children, 219
electrical fuse, 49 trade-off, 22, 29, 31
in need analysis, 9, 23, 33, 192 transformation, 60, 62, 66, 78, 80
specic energy, 91 trihedral cavity, 200
specications, see design turbine, 109
specications and product typewriter, evolution of, 104
specications
spring energy storage, 142, 147 undersea remotely operated
strain vehicle, 245
effective, 163
true, 158 value
strain gage, 111 airow sensor, 107
stress shielding, 83 ball mover, 190
submarine, 77 cut-edge sensor, 126
substitution, see parameter electrical fuse, 46
substitution equal-channel-angular-
Suez Canal, see canal, ship extrusion, 160
Suh, N. P., 220 in need analysis, 9, 23, 29, 46,
system integration team, 245, 36 1067, 126, , 138, 141, 160, 190
indexing system, 141
task statement, 14, 16, 20, 80 velocity prole, 189
technology identication, 5, 6, Venturi, 109
725, 96, 108, 127, , 142, 161,
192, 217 Wankel engine, see engine, rotary
technology identication water heater, 208
airow sensor, 108 work hardening, 157
ball mover, 192
cut-edge sensor, 127 zero emission vehicle (ZEV), 15,
equal-channel-angular- 914
extrusion, 161
232

Innovative Conceptual Design - Theory and Application of Parameter Analysis

Uploaded by

Copyright:

Available Formats

Innovative Conceptual Design - Theory and Application of Parameter Analysis

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Innovative Conceptual Design - Theory and Application of Parameter Analysis

Uploaded by

Copyright:

Available Formats

Conceptual design, along with need identication and analysis, make

Ehud Kroll is on the faculty of the Mechanical Engineering Depart-

Sridhar S. Condoor is a professor in the Department of Aerospace

David G. Jansson is the principal in David G. Jansson & Associates. He

Cambridge University Press 2001

First published in printed format 2001

A catalogue record for the original printed book is available

ISBN 0 511 01928 9 virtual (netLibrary Edition)

3.3 Need Identication 43

7.2 Abbreviated Need Analysis 124

10.7 Parameter Analysis of Conceptual Design II 199

Over the past several years, we have taught project-based courses on

ment of a design task into a set of design requirements. In this trans-

designer to continually incorporate new ideas into the design. It is

Conceptual design is the thought process of generating and imple-

1.1 What Is Conceptual Design?

The core technical concepts developed during conceptual design fun-

Core Technical Concepts of Computer Printers

The context in which engineering design activities occur is

1.2 Parameter Analysis: A Conceptual Design

The primary focus of this book is the conceptual design process

process since it involves the creation of something new. (The

1.3 Overview of the Engineering Design Process

Numerous terms are used to refer to the process or parts of the

Figure 1.1 Overview of the engineering design process.

tion, and, of course, engineering design. Practitioners and academ-

A stage of identifying and analyzing a need prior to initiating

A high-level view of the engineering design process is represented by

The purpose of the upstream stage called need identication and

1. Discovering what the real need is, and

In a sense, both the roles of need identication and analysis

The broadly dened downstream design and development

1.4 Structure of the Book

This book is heavily dependent on examples to elucidate the

hand the value of the coaching method called parameter analysis.

Every design process begins by identifying and analyzing the need,

2.1 The Importance of Need Identication

Design tasks arise from a variety of sources, sometimes referred to

ing full understanding of what should really be designed. This ten-

Market Needs: Insights from the Evolution of Typewriters

The typewriter, a great American invention, shaped our society

end of the top row). This arrangement, which helped to avoid

" < > ? +

Figure 2.1 Dvoraks keyboard arrangement which separates fre-

would require retraining people and buying new typewriters. This

foreign countries. By producing different-language typewriters,

2.2 Need Identication in Practice

Design tasks may originate from sources external or internal to the

Zero Emission Vehicles

Automobiles contribute well over half of the emissions that dam-

To suppress the tendency of being conguration- or product-ori-

NASAs Mars Sample Return Mission

One possible initial task statement for NASAs Mars sample

Let us examine another task statement, that of design a ship

To develop a good need statement, it is often useful to view the

2.3 The Need Analysis Methodology

Figure 2.4 Schematic of a solution space bounded by constraints. The task

In an ideal situation, the designer is concerned with satisfying

Constraints can be generated by examining the original design